On target: ensuring geometric accuracy in radiotherapy

Transcription

1 On target: ensuring geometric accuracy in radiotherapy The Roya Coege of Radioogists Institute of Physics and Engineering in Medicine Society and Coege of Radiographers

2 Contents Foreword 6 Executive summary 7 1 Introduction Purpose Objectives Background 8 2 Principes of geometric verification What is verification? Verification definitions Verification Reference image Pretreatment verification Image-guided radiotherapy (IGRT) Off-ine treatment verification Onine treatment verification Interfractiona verification Intrafractiona verification Rea-time treatment verification Set-up error definitions Gross error Systematic error Random error Set-up error measurement Set-up error and treatment margin 14 3 How to verify: creating geometric verification processes Equipment and technica infrastructure Generation of reference and treatment images Data transfer and storage Quaity assurance and image-matching accuracy Consistency of co-ordinate systems, error reporting and corrective actions Verification data management software 17 On target: ensuring geometric accuracy in radiotherapy 3

3 3.2 Personne, responsibiities and training Manageria responsibiities Cinica responsibiities Training and competency programmes Imaging protocos Image acquisition Frequency and timing of imaging Measurement of set-up errors Identification of gross error Determining systematic and random errors for individua patients Toerances, action eves and correction strategies Toerances Action eves Corrective strategies Assessment and correction of systematic errors Dose considerations concomitant exposure Audit Use of audit to evauate imaging protocos Use of audit to evauate techniques and immobiisation systems Identification of systematic errors in the departmenta processes 28 4 Derivation of systematic and random set-up errors and reationship to the CTV-PTV margin Set-up error Systematic set-up errors Random set-up errors Reationship between the CTV-PTV treatment margin and treatment verification Effect of correction protocos on margins Margin derivation Margin contro 35 5 Training and competency assessment Training and competency assessment Training manua covering the foowing subjects Competency assessment methods Cinica competency knowedge and skis Further competency assessment for advanced practice 37 6 Equipment used for geometric verification Reference images Image acquisition modes 39 4 On target: ensuring geometric accuracy in radiotherapy

4 6.3 Traditiona equipment and techniques Nove equipment and techniques 41 7 Site-specific protocos for geometric verification Brain Suggested protoco for brain verification Evidence for brain verification guideines Head and neck Suggested protoco for head and neck verification Evidence for head and neck verification guideines Thorax and mediastinum Suggested protoco for thoracic and mediastinum verification Evidence for thoracic and mediastinum verification guideines Breast Suggested protoco for breast verification Evidence for breast verification guideines Pevis (prostate, badder, gynaecoogica and coorecta cancers) Suggested protoco for pevis verification Evidence for pevis verification guideines Evidence for prostate verification guideines Evidence for badder verification guideines Evidence for coorecta verification guideines Evidence for gynaecoogica verification guideines Spine Suggested protoco for spina verification Evidence for spina verification guideines Limb Suggested protoco for imb verification Evidence for imb verification guideines Paediatric Suggested protoco for paediatric verification Evidence for paediatric verification guideines Paiative Suggested protoco for paiative verification Evidence for paiative verification guideines 63 8 Gossary 64 9 References Working Party Members 76 On target: ensuring geometric accuracy in radiotherapy 5

5 Foreword This report provides guidance on how to improve the accuracy of radiotherapy. It expains the significance of the systematic and random errors present in even the best-run departments and ays out methodoogies for measuring and minimising such errors. By providing cear guidance it wi assist in accurate and reproducibe radiotherapy deivery. Cinicians and radiographers shoud ensure that for each patient the random uncertainties of set-up are minimised by attention to immobiisation procedures. Systematic error (in the same direction on a fractions) shoud be identified and then eiminated, by the use of an agreed correction strategy. Department heads wi need to ensure that these processes are in pace not ony for individua patients but aso for every radiotherapy technique in use in their department. In this way, popuation-based random and systematic errors can be identified and minimised by changing departmenta practices. Measurement can be used to provide the basis for cacuating the correct margin between the cinica target voume (CTV) and the panning target voume (PTV). Competion of this cyce of individua and popuation-based measures to address random and systematic errors wi aow each of us in our own departments to be certain of treating our patients with the minimum margin required without missing the target. I woud ike to thank Peter Hoskin (Chair), Mark Gaze, Tony Greener, Mike Kirby, Heen McNair, Meanie Powe, Donna Routsis and Diana Tait for their work in producing this document. Thanks aso go to Dr Michae Wiiams who initiated this project during his time as Dean of the Coege. I woud aso ike to thank Karen Seby for her assistance. Jane Barrett Dean of the Facuty of Cinica Oncoogy The Roya Coege of Radioogists 6 On target: ensuring geometric accuracy in radiotherapy

6 Executive summary Treatment verification is an important component of radiotherapy. The roe of verification is primariy to detect treatment deivery errors and secondy, to assess the suitabiity of the size of the margins panned around the cinica target voume that aow for the uncertainties in the radiotherapy process. This document describes and recommends the best evidence-based practices for geometric treatment verification. It aso provides guideines as to how individua centres may impement geometric verification processes ocay. Summary of main recommendations Geometric verification is mandatory for a megavotage X-ray externa beam radiotherapy. The geometric verification process must be carried out within a ceary defined structure, adhering to ocay defined protocos. Each radiotherapy department shoud determine the verification protocos and panning margins required for their own practice. This is because the frequency of imaging, the toerances and action eves used, and the panning margins wi vary according to oca use of techniques, processes, anatomica site, equipment and immobiisation. Cinica impementation of geometric verification protocos shoud be co-ordinated by a designated mutiprofessiona team. Training and competency assessment must be undertaken by a professiona groups invoved in verification. A gross error is an unacceptaby arge set-up error. Methods must be in pace in each radiotherapy department to detect and act on gross errors for each patient at the start of each radiotherapy treatment course. Set-up errors have both systematic and random components. Verification protocos are necessary to identify each component. Systematic errors must be identified and minimised using correction protocos for every patient having a muti-fraction course of radiotherapy. Random errors shoud be minimised by carefu attention to immobiisation and patient preparation techniques. Popuation random and systematic errors shoud be determined for a techniques in use and a anatomica sites treated within individua departments. These data shoud be used to inform the cinica target voume (CTV) to panning target voume (PTV) margin for each technique. Additiona exposure for verification must be justified within the Ionising Radiation (Medica Exposure) Reguations (IR(ME)R). On target: ensuring geometric accuracy in radiotherapy 7

7 1 Introduction 1.1 Purpose The purpose of this report is to recommend best practices for geometric treatment verification within megavotage externa beam radiotherapy and to provide guideines for the oca cinica impementation of these practices. The scope of this report incudes a current and emerging treatment verification methods in UK departments, for a treatment sites, compexities and intents. 1.2 Objectives This report aims to: Provide evidence-based guideines for impementing geometric verification into cinica practice Provide guidance for each radiotherapy centre to create oca management structures, processes and protocos that woud aid the impementation of geometric verification practices. This incudes describing methods by which each centre can determine: The oca verification protocos required Site-specific and individua patient systematic and random set-up errors, which can be used in defining treatment panning margins. 1.3 Background An audit of UK verification practice carried out in 2004, 1 identified a wide diversity in practice across the UK; as a resut, the need for guidance to direct verification in practice was identified. This report detais the impementation process; additiona information to support this is covered in other radiotherapy pubications, specificay: Deveopment and Impementation of Conforma Radiotherapy in the United Kingdom 2 Internationa Commission on Radiation Units and Measurements (ICRU) Reports 50 3 and 62 4 Geometric Uncertainties in Radiotherapy 5 Guidance for the Cinica Impementation of Intensity Moduated Radiation Therapy 6 Baancing Costs and Benefits of Checking in Radiotherapy 7 Towards Safer Radiotherapy. 8 Image-guided radiation therapy (IGRT) is the atest deveopment for increasing the precision and accuracy in radiation deivery. Many of the concepts detaied in this pubication are immediatey transferabe to these new technoogies. The main procedura recommendations, such as the use of speciaised, mutiprofessiona teams for co-ordinating geometric verification, standardising methods of training and authorisation, and the use of ceary defined protocos, are essentia for IGRT. 9,10 The acquisition of voume image data throughout a course of treatment introduces other areas of verification (for exampe, soft tissue verification and adaptive radiation therapy), which need to be considered. These are beyond the scope of this pubication, but readers may be interested in recent pubications highighting the new probems being uncovered by IGRT and their potentia soutions On target: ensuring geometric accuracy in radiotherapy

8 2 Principes of geometric verification 2.1 What is verification? Radiotherapy verification is the process that enabes us to be certain we are treating the tumour voume as panned. In ensuring that the right radiation dose has been given to the right pace, two measures are needed geometric and dosimetric verification. This report concentrates on geometric verification. A typica exampe of a fowchart for the process is shown in Figure 1. The aim of geometric verification is to ensure that the geometric accuracy of the radiotherapy deivered is within the imits set by the uncertainty margin aowed in the treatment pan. This is achieved by comparing information from the deivery against that panned. Verification is ony one component of the treatment process. Accurate and reproducibe panning procedures, incuding the acquisition of good quaity reference images, are essentia to successfu verification. Figure 1. An exampe of a typica fowchart for the process of geometric verification PRETREATMENT Acquire reference images TREATMENT Fraction 1 Acquire treatment images Review images onine against reference Above gross error action eve? Fraction 2 Acquire treatment images No Continue with current set-up Yes STOP Revise set-up, re-image Fraction 3 Acquire treatment images Cacuate mean dispacement (systematic set-up error - SSE) in a 3 axes Above SSE action eve? No Yes Continue with current set-up First time cacuated: Adjust set-up. Re-image for at east next 2 fractions Fraction 4 Fraction 5 No imaging No imaging Acquire treatment images Acquire treatment images Cacuate mean dispacement for 2 images (systematic set-up error - SSE) in a 3 axes Above SSE action eve? No Yes Weeky Acquire treatment images Continue with current set-up STOP Investigate further Each imaged fraction: Review images off-ine against reference Out of toerence? No Continue with current set-up Yes Re-image for at east 2 fractions If sti out of toerance, image again and recacuate SSE On target: ensuring geometric accuracy in radiotherapy 9

9 2.2 Verification definitions Many verification terms are used in this report. The foowing section describes their specific meanings to set the common anguage used for the report. Other definitions used in this report are given in the Gossary (page 64) Verification This is the process by which the accuracy of radiotherapy is assessed. It is achieved by comparing images (or data) of the treatment deivered with that panned. This wi use information from either 2D or 3D systems to give different degrees of transationa and rotationa set-up accuracy data Reference image The reference image obtained shows the panned geometry of the treatment fied pacement reative to interna anatomy or anatomica surrogate such as bone or markers. This is used as the standard against which treatment images are assessed. Reference images are produced in numerous ways incuding: digitay reconstructed radiographs (DRRs), digitay composited radiographs (DCRs), simuator images, digitised fims, utrasound, or the entire voumetric panning data set. In this document, image is used to encompass a of these modaities Pretreatment verification This is the process that compares the reference images with the panned treatment before the course of radiotherapy is started. It usuay occurs away from the treatment deivery room Image-guided radiotherapy (IGRT) In its broadest definition, this appies to a parts of the radiotherapy process from using imaging to define and deineate the target voume to evauating treatment response. The most widey used concept of IGRT is using imaging in the treatment room either immediatey before or during treatment to evauate and correct set-up errors. For onine treatment verification, IGRT uses images obtained immediatey before treatment deivery and intervention to correct set-up before deivery. Images may be acquired using computed tomography (CT) (kiovotage [kv] and megavotage [MV]), porta images (MV), kv panar radiographs, utrasound or other methods. Despite improved imaging, it is not possibe to correct for a components of geometric error in radiotherapy. There are inevitaby residua errors (to be accounted for in margin cacuations), which may arise from sources such as: Target deineation uncertainty (cannot be detected by imaging) Movement of the patient or interna motion of organs as the treatment is being deivered. Intrafractiona verification may be needed to quantify this Off-ine treatment verification This compares the reference images with the images taken in the treatment deivery room, and anayses the set-up accuracy at some time after the treatment has been given. The set-up data are not acted on unti the next treatment Onine treatment verification This compares the reference images with images taken in the treatment deivery room, immediatey prior to the treatment being deivered. Any necessary corrections are appied before the treatment is deivered. Ideay, the time taken between onine verification and treatment deivery shoud be as short as possibe (a few minutes), to reduce the variation that may occur from patient movement during this time. Beyond this timescae, the information may no onger represent the patient s true position during the therapy Interfractiona verification This compares the set-up accuracy between different treatment fractions. 10 On target: ensuring geometric accuracy in radiotherapy

10 2.2.8 Intrafractiona verification This compares the set-up accuracy during a singe treatment fraction and may be assessed over the course of a singe beam being deivered or over a singe fraction. The effect of intrafractiona movement can be compensated for when panning treatment margins, or reduced by the foowing methods: Terminating the treatment beam if movement occurs outside predefined toerances Timing the treatment beam (puse) to ensure deivery of radiation coincides with a known position of the patient s interna anatomy (gating) Restricting variation in the position of interna anatomy Rea-time treatment verification This is where comparisons are made between the reference images and images taken in the treatment deivery room, as the radiation is being deivered. Most rea-time verification methods detect dispacements over a predetermined eve, so that the operator or automated system can stop, or gate, the treatment. Other rea-time systems use the reationship between externa references and interna anatomy. Optica surface detection systems work on a simiar principe, by stopping treatment if the patient s externa skin contours or reference points move outside a set toerance eve. This non-radiation method of rea-time verification reies on the assumption that the reationship between externa reference points and interna anatomy remains constant. 2.3 Set-up error definitions The term set-up error is used in this document to describe the discrepancy between intended and actua treatment position. It comprises a systematic and random component. It is normay cacuated as a shift in treatment fied position when a treatment image is compared against its corresponding reference. The set-up error may be determined reative to the isocentre, the fied borders or both and can contain transationa and rotationa information Gross error A gross error is an unacceptaby arge set-up error that coud underdose part of the cinica target voume (CTV) or overdose an organ at risk. CTV to panning target voume (PTV) treatment margins do not account for errors of such magnitude and therefore gross errors must be corrected before any treatment commences. Mechanisms to detect gross errors must be in pace before any treatment course starts. The usua way of achieving this is to perform imaging after treatment panning, but before the start of treatment. The methods which can be used are detaied in Section The preferred method is to take and assess an eectronic image on the first treatment fraction, immediatey prior to treatment deivery. Possibe causes of gross error woud incude: Incorrect patient, anatomica site or patient orientation Incorrect fied size, shape or orientation Incorrect isocentre position of unacceptabe magnitude. Each department must decide on an appropriate magnitude for gross error and this may differ between treatment sites. The chosen vaue must exceed any that coud occur from expected day to day fuctuation in treatment position. In practice a gross error action eve of 10 mm is appropriate for a wide range of sites and techniques Systematic error The systematic component of any error is a deviation that occurs in the same direction and is of a simiar magnitude for each fraction throughout the treatment course. On target: ensuring geometric accuracy in radiotherapy 11

11 When considering geometric uncertainties in radiotherapy, the term systematic error may be used when referring to the individua patient, or to the treatment popuation, and this distinction needs to be carified to avoid confusion. Individua the systematic error for an individua patient is the mean error over the course of treatment. Popuation the systematic error for a group of patients is an indication of the spread of individua mean errors. It is cacuated as the standard deviation (SD) of the distribution of mean errors for each individua patient and is usuay given the capita sigma symbo Σ error where the subscript error refers to the particuar error considered (for exampe, Σ set-up for the measured systematic set-up error). Systematic errors may be introduced into a patient s treatment at the ocaisation, panning or treatment deivery phases. For this reason these types of errors are often referred to as treatment preparation errors. 17 Once frozen into the process, systematic errors wi occur in each treatment fraction. Possibe treatment preparation errors 5 are summarised beow. Target deineation error this may be introduced when the CTV is first deineated and represents the difference between the defined and idea CTV. Target position and shape this is a change in target position and shape between deineation and treatment. Possibe causes incude tumour regression or growth, badder fiing and recta distension. Phantom transfer error this is the error that accumuates when transferring image data from initia ocaisation through the treatment panning system to the inear acceerator. It is measured using a test phantom and may be sub-divided into geometric imaging, treatment panning system and inac geometry errors. 18 Possibe causes incude differences in aser aignment between CT and inear acceerator, CT couch ongitudina position indication, image resoution, margin growing agorithm, fied edge and mutieaf coimator (MLC) eaf position, isocentre ocation, source to surface distance indication, gantry and coimator ange accuracy. Many of these parameters are subject to routine checks as part of a machine quaity contro programme and this shoud ensure that any differences ie within aowed toerances such as +2 mm for a distance and +1 for an ange indication. Phantom transfer errors are cassed as systematic because their causes either do not change (image resoution, margin agorithm) or are assumed to vary sowy (isocentre position, eaf position accuracy) and are therefore taken as constant over the typica treatment duration. Patient set-up error this describes a causes of treatment set-up error not accounted for by the phantom transfer error and incudes a the errors isted under gross error. Possibe causes incude changes in the patient s position, shape or size (for exampe, weight change, hair oss). It aso encompasses more subte effects such as the dispacement of the target reative to skin set-up marks caused by the CT scan and treatment being performed on different couches. Patient set-up error is ony one possibe component of the overa measured systematic set-up error. The chosen method of treatment verification wi determine how many of the above sources of systematic error wi be incorporated into the measured set-up error Random error The random component of any error is a deviation that can vary in direction and magnitude for each deivered treatment fraction. When considering geometric uncertainties in radiotherapy, the term random error may be used to refer to the individua patient or to the treatment popuation and, as for systematic errors above, this distinction needs to be carified to avoid confusion. Individua the random error for an individua patient is the standard deviation (SD) of the measured errors over the course of treatment and quantifies the spread of errors. Popuation the random error for a group of patients is cacuated as the mean of the individua random errors and is given the ower case sigma symbo σ error where the subscript error refers to the particuar error considered (for exampe, σ set-up for the measured random set-up error). 12 On target: ensuring geometric accuracy in radiotherapy

12 Random errors occur at the treatment deivery stage and for this reason are often referred to as treatment (or daiy) execution errors. 18 They are summarised beow. Patient set-up error these are varying, unpredictabe changes arising from change in a patient s position, treatment equipment or set-up methodoogy between each deivered fraction. 19 Target position and shape the change in target position and shape between fractions. This error is essentiay the same as that described above for systematic errors but accounts for motion between fractions rather than from deineation to treatment. Intrafraction errors this describes changes in the patient s position and interna anatomy arising during the deivery of a singe fraction, for exampe, due to breathing. Random errors are infuenced by the immobiisation system, patient compiance and department protocos. If a new immobiisation device is introduced, it is ikey that the random error wi be affected. An off-ine correction strategy cannot predict the random error component in subsequent fractions and so treatment margins must be cacuated to incude these variations. Onine correction strategies can be used to contro random errors. Figure 2 shows the difference between systematic and random errors. The daiy set-up errors potted (in miimetres) from anterior-posterior images acquired for two patients over the course of their treatment. Patient 1 exhibits a sma systematic (mean) set-up error compared to Patient 2. Patient 1 has a arger, random spread of errors than Patient 2. Athough Patient 1 has an overa treatment accuracy cose to that intended, any individua image taken is a poor indicator of this mean position. Action on any individua image must therefore be undertaken with caution as it can ead to overcorrection. Sup - inf Sup - inf Sup- inf Sup- inf Set-up errors in Patient 1 (in mm) Set-up errors 10 in Patient 1 (in mm) Patient 1 set-up dispacements 8 Left - right Average (systematic) error Patient 1 set-up dispacements 10 Left - right Average (systematic) error Set-up errors in Patient 2 (in mm) Set-up errors 10 in Patient 2 (in mm) Patient 2 set-up dispacements 8 Left - right Average (systematic) error 10 Patient 2 set-up dispacements Left - right Average (systematic) error On target: ensuring geometric accuracy in radiotherapy 13

13 2.3.4 Set-up error measurement The set-up error measured from a singe image wi contain both systematic and random components. As outined above, the systematic part of the measurement wi nominay be constant from one fraction to the next whereas the random part wi vary in an unpredictabe way. The difference between systematic and random error is demonstrated in Figure 2, where the daiy treatment verification data have been potted for two patients. The information shows that each patient has different systematic and random errors occurring during their treatment; Patient 1 has a sma systematic error but arger random errors, whie Patient 2 has a arger systematic error but smaer random errors. The random error for Patient 2 is characterised by the individua points being grouped more cosey around the mean (systematic error) position. These exampes aso demonstrate that more than one image must be acquired to distinguish between the systematic and random components and provide a good estimate of any correction to be appied. Actions based on a singe image must be undertaken with caution as it can ead to magnification of errors. For exampe, if the image of Patient 1 associated with the set-up error of 8 mm right and 5 mm inferior was used in isoation to correct subsequent treatments, a considerabe overcorrection woud be made. Further images taken and acted on independenty woud aso be subject to the same outcome eading to a series of unnecessary corrections around the mean position. For this reason, most off-ine porta imaging correction strategies acquire images over the first few fractions to provide a more accurate estimate of the mean. Ideay, a departments shoud determine their own popuation systematic and random set-up error components for each site-specific group. Section 4 of this report describes a practica method of performing this aong with worked exampes Set-up error and treatment margin Set-up errors and CTV-PTV geometric margins are interinked. Figure 3 (page 15) demonstrates the impact of systematic and random errors on CTV coverage. It demonstrates that random errors, which vary from day to day, ead to a burring of the cumuative dose distribution around the CTV, whereas systematic errors coud ead to a cumuative underdose to a portion of the CTV. Because of this atter effect, most of the CTV-PTV margin is needed to ensure adequate coverage from the various sources of systematic error. The CTV-PTV margin may be modified depending on the number of contributing errors that can be detected and corrected during the course of treatment. This wi be dependent on the treatment verification method used and which contributing error can be imaged. These may be summarised as foows. Off-ine imaging of bony anatomy Detects phantom transfer and patient set-up errors. Off-ine imaging of target As above pus the systematic error associated with target position and shape that can occur between deineation and first treatment. Onine imaging of bony anatomy Detects phantom transfer and patient set-up errors. Onine imaging of target As above pus the random and systematic errors associated with target position and shape. Athough onine and off-ine imaging of bony anatomy measure the same parameters, an onine approach measures the set-up error before treatment and enabes correction of the tota set-up error for that treatment; that is, systematic pus that day s random error. The target deineation error cannot be measured for an individua patient and is present for the treatment course. This and any other uncorrected errors shoud be incorporated into the geometric CTV-PTV treatment margin. These concepts are described in more detai in Section 4 aong with worked exampes. 14 On target: ensuring geometric accuracy in radiotherapy

14 Figure 3. The impact of geometric deviations on the dose distribution reative to the CTV. Random (treatment execution) deviations ead to a burring of the dose distribution. Systematic (treatment preparation) deviations ead to an unknown shift in the cumuative dose distribution reative to the CTV, 17 as shown in Figure 3, beow. The norma situation comprises a combination of systematic and random components. If eft uncorrected, the systematic error wi remain throughout the course of treatment potentiay compromising dose coverage to the CTV. Treatment verification concerns quantifying this unknown systematic deviation and ensuring it ies within acceptabe toerances. An off-ine correction strategy aims to quantify and correct for the systematic errors occurring over a course of treatment, so that ony random errors remain. CTV Ideaised dose enveope around CTV RANDOM DEVIATIONS ONLY SYSTEMATIC DEVIATIONS ONLY CUMULATIVE DOSES RANDOM & SYSTEMATIC SYSTEMATIC DEVIATION CORRECTED On target: ensuring geometric accuracy in radiotherapy 15

15 3 How to verify: creating geometric verification processes This section outines an evidence-based impementation process that can be adapted according to the needs of the individua radiotherapy department. When starting this process, the first step is to identify the scope of the verification practice needed for the oca cinica service. The equipment, management and protocos chosen wi vary according to the verification systems avaiabe within the department. With certain verification processes, the cinica workoad for the department may increase, and it is essentia that the resource impications are considered at an eary stage. Services can be panned to optimise the baance between best verification practices and workoad. The verification process chosen shoud consider the foowing: Equipment and technica infrastructure Personne, responsibiities and training Imaging protocos Image acquisition Frequency and timing of imaging Measurement of set-up errors Gross error Systematic and random error Toerances, action eves and correction strategies Dose considerations concomitant exposure Audit. 3.1 Equipment and technica infrastructure A cear method for verification shoud be decided by each department. This shoud start with identifying the equipment needed for verification (for reference image acquisition, treatment image acquisition, image matching and data storage), estabishing the connectivity methods between the equipment, and determining the quaity eves and tests needed to maintain verification standards Generation of reference and treatment images Good quaity images are essentia. CT image quaity is reated to sice width. These shoud be 5 mm or ess. The optima image quaity wi vary with different systems and shoud be chosen accordingy. 21 Where the DRR is of poor quaity, X-ray simuator reference images shoud be obtained. However, the risk of introducing a further source of systematic error by using this additiona step needs to be considered. Treatment image quaity shoud ideay have fine spatia resoution and high contrast with a high contrast-tonoise ratio. Acquisition methods shoud be optimised to produce the quaity required and may invove the use of image-processing software. The images shoud be of sufficient quaity to identify the foowing on both the reference and treatment image: Isocentre and/or fied edge Tumour surrogate (bone, soft tissue, impanted markers) Data transfer and storage Image storage and distribution shoud be considered before impementing verification processes as image fie sizes from eectronic systems can be arge. Transfer of data between the panning system, CT scanner, simuator, CT 16 On target: ensuring geometric accuracy in radiotherapy

16 simuator and treatment machine shoud be: Secure and maintain integrity of reference and verification data Efficient and reiabe transferred data shoud be avaiabe within a few minutes. The data coected (images and match anaysis) shoud be: Backed up and archived Stored so that historica data is readiy retrievabe Quaity assurance and image-matching accuracy Quaity assurance programmes shoud be created to ensure image quaities and verification data coection standards are reguary assessed and maintained. Each component of the verification process from the acquisition of panning data to the subjectivity in decision-making by individuas may have a certain eve of error or uncertainty within it. Ideay, these shoud be measured so that the overa accuracy of the verification process is known. This can be taken into consideration when assessing the vaidity of the image match data. This is an important measure when determining panning margins. 5 Verification systems shoud be assessed in terms of: Absoute accuracy how accurate is the dispacement information given from a particuar measuring system? This is tested by anaysing data containing a known dispacement User accuracy how do the measured resuts vary between users? Registration technique how is the accuracy of registration affected by the different agorithms avaiabe, such as tempate, fiducia and chamfer automated matches? Image processing some processing may affect the measured dispacement Consistency of output how accurate is the conversion of 2D data to 3D movements, and how imiting is the accuracy of couch movements? Consistency of co-ordinate systems, error reporting and corrective actions It is essentia within a department to deveop cear and consistent conventions for the reporting and correction of set-up errors. Consideration shoud be given to each of the possibe patient orientations on the treatment couch, ensuring that this information is dispayed on the images and incorporated into the offset measurements. Ideay, a singe co-ordinate system shoud be used across the department, specifying the isocentre position reative to a set-up point, stating x, y, z, transationa directions and rotation (pitch, ro, and yaw) around these axes. If such consistency between different systems that specify movement within a department is not possibe, other co-ordinate systems used in the verification pathway shoud be ceary documented, aong with the conversion process from one to another. Specify direction of movement. Specify what is moving, treatment fied or couch Verification data management software Software pays an important roe in automating the verification process. It must address the specific requirements of a department. 22,23 A recommended ist of functions that such software shoud provide is given beow: Enabe the design and impementation of imaging protocos that dea with the fu range of possibe outcomes Incorporate a patient-based database that aows a chosen protoco to be appied to an individua patient. This incudes highighting on the treatment unit when and which images are required for each treatment Aow the input of measured set-up errors Cacuate any required shifts based on the chosen protoco and convert to new isocentric co-ordinates (tabe positions) On target: ensuring geometric accuracy in radiotherapy 17

17 Incorporate any shifts correcty into the ongoing anaysis Cacuate popuation statistics Generate reports Anayse trends Enabe audit. This functionaity ies somewhere between that of an eectronic porta imaging and a record and verify (R&V) system and coud be incorporated within either. Ideay there shoud be a ink both to the imaging system for automated recording of set-up errors and to the record and verify system to use the patient management database. Summary. Equipment and technica infrastructure Images used shoud be of sufficient quaity to be abe to determine information required. Good connectivity is required. Cear conventions needed for co-ordinate system and error reporting. Quaity assurance (QA) is important for ensuring accuracy of data. The contribution of uncertainties in the verification pathway shoud be measured. The software packages used must be suitabe for managing the compete verification process. 3.2 Personne, responsibiities and training Responsibiity for verification shoud be carefuy assigned within each department. The foowing components shoud be considered Manageria responsibiities A verification team comprising senior experienced physicists, radiographers and cinicians, shoud undertake responsibiity for managing the verification process. It may be appropriate to have either one dedicated team or severa tumour site-specific teams. The foowing areas of responsibiities shoud be incuded: Design of the overa structure and process which incudes: Verification protocos Set-up error action eves Timescaes for action on verification anomaies Personne responsibiities and authorities Pathways for referra Action to take outside the protoco Impementation of training programmes Instigation of quaity contro programme Impementation of audit programmes Cinica responsibiities Day-to-day responsibiity for acquisition and approva of verification data wi rest with individuas and is based on training and competency assessments. Treatment deivery staff, usuay radiographers, are often considered best paced to make these decisions and, in particuar, onine corrections, as they have a the information regarding patient set-up and compiance at their disposa. 24 However, different skis are required for each aspect of the verification process and different personne may be used for a or part of it. These discipines coud incude: radiographers, physicists, cinica oncoogists and aso perhaps nonregistered staff (for exampe, assistant practitioners, cinica technoogists, dosimetrists), provided that they are 18 On target: ensuring geometric accuracy in radiotherapy

18 suitaby trained and authorised in accordance with ocay agreed protocos and any pubished guidance An individua may be permitted to undertake a or ony some of the tasks required, based on their competencies. The foowing areas of responsibiities shoud be incuded: Image acquisition Gross error detection and management Set-up error detection and correction Image manipuation and processing Making cinica decisions using the verification data. The number of independent checks needed when anaysing and making cinica decisions on verification data is dependent on the oca eves of training and competency assessment. This may vary between oca cinica settings and take into account changing roes and differing modes of service deivery. A risk anaysis is recommended to determine that the oca practices used are safe. 8 There must be agreement across the service as to who is responsibe for image approva and actioning. Specific recommendations are: If the procedure is carried out by a singe operator, s/he must be fuy trained to the eve required ocay and assessed as competent Two independent checks are required for technique audit Training and competency programmes A competency programme requires: Identification of each task in the verification process Determination of the knowedge and skis necessary for each task A forma training programme Assessment of competencies. The frequency of repeating competency assessments shoud be stated. Generay a maximum time interva of one year is required. However, it has been suggested that accuracy of quantitative anaysis depends on the amount undertaken 28 and refresher training may be required within this time. An outine programme is shown in Section 5 of this report. Summary. Personne, responsibiities and training Verification team required to set guideines on structure, process and responsibiities. Training and competency assessments are necessary. Action to take outside protoco to be cear. The number of independent checks needed for making cinica decisions is dependent on the eves of training and competence. A risk anaysis is recommended for each oca setting. 3.3 Imaging protocos An imaging protoco is simpy a series of instructions (or preferaby a fow diagram) that gives the user guidance on the appropriate images and action to be taken at a stages of the treatment course. It shoud identify and manage a possibe scenarios (what to image, when and how often) and wi normay incude one or more correction strategies. The protoco required wi vary between centres based on oca needs and practices and may differ with each anatomica site or technique or patient preparation method used. An imaging protoco shoud ceary indicate: The need to act immediatey for gross error The number of fractions aowed before action is taken on non-gross errors. On target: ensuring geometric accuracy in radiotherapy 19

19 3.3.1 Image acquisition If panar MV images are used in the geometric verification process, they can be acquired using various modes such as singe exposures (short or ong), doube exposures or movie oops, as described in Section 6. The method of image acquisition (for exampe, interfractiona or intrafractiona) wi vary for different anatomica sites and verification purposes. It is dependent on the anatomy visibe within a treatment fied and by how much that anatomy moves during irradiation. The need for irradiating norma tissue in order to make a decision on the accuracy of fied pacement shoud be baanced against the radiation dose given to these tissues. Where doube exposures are required, care shoud be taken to avoid unnecessary exposure to critica structures. This is outined further in the site-specific protocos in Section 7. Where possibe, fied pacement information shoud be acquired in a three transationa dimensions (x, y, z). If using panar images, acquiring an orthogona pair of images wi give dispacements that can easiy be used to cacuate the corrective couch movements required to reposition the patient. If using panar images obtained at non-orthogona anges, the corrective couch movements may sti be determined foowing appropriate correction. 29 Where software aows, information on in-pane rotations may aso be gained. If panar images are acquired from a singe direction, or opposing directions, the accuracy of the fied pacement can be assessed in two dimensions ony. Ideay when using panar imaging, the two images shoud be acquired simutaneousy to be truy representative of the patient s position. In practice, there is usuay a few minutes deay. This shoud be minimised wherever possibe Frequency and timing of imaging Images shoud be taken over a sufficient number of fractions to determine the set-up error. 30 The number required for each disease site/technique wi vary. Accuracy increases with the number of images obtained, but image acquisition over the entire treatment course is resource-intense 31 and may increase concomitant radiation exposure. The best distribution of image acquisition over the treatment course is uncear. The vaue of first day images has been debated since patient anxiety may give an unrepresentative picture. 32 One study has recommended that isocentre correction decisions shoud not be based on the first day except in the case of gross errors. 33 Other studies have suggested the worth of first day images depends on the efficacy of the immobiisation technique. Optima correction and imaging strategies are discussed beow and are summarised for individua site-specific protocos in Section 7. A exposures must be justified by a practitioner under The Ionising Radiation (Medica Exposure) Reguations (IR(ME)R). 34 Summary. Imaging protocos The image acquisition method to use wi depend on the quantity of stabe anatomy seen within the fied. Critica structures shoud be avoided when exposing norma tissue for data gathering. The timing of the image acquisition changes the information gathered. For panar imaging, an imaging set (minimum two orthogona images) acquired in as sma a time frame as possibe, is needed to verify accuracies in a three directions. The accuracy of assessing the systematic error increases with the number of image sets acquired. In genera for radica treatments, three to five imaging sets are needed to assess the systematic error. The number of imaging fractions wi depend on site and treatment technique. The actua number of images wi aso be infuenced by perceived risk of additiona radiation exposure and resources. 20 On target: ensuring geometric accuracy in radiotherapy

20 3.4 Measurement of set-up errors Identification of gross error Gross error determination must be undertaken in a centres before the first treatment fraction is deivered. Ideay, this shoud be on the treatment unit as set-up errors may go undetected when using pretreatment simuation aone. Gross error must be determined by one of the foowing methods, in order of preference. Undertaking a rapid estimate of gross error on the first treatment fraction, immediatey prior to treatment deivery. A quick check for gross error can be made by interrupting the treatment as soon as an eectronic image has been acquired and reviewing it for fied size and orientation, shieding and approximate position accuracy. Using the first treatment unit session for verification aone. Using the simuator or CT scanner to provide pretreatment verification. Images must sti be taken on the first fraction on the treatment unit and reviewed before the next fraction. Where a beam arrangement cannot be accuratey visuaised on imaging, gross error can be investigated by reviewing the ight fied in reation to surface anatomy. This is often used for eectron or skin treatments and vertex fieds Determining systematic and random errors for individua patients The systematic component of the set-up error of a individua patients shoud be determined for a muti-fraction courses (>5 fractions) so that it can be minimised for the rest of the treatment. For three-dimensiona treatment pans, verification data shoud be obtained from two imaging panes. Where possibe, verification data shoud be acquired from an orthogona image pair. This enabes corrections to be easiy resoved into the required treatment couch adjustments (Section 3.3.1). Anatomy, to which comparisons can be made, can be identified on the reference images and contoured using drawing toos. At east three features are required for most automatic image registration software. These same features or measurements can then be compared with each treatment image. These anatomica or surrogate reference points shoud be defined and be constant for any given site. If port fims are used, these images shoud be digitised and compared eectronicay. If a digitisation method is not avaiabe, fim and reference images can be marked and measured manuay, but the data wi contain a greater error component. The eve of observer error shoud be determined to assist decision-making. Out-of-pane rotations are difficut to measure and require the use of speciaist software. These measure the discrepancy in magnification in different parts of the image. Currenty, software gives an indication of the presence of an out-of-pane rotation but is unabe to quantify it, athough it is possibe to create agorithms to do so. This is of importance as out-of pane rotations affect the accuracy of the two dimensiona transationa dispacement. 35 Further deveopments are to be expected in this area. Summary. Measurement of set-up errors Gross error detection on the treatment unit is an essentia requirement prior to first treatment. Individua patient systematic set-up error shoud be measured and minimised. 3.5 Toerances, action eves and correction strategies Toerances The toerance of any measurement or parameter may be defined as the permitted observed variation in that measurement or parameter. Toerance eves set the optimum conditions based on a therapeuticay desired vaue, athough these may not be enforceabe or achievabe in a circumstances. On target: ensuring geometric accuracy in radiotherapy 21

21 In the case of treatment verification, the toerance wi be the permitted range of set-up error from the reference point. For exampe, we might aow a 5 mm eft/right toerance for a measured set-up. This wi mean that the measured set-up error in the eft/right direction may range from 5 (eft) to +5 mm (right) from the reference (desired) position. For treatment verification, the chosen toerance of a particuar measure of set-up error is reated to the margins used for treatment panning, the aim being to aways maintain adequate dose coverage of the CTV (Section 4). Input data to cacuate both CTV-PTV margins and set-up toerances may be derived from a porta imaging study undertaken for the particuar treatment site and technique. 5 For radiotherapy centres that are sti deveoping their treatment verification protocos, or are commissioning new treatment techniques (incuding equipment), data summarised by Hurkmans 36 and in Section 7 shoud form a good starting point. The toerances used in practice wi take into account severa other factors, 5 incuding: Immobiisation method Toerances in equipment movement and set-up Interna organ motion. Each radiotherapy centre shoud define toerances for each anatomica site and treatment technique used. Information on how these studies may be conducted is presented in Section 4. Individua patients may require specia consideration; for exampe, those in pain, the anxious or the obese. The toerance to be used in an imaging protoco therefore depends on: Anatomica site Treatment technique CTV-PTV margin Patient compiance Action eves An action eve for a measurement or parameter is the point at which action is necessary. Action eves set minimum conditions, beyond which performance is deemed unacceptabe. The action to be taken depends on the importance of the parameter and the risk of not making any ateration. For instance, one may set a series of action eves whereby for a sma deviation from the toerance, observation is recommended. For a arger deviation, immediate amendment is required. Actions in treatment verification wi incude: Further imaging (in case the resut was simpy random) Re-assessment of any systematic error Immediate modification such as stopping and adjusting the patient s position if the measurement is at a eve which may be regarded as a gross error. The magnitude of the chosen action eve is reated to panning margin, frequency of imaging, the random (daiy set-up) error, and the chosen strategy used to constrain the set-up error over the treatment course. For exampe, in the case of a singe acquired image, an action eve of twice the standard deviation of the random set-up uncertainty may be used. 5 The action eve for immediate intervention (that is, a gross error) may be set at three times the random set-up uncertainty that wi typicay come to around 1 cm for a wide range of treatment techniques Corrective strategies The key requirement in any imaging protoco, apart from gross error detection, is to provide an accurate estimate of systematic set-up error. Depending on action eve, the chosen correction strategy can then be used to remove this error. The protoco must confirm any appied correction is vaid and remains so for the duration of the treatment. Figure 4a (page 23) shows the effect of a poor correction strategy. The bue trianges indicate the set-up error measured from a porta image. The red ine indicates the systematic set-up error, cacuated as the average 22 On target: ensuring geometric accuracy in radiotherapy

22 dispacement over four days. For this patient, the systematic set-up error (SSE) has been determined as being the argest (or ast) vaue seen over the four days and a correction appied to that vaue. This method is repeated for the next four sets of images. The resut is an exaggerated osciation in the accuracy of the patient set-up. Figure 4b (page 23) shows the outcome on set-up if the SSE had been correcty cacuated and actioned. Figure 4a. Effect on set-up of poor correction strategy Set-up error during radiotherapy course poor correction protoco Set-up error (mm) Dispacement (mm) SSE per cacuation set Figure 4b. Outcome on set-up if the SSE is correcty cacuated and actioned Set-up error during radiotherapy course good correction protoco Set-up error (mm) Dispacement (mm) SSE per cacuation set Assessment and correction of systematic errors It is important to assess the SSE accuratey within an appropriate number of fractions so that (a) a robust estimate of the true systematic error can be made, whie (b) the minimum number of fractions is deivered with the systematic error present. Two possibe approaches are the no action eve (NAL) and shrinking action eve (SAL) correction strategies. The NAL 37,38 is most commony advocated for radica treatments. It invoves the systematic error being On target: ensuring geometric accuracy in radiotherapy 23

23 cacuated after 3 4 fractions and a correction performed which is the tota magnitude of the systematic error, regardess of the toerance for that treatment site. Since the NAL approach does not define an action eve for corrections, there is aso a subpopuation of patients in whom the systematic error is so sma that appying a correction woud be impractica; for exampe, moving the couch <2 mm. It is suggested that ony systematic errors of >2 mm shoud be corrected. The extended NAL protoco (enal) incudes once-weeky imaging in addition to imaging in the first 3 4 fractions if the resut is within toerance there is no action. If out of toerance, further images are obtained to determine systematic error, see page 9. This is usefu in detecting trends and systematic changes to the patient s set-up over the treatment course. 39 The NAL protoco does not act on gross errors. Such errors shoud be corrected before a further fraction is given. The SAL 40 uses an action eve that reduces according to the number of fractions imaged. The running mean error over a acquired images is compared with the current action eve and treatment set-up adjusted by this amount if the discrepancy exceeds the action eve. The fina action eve is determined by the initia action eve and number of images considered appropriate for the particuar technique. The SAL avoids a set-up being corrected prematurey, where discrepancies observed at the start of treatment may have arisen through random rather than systematic error. A disadvantage of the SAL is that foowing any correction, the process is restarted and information obtained prior to the restart is ost. 20 A direct comparison of the SAL and the NAL protoco using an average of ten imaged fractions per patient found the NAL protoco to be more efficient in terms of number of images per reduction in systematic error. 38 Figure 5 (page 25) iustrates the difference between toerance and action eve using a NAL protoco. Summary. Toerances, action eves and corrective strategies Gross errors shoud be acted upon immediatey. Each radiotherapy department wi need to evauate their own toerances and action eves. A imaging protocos must incude correction strategies for a treatment sites. A no action eve strategy wi correct the systematic component of the set-up error after at east three fractions. A shrinking action eve strategy uses an action eve which decreases according to the number of fractions imaged, in order to remove the systematic component of the set-up error. Workoad may increase as a consequence. A corrections appied to the treatment set-up must be verified. Weeky imaging is recommended in addition to the correction poicy. 3.6 Dose considerations concomitant exposure Concomitant exposures are defined as a exposures within the course of radiotherapy other than treatment exposures. 37 Exampes are: Exposures for re-simuation or repeated CT scans during the radiotherapy course A non-treatment fieds Large, open segments for doube exposures Orthogona, open fieds used purey for imaging Exposures from nove image-guided techniques (panar views) Orthogona radiographs (in-room kv, in-room CT [scout view], kv CBCT [panar imaging], tomotherapy [scout view]) Exposures from nove image-guided techniques (voume imaging) In-room CT IGRT MV CBCT IGRT kv CBCT IGRT Tomotherapy MV IGRT. The Ionising Radiation (Medica Exposure) Reguations (IR(ME)R) 34 requires that a medica exposures shoud be justified. The practitioner must ensure that exposures of target voumes are individuay panned, taking into account that doses to 24 On target: ensuring geometric accuracy in radiotherapy

24 Figure 5. Changes in set-up depicted during a fractionated course of radiotherapy to iustrate the concepts of gross error, toerance and action eves using a NAL protoco (read eft to right, top to bottom). A - Fraction 1 B - Fraction 1 Ref Ref AL -G -T +T AL G AL -G -T +T AL G Onine verification image taken with the first few MU of the fied on day 1 (first fraction), and anaysed prior to deivering rest of exposure. Gross error found; action taken immediatey Error identified and set-up adjusted repeat image taken; Ok to continue with rest of exposure C - Fractions 2-5 Ref D - Fraction 6 Ref AL -G AL -NAL AL +NAL -T +T AL G AL -G -T +T AL G Four more images taken (on fractions 2 5) for NAL protoco to cacuate systematic component of the set-up error for this particuar patient. Off-ine verification protoco. AL NAL (or AL -NAL ) Action Leve for the NAL Protoco Systematic component of the set-up error cacuated and reference moves adjusted for the patient. Image taken on fraction 6 (week 2). Within toerance E - Fractions 11 and 16 Ref F - Fraction 21 Ref AL -G -T +T AL G AL -G -T +T AL G Weeky imaging (fractions 11 and 16, weeks 3 and 4). Both within toerance Weeky imaging (fraction 21, week 4) is out of toerance (beyond the weeky imaging action eve). Action taken check set-up instructions and annotations and repeat the verification image at the next fraction G - Fraction 22 Ref H - Fraction 22 Ref AL -G -T +T AL G AL -G -T +T AL G Repeat imaging (Onine protoco) (fraction 22, week 4) is sti out of toerance. Action taken re-check set-up instructions and annotations. Error in set-up moves identified corrected Further image taken OK. Continue with the rest of fraction I - Fractions 26 and 31 Ref AL -G -T +T AL G Weeky imaging (fractions 26 and 31, weeks 5 and 6). Both within toerance KEY T Toerance AL G (or AL -G ) Action eve for gross errors On target: ensuring geometric accuracy in radiotherapy 25

25 non-target voumes and tissues sha be as ow as reasonaby practicabe and consistent with the intended radiotherapeutic purpose of the exposure. Most imaging protocos for treatment verification are sufficienty detaied to incude methods of image acquisition, recommended machine units (MUs) for acceptabe image quaity, standard frequency of imaging (for the idea set-up) and compensation for target dose deivery from verification exposures but may fai to ink sufficienty these exposures with concomitant dose to non-target tissues. For practica purposes, the medica and denta guidance notes 41 advise that the IR(ME)R practitioner responsibe for treatment can justify the concomitant exposures at the outset or during the radiotherapy course, and in doing so must be aware of the ikey exposures and the resuting dose so that the benefit and detriment can be assessed. This can be achieved by incuding ikey concomitant exposures within site-specific protocos with an effective dose agreed. Estimates of the magnitude of concomitant exposure to non-target tissues, especiay organs at risk shoud therefore be incuded in a verification imaging protoco under idea and non-idea conditions It is important to emphasise the benefit from imaging which wi permit accurate, geometric verification of the irradiated voume which can then permit reduction of treatment margins and reduced norma tissue exposure, hence achieving increased compiance with the IR(ME)R requirements. 12 A practitioners and operators shoud be aware of the doses invoved in reevant radiotherapy procedures in order to justify exposures based on the benefit over risk. 34,42,44,49,51,52 Steps shoud be taken to reduce the concomitant exposure dose wherever possibe; this may be as simpe as using taiored, asymmetric fieds for doube or singe open fied exposures (so that ony desired anatomy is visibe). For more compex techniques and in particuar emerging technoogies, it may require more detaied assessments of true verification doses. 42,45,46,48,50,53 56 A sources of non-therapeutic and concomitant exposure shoud be propery understood throughout the radiotherapy process. These wi incude doses from: CT scanning (panning scans etc) Pretreatment verification procedures Porta imaging (especiay non-target tissue doses) Transmission through the MLC during treatment deivery Leakage and scatter from the inac during treatment deivery Compex therapeutic procedures (such as dynamic IMRT) Nove image-guided practices. The roe of verification techniques which do not require radiation exposure shoud aso be considered, such as utrasound-based techniques, video-based surface tracking and most recenty wireess detection of impanted transponders (see Section 6). Some imaging techniques wi sti use ionising radiation, and the concomitant dose wi aso depend upon the treatment technique itsef. For exampe, for treatments which use impanted soft-tissue markers, fieds used for imaging do not need to be enarged to see sufficient anatomy; the fied sizes need ony be equivaent to the treatment fieds themseves. The risks in terms of secondary cancer induction and increased toxicity for organs at risk in the vicinity of the PTV need to be carefuy weighed against the overa benefits. This is especiay so for modern techniques which increasingy use technoogica advances ike CT simuation, IMRT and IGRT. 42,43,45 47 Concomitant exposure may need further consideration for organs at risk, particuary those near to the PTV which are aready cose to acceptabe toerance from the panned dose distribution. 49,57 Where critica structure doses are potentiay important, imaging frequency and acquisition method imitations may be imposed at the time of panning, based upon the overa TP and cacuated dose-voume-histograms (DVHs). For secondary cancer induction, doses estimated from idea and non-idea imaging protocos (incuding scatter and eakage from the inac itsef) may be used to cacuate effective doses deivered and therefore the subsequent probabiities. 42,44,45,57 61 Estimates of typica non-target doses may be made from simpe cacuations 42,49 or through panning computer modes and measurements. 48,50 An exampe of a simpe approach is shown in Box 1, page It is dependent upon the type of imaging protoco being used, which in turn is anatomica site-dependent. Different imaging 26 On target: ensuring geometric accuracy in radiotherapy

26 technoogies wi require different concomitant exposures for adequate imaging purposes. 46,55,56 The exampe shown is for iustration of a method which may be used to discuss and quantify the doses and their possibe cinica effects. It uses representative effective doses and probabiities for fata cancer induction; 50 the associated caveats for those figures shoud be carefuy noted by the reader. A simiar tabe may be formaised for, say, cone-beam and other IGRT procedures, derived from the appropriate dose assessments for those imaging methods 46,55,56 and their expected imaging frequencies throughout the treatment course. Limits can be appied to frequency and number of repeat exposures in imaging protocos by estimating the second cancer risk from cacuated doses to non-target critica structures derived from the recording and monitoring of effective doses from a concomitant exposures. An action eve may then be set of 100 msv, for exampe, for the chance of second cancer induction to be 0.5%, compared with the natura ifetime risk of approx 25%. 50 Any consideration of the impact of concomitant exposures must aso be set in the context of the cinica situation; the actua risk of second maignancy is ceary greater in a young woman having treatment for Hodgkin s ymphoma or breast cancer with a surviva and risk period of severa decades than an edery man having treatment for advanced ung cancer having a prognosis of a few months. Summary. Dose considerations Concomitant exposures shoud be considered with every verification procedure and kept to a minimum consistent with acquiring the necessary information to deiver accurate treatment. It is better to image appropriatey to ensure adequate and accurate coverage of the PTV than to avoid imaging for fear of a minima risk associated with the additiona exposure. Box 1. Hypothetica protoco for radica pevis imaging Basic assumptions Six-week radica course of treatment AP and atera images used throughout (constituting an imaging session) Images are approximatey 16 x 16 cm, 6 MV, singe open-fied exposures Isocentric depth approximatey 10 cm Imaging frequencies Routine: AP and atera images acquired once a week Imaging technoogies Amorphous Siicon (asi) Difficut set-up case: Assume three imaging sessions needed in first week; weeky thereafter 39 No action eve (NAL) protoco: Assume five imaging sessions in first week; no imaging thereafter 38 asi EPIDs: 4 MUs per image, 8 MUs per imaging session Fim or oder EPID: 8 MUs per image, 16 MUs per imaging session Concomitant doses EPID Fim/oder EPID Isocentre dose per session 0.07 Gy 0.14 Gy Routine imaging 0.42 Gy per course 0.84 Gy per course Difficut set-up case 0.56 Gy per course 1.12 Gy per course NAL protoco 0.35 Gy per course 0.70 Gy per course Effective doses (adut mae): Routine imaging 16 msv per course 32 msv per course Difficut set-up case 21 msv per course 42 msv per course NAL protoco 13 msv per course 26 msv per course Chance of second cancer induction: Routine imaging 0.08% 0.16% Difficut set-up case 0.11% 0.22% NAL protoco 0.07% 0.14% On target: ensuring geometric accuracy in radiotherapy 27

27 3.7 Audit Audit is an essentia process. It is used to assess both the success of the impementation of the verification processes and measure the standards of the ongoing service provided. Reguar audit of practice assesses the continuing effectiveness and appropriateness of the verification processes and gives the opportunity to update protocos in ine with current evidence. Guidance in creating audit programmes is given by the Nationa Institute for Heath and Cinica Evidence (NICE) 62 and the NHS Cinica Governance Support Team. 63 Cinica audit evauates services against a standard that has aready been set by examining: Whether what ought to be happening is happening Whether current practice meets required standards Whether current practice foows pubished guideines Whether cinica practice is appying the knowedge gained from current evidence. Usefu assessment criteria for the verification process woud incude identification of, adherence to and effectiveness of: The management process whether there was a designated verification team and whether there was a singe team for the department or it was tumour-specific and the components of that team in terms of specifying the discipines invoved Quaity assurance programmes for maintaining the safety and accuracy of verification equipment Protocos for image acquisition and anaysis for each tumour site Action eves and correction strategies and responsibiities within these protocos Protocos for incorporating the set-up accuracies measured in margin cacuations for common treatment sites where verification is particuary reevant to the appication of highy technica radiotherapy; for exampe, prostate cancer and head and neck cancer Training and competencies programmes, preferaby benchmarked to standards set out esewhere Incusion of peer review standards into any audit once these have been revised Adherence to IR(ME)R reguations Use of audit to evauate imaging protocos The recommendations made within this document shoud be evauated for their appicabiity to oca practice. It is suggested that each anatomica site shoud be seected for detaied review of imaging protocos. Specificay the procedures for image acquisition, the quaity of images and their subsequent use for treatment modification shoud be audited Use of audit to evauate techniques and immobiisation systems Under the site-specific sections of this document, patient immobiisation is incuded in the suggested protocos for breast, brain, head and neck, prostate, thorax and paediatrics. If these suggestions are foowed, it is recommended that a oca audit shoud be carried out to assess their appicabiity. Initiay, this might be confined to the major sites of breast, prostate and thorax for which there is the most supporting evidence, giving greater strength to the recommendations. Reguar audit of the set-up data provides a method for assessing treatment techniques and the effectiveness of immobiisation systems. Set-up accuracies can be monitored over time, to identify trends or changes in standards. Comparisons can be made between different immobiisation systems or techniques to assess the benefit of one method over another Identification of systematic errors in the departmenta processes Auditing popuation data for each treatment site can identify probems in the treatment process; systematic errors detected in a patients for a particuar treatment technique can indicate a probem with that technique. Summary. Audit Reguar audit is necessary, assessing process, management, training, protocos, equipment and techniques. 28 On target: ensuring geometric accuracy in radiotherapy

28 4 Derivation of systematic and random set-up errors and reationship to the CTV- PTV margin This section describes: How the random and systematic set-up errors for a group of patients may be derived from porta images. An exampe is presented to demonstrate the method The interreationship between CTV-PTV margin and treatment verification. 4.1 Set-up error The set-up error (Δ) is defined as the deviation between actua and expected position, normay cacuated as a shift in the isocentric position when an image is compared against its corresponding reference. For ease of anaysis, and more importanty subsequent set-up correction, the set-up error shoud be resoved into orthogona directions (for exampe, A P and S I from a atera image and R L and S I from an anterior image). Conversion of these measurements into the required co-ordinate axes may be made if the acquired images are not orthogona. 29 It is crucia that vector quantities are cacuated so the correct direction information is maintained. For exampe, if shifts in the anterior direction are given a negative sign then those in the posterior direction are aways positive. Necessary de-magnifications must be appied to a measurements at this stage. Definitions and causes of random and systematic errors are given in Section 2.3. The equations used to cacuate these are given beow and spit into two basic forms, those cacuating a mean and those a standard deviation (SD). The SD is a measure of how widey vaues are dispersed from the mean vaue and in this context defines the size of the error. The term treatment popuation is used to represent a patients treated with a specific technique (treatment site and immobiisation method). The set-up errors for this popuation are estimated by cacuating the errors for a group of patients whose resuts are assumed to accuratey represent those of the popuation from which they are drawn Systematic set-up errors Individua mean set-up error The systematic error (m individua ) is the mean set-up error for an individua patient. It is cacuated by summing the measured set-up error for each imaged fraction (Δ 1 + Δ 2 + Δ 3 ) then dividing by the number of imaged fractions (n). This can be expressed by the formua: m individua = Δ 1+ Δ 2 + Δ Δ n n (E1) Overa popuation mean set-up error The overa mean set-up error (M pop ) is the overa mean for the anaysed patient group and shoud ideay be zero. Significant departures from zero indicate an underying error common to this patient group and requires investigation. This parameter is a strong indicator of the efficacy of any given treatment technique and is often omitted. The equation is essentiay the same as equation 1 with the means for each individua patient (m 1, m 2, m 3 ) now being summed and the tota divided through by the number of patients in the anaysed group (P). M pop = m 1+ m 2 + m m p P (E2) On target: ensuring geometric accuracy in radiotherapy 29

29 Popuation systematic error The systematic error for the popuation ( set-up ) is defined as the SD (spread) of the individua mean set-up errors about the overa popuation mean (M pop ). It is cacuated by summing the squares of the differences between the overa popuation mean derived from equation 2, and each individua patient mean derived from equation 1, in turn. Note that the resutant sum is divided by the number of patients minus one and the square root of the resutant vaue is required to give set-up. 2 set-up = (m 1 M pop ) 2 + (m 2 M pop ) 2 + (m 3 M pop ) (m n M pop ) 2 (P 1) (E3) Random set-up errors Individua random error For each individua, the interfractiona random (daiy) set-up error (σ individua ) is the SD of the set-up errors around the corresponding mean individua vaue (m) derived from equation 1. It is cacuated by summing the squares of the differences between the mean and set-up error from each image in turn. Note that the resutant sum is divided by the number of images minus one and that the square root of the resutant vaue is required to give σ individua. 2 σ individua = (Δ 1 m) 2 + (Δ 2 m) 2 + (Δ 3 m) (Δ n m) 2 (n 1) (E4) Popuation random error The popuation random error (σ set-up ) is the mean of a the individua random errors (σ 1, σ 2, σ 3.). This equation assumes that the number of images acquired per patient is identica or that the ikey differences wi have minima effect on the fina resut. σ set-up = σ 1 + σ 2 + σ σ p P (E5) 30 On target: ensuring geometric accuracy in radiotherapy

30 Worked exampe Using equations 1 to 5 incusive, a worked exampe is presented in Tabes 1 to 4, pages 31 and 32. A measurements are in miimetres. Tabe 1. Individua set-up errors (Δ) acquired from three patients. The data presented do not incorporate any corrections made during the course of treatment. Anterior fied Latera fied Patient Image fraction R L S I S I A P ND ND ND ND ND ND = No data measured. On target: ensuring geometric accuracy in radiotherapy 31

31 Tabe 2. Individua mean and random set-up errors for Patients 1, 2 & 3 derived from the data in Tabe 1 using equations 1 and 4 Orthogona directions Patient Mean error R L S I (Ant) S I (Lat) A P 1 m m m Patient Random error R L S I (Ant) S I (Lat) A P 1 σ σ σ Tabe 3. The mean set-up errors for a ten-patient group, incuding Patients 1 3 from above, are given beow aong with the cacuated overa popuation means (equation 2) and the resutant popuation systematic set-up errors in each orthogona direction (equation 3) Orthogona directions Patient Mean error R L S I (Ant) S I (Lat) A P 1 m m m m m m m m m m Overa mean M pop Popuation systematic set-up errors Σ set-up Tabe 4. Individua random errors for the ten-patient group, incuding Patients 1 3, are given beow aong with the cacuated popuation random set-up errors in each orthogona direction (equation 5) Orthogona directions Patient Random errors R L S I (Ant) S I (Lat) A P 1 σ σ σ σ σ σ σ σ σ σ Popuation random set-up error σ set-up On target: ensuring geometric accuracy in radiotherapy

32 Comments on the exampe Athough some reported studies have anaysed images from as few as ten patients, it has been shown 37 that sma patient studies of this size can resut in a arge uncertainty in the popuation systematic set-up error. The random set-up error is ikey to be more accurate even for sma patient numbers as ong as sufficient images are acquired per patient and inter-patient variabiity is not excessive. It is recommended therefore that at east 20 patients are incuded in a study with at east five images per patient. The equations above appy equa weight from each patient to the overa resuts rather than the number of images acquired per patient. This is the generay accepted method Summary. Why perform a porta imaging study? To determine the overa mean set-up error. This is a strong indicator of any unwanted systematic component acting on a patients within the anaysed group. To aid in the design of a porta imaging correction strategy using evidence-based action eves. For comparison with simiar immobiisation and treatment techniques in the iterature. To provide oca input data into CTV-PTV margin cacuations or used as evidence to support currenty appied margins. 4.2 Reationship between the CTV-PTV treatment margin and treatment verification Effect of correction protocos on margins This section describes the interreationship between treatment margins and treatment verification. A generic CTV- PTV margin cacuation recipe is described and the input data required for this cacuation are discussed. The systematic and random set-up errors derived from a porta imaging study may be fed into this margin recipe but these are not the ony contributing components. The use of a treatment verification protoco to contro set-up errors may be used as a basis to reduce or justify currenty appied margins. The eves of margin reduction achievabe are based on the type of imaging protoco undertaken (see Section 2.3). This is summarised in Figure 6 opposite and expained in more detai in the exampes beow. Three cases are presented in Figure 6 opposite to iustrate this. Case A. No correction to measured set-up errors. The cacuated margin must incude a contributing sources. Case B. An off-ine protoco imaging bony anatomy is used to correct set-up errors. This enabes correction for the phantom transfer and systematic patient set-up components of the treatment preparation error (see Section ) and may be used to justify a reduction in the corresponding margin. By definition, an off-ine protoco appies any correction at the next fraction and so cannot account for the random patient set-up variations occurring from one fraction to the next. The contribution to the treatment margin from the treatment execution errors therefore must remain unchanged between Case A and B. Case C. An onine protoco imaging the target is used to correct set-up errors. This requires imaging, anaysis and set-up correction before each fraction and compared to Case B wi additionay detect the random patient set-up error as we as both errors associated with variations in target position and shape. This may be used as a basis to further reduce the random (treatment execution) and systematic components of the margin. On target: ensuring geometric accuracy in radiotherapy 33

33 C Onine correction Margin B Off-ine correction Margin CTV CTV A No correction PTV Extent of CTV Systematic (treatment preparation) errors Random (treatment execution) errors Figure 6. The CTV to PTV margin is governed by the combined systematic (treatment preparation) and random (treatment execution) errors occurring from a possibe sources (see Section 2.3 for definitions). The appication of treatment verification with imaging protocos designed to quantify and reduce some of these contributing sources can provide justification to reduce the appied margin Margin derivation It is beyond the scope of this document to discuss the derivation and cacuation of CTV-PTV margin cacuations in detai. Severa popuation-based margin cacuation recipes have been proposed. 17,18,68 These address the differences between treatment execution and treatment preparation errors and how these are to be combined to produce an appropriate margin. A of these margin cacuation recipes can be expressed as foows: CTV - PTVmargin = a + bσ + c (E6) where and σ are the combined sum of the SDs of a contributing systematic and random errors respectivey (Section 2.3) and a, b and c are constants. The constant c is incuded to account for parameters that affect the margin in a inear manner, such as breathing. 69 The two constants a and b characterise the reative contributions of the systematic and random components and these depend on factors such as the beam arrangement and chosen coverage probabiity. 17,68 Typicay a is 3 4 times greater than b and is generay much arger than σ indicating that the key contributor to the margin is the combined systematic error. 70 Figure 3, (page 15) in Section 2.3 demonstrates the reative effects systematic and random errors have on the cumuative dose to the CTV. The combined systematic error incudes a possibe sources of error as described in detai in Section 2.3. The SDs of these four contributing sources ( deineation = target deineation, motion = target position and shape, transfer = phantom 34 On target: ensuring geometric accuracy in radiotherapy

34 transfer and patient set-up = patient set-up error) are assumed to be normay distributed and independent of each other, and may be combined in quadrature (equation 7) to produce the combined systematic error. Recent work 71 suggests that deineation may require handing in a different way than the other components and needs an aternate theoretica approach. For the purposes of the anaysis beow, it is assumed to be normay distributed = deineation + motion + transfer + patient_set-up (E7) The components contributing to the combined random error are σ patient set-up and σ motion where σ patient set-up is the random patient set-up error and σ motion the random variation in organ position and shape (except breathing). These two components can aso be combined in quadrature in a simiar manner to equation 7 to give the combined treatment execution error σ. It can be seen that a porta imaging study wi provide important input data to the margin cacuation for both the random and systematic components Margin contro A porta imaging protoco designed to reduce set-up errors may be used either as a basis to reduce margins or to justify currenty used margins. The question is, what components reevant to the CTV-PTV margin wi be controed and what is the ikey effect on the overa margin? Exampe 1. Contro of systematic components For a given treatment popuation, it is decided to impement a protoco designed to correct mean set-up errors greater than 2 mm. For an uncorrected patient group, a porta imaging study for this popuation reveas a systematic set-up error ( set-up ) of 3 mm. Appication of an off-ine correction strategy has the effect of reducing the accumuated contributions of both the patient set-up and the phantom transfer error (see Section 2.3). It has been shown that correcting the mean set-up over the course of treatment to within + X of the expected position gives a theoretica approximation to this combined SD of X/ This corresponds to 1.2 mm for the exampe of X=2 mm. Tabe 5 gives representative vaues for the contributing systematic components for a prostate treatment 73 and shows the combined systematic error for the uncorrected and corrected cases. Tabe 5. The effect of an off-ine correction strategy on the systematic components of the CTV-PTV margin Systematic errors (mm) No correction Correction deineation 2 2 motion 3 3 transfer 3 Combined error = 2/ 3 = 1.2 patient set-up 3 (sum in quadrature, see Equation 7 ) There is, therefore, a reduction from 5.6 to 3.8 mm in the combined systematic set-up error as a resut of empoying a correction protoco. For a typica margin recipe, 17,18 the constant a has a vaue of 2.5 eading to a theoretica margin reduction of 2.5 x 1.8 = 4.5 mm for this exampe. Foowing fu impementation of the correction protoco, a porta imaging study repeated on the corrected patient group data shoud give a theoretica vaue coser to set-up = 1.2 mm. This exampe demonstrates how a correction strategy designed to imit the mean set-up error constrains the combined effects of patient set-up and transfer and can ead to a reduction in the cacuated CTV-PTV margin Exampe 2. Contro of random components Measured random patient set-up error can be controed on an individua basis using onine imaging. This assumes that the daiy set-up variation can be assessed and corrected before any further intrafractiona movement occurs. Direct or indirect imaging of the target coud aso enabe correction for target position and On target: ensuring geometric accuracy in radiotherapy 35

35 shape, σ motion. It shoud be noted that the systematic patient set-up and phantom transfer errors wi aso be corrected with an onine strategy, just as they are with an off-ine approach. On a popuation basis, random errors can be reduced by improved immobiisation, adherence to set-up protocos and training. This document is concerned with treatment verification and impicit within this is the contro of set-up errors. Figure 6 and the two exampes above demonstrate the reationship between this process and the CTV-PTV geometric margin (Equation 6). Any such approach used to justify margin reduction shoud be appied with caution and not used to override cinica judgement. Summary. Verification and treatment margins The magnitude of the CTV-PTV margin is argey governed by the combined systematic (treatment preparation) errors. An off-ine porta imaging correction strategy can be used to constrain the patient set-up and phantom transfer components of the preparation error. An onine correction strategy can aso contro random patient set-up errors. An onine correction strategy capabe of detecting target position can additionay contro both the systematic and random errors associated with organ motion. 36 On target: ensuring geometric accuracy in radiotherapy

36 5 Training and competency assessment 5.1 Training and competency assessment Training and maintaining competency is an integra part of the eectronic porta imaging (EPI) process. 74 A training programme wi ensure that each individua is trained to a consistent eve which can reduce inter-observer variabiity. 75 The content of a training and competency manua coud incude the foowing Training manua covering the foowing subjects Abiity to acquire images. Understanding the use of different acquisition modes. Abiity to create anatomica tempates from a reference image. Knowedge of appropriate anatomy to use for each site. Abiity to perform a matching technique resuting in a record of isocentre dispacement. Abiity to anayse images with awareness of in-pane rotations. Abiity to appy current isocentre correction methods under protoco. Knowedge of systematic and random errors. Knowedge of quaity assurance necessary on the EPI systems. Knowedge of data movement/archiving/deeting and retrieving fies Competency assessment methods The aims of the competency assessment shoud be to: Identify current eves of knowedge and ski Cinica competency knowedge and skis As covered in training manua above Further competency assessment for advanced practice Abiity to train other members of staff. Knowedge of different isocentre correction methods. Abiity to identify times when and what type of protoco shoud be used for an individua patient, such as onine imaging for a patient demonstrating arge random errors. Knowedge of dosimetric considerations; such as, identify if repanning is required. Abiity to identify causative factors reating to dispacements; for exampe, immobiisation causes or patientspecific causes. Awareness of imitations of EPI. Management of quaity assurance programmes. Understanding of risk/benefit of ionising radiation. Document reevant experience and refection on decision-making a) Outine objectives to achieve the required eve of competence b) Record of cinica competencies and reated refections on practice. On target: ensuring geometric accuracy in radiotherapy 37

37 6 Equipment used for geometric verification 6.1 Reference images Image type DRRs digitay reconstructed radiographs Exampe DCRs digitay composited radiographs 76 Simuator images fim (digitised) Simuator images digita image intensifier Simuator images fat pane 38 On target: ensuring geometric accuracy in radiotherapy

38 6.2 Image acquisition modes Acquisition Mode Comment Exampe Singe exposure short Simpest image taken at the start of the treatment fied. Fine if sufficient anatomy within fied to assess set-up errors. Good for interventiona (onine) porta imaging, where judgements can be made before deivering the rest of the treatment fied. Energy may be chosen different to that of the treatment fied. Minimises burring due to respiration and interna organ motion Singe exposure ong Image may be of reduced quaity (since short exposure). Fu onine correction is time-consuming Usuay taken throughout the treatment fied deivery. Improved image quaity for static anatomy. Reduced image quaity for fieds invoving moving anatomy (burring) Doube exposure Often used for instances where there is insufficient anatomy within the treatment fied to accuratey assess any set-up error. Coimator jaws are opened to an appropriate size (open fied segment) to see the desired anatomy for a short exposure (usuay 1 5 MU). Coimators are then returned to the treatment fied size/ shape and a second exposure taken. The two images are added digitay to produce a doube exposure Some commercia companies simpy dispay the treatment fied edge overaid onto the open fied segment Singe exposure open fied Deivers extra (concomitant) dose to the patient outside the target voume Identica in purpose to the doube exposure, but makes use of the open fied segment ony For systems which digitay add both segments together in a doube exposure, the contrast is often improved for the open fied segment Movie oops For this and the doube exposure, coimator settings are chosen to maximise usefu anatomica information, but minimise concomitant exposure Mutipe images acquired throughout a or part of the treatment fied Idea for showing anatomica movement for cinica sites (such as ung, breast) where it may significanty affect the coincidence of the target and irradiated voumes On target: ensuring geometric accuracy in radiotherapy 39

39 6.3 Traditiona equipment and techniques Equipment Simuator Comment Reference (kv) panar images in the panned treatment position Images may be radiographic fim, digita fuoroscopy or digita fat pane CT simuator and Virtua Simuation software TPS EPID Fim Computed radiography (CR) Impanted markers In-room utrasound If acquired for pretreatment verification, shoud be compared (registered) with panned treatment source data (usuay CT) Mutisice CT scanner with software for performing virtua simuation Images for geometric verification are usuay DRRs (or more recenty DCRs) For nove technoogies, reference data may be the fu 3D CT dataset with appropriate isocentre reference Imports CT/MR/PET data for treatment panning Images for geometric verification are as for CT simuator Eectronic detector for acquiring images (porta) during treatment deivery, using the MV treatment beam Traditiona radiographic medium for acquiring images (porta) during treatment deivery, using the MV treatment beam. As with diagnostic radiography, fim is paced in an imaging cassette on the beam exit side of the patient 30,78,79 A repacement for radiographic fim for acquiring simuator (kv) or porta (MV) images. A specia pate repaces fim inside a traditiona cassette. However, the pate is insensitive to ambient ight (no darkroom is necessary) and is reusabe. The pate is examined in a specia reader and the image is produced in a digita format straight away 30,78,80 Fiducia markers (usuay sma god seeds or wire cois) impanted in soft tissue within the target voume. Markers are arge enough and dense enough to be seen on porta images taken with fim and EPIDs. Are particuary cear when using asi-type EPIDs Utrasound used in the treatment room for visuaising soft tissue targets (for exampe, the prostate). Position of the transducer is inked (either mechanicay or opticay) to a detection system that reates its position to the isocentre of the treatment machine. Target position is determined with patient on the treatment couch immediatey prior to treatment deivery; isocentre position is often moved at this point using an onine protoco On target: ensuring geometric accuracy in radiotherapy

40 6.4 Nove equipment and techniques Equipment In-room kv imaging In-room CT imaging kv CBCT MV CBCT Cone-beam simuator Tomotherapy Surface and marker tracking Impanted transponders MR inac/mr cobat source Gating Comment Linac (standard): Foor and ceiing mounted diagnostic X-ray tubes and image intensifiers to give orthogona views of the patient on the treatment couch. Static or dynamic (fuoroscopy) images may be used to evauate patient set-up both before and during fied deivery. In dynamic mode, it may be used to infuence the beam on timing of the inac Linac (robotic): Imaging component for a robotic inac. Identica to the standard inac (above), except that it can aso infuence the position and beam on timing of the inac Often described as CT-on rais ; uses a standard CT scanner at one end of the treatment couch. However, instead of the couch moving through the CT gantry, the CT scanner itsef moves ineary over the couch. Fu CT voume data is acquired (precisey the same as panning CT scans) prior to treatment. Patient and couch must rotate by 180 (or 90) degrees between imaging and treatment. Direct geometric reationship between the CT scan co-ordinates and the isocentre of the treatment machine 53,95,96 Diagnostic X-ray tube and associated imaging pane are attached orthogonay to, or diametricay opposite, the treatment machine gantry arm and EPID. Fu-voume cone beam CT scan may be acquired in one singe rotation of the inac gantry prior to treatment. Patient is in the treatment position. Static diagnostic X-rays and dynamic fuoroscopic images may aso be acquired. Direct geometric reationship between the imaging and treatment isocentres 46,54,56,97 99 No extra arms need be attached to the treatment machine gantry for MV CBCT. An MV X-ray beam and EPID are used to acquire a fu voume cone beam CT scan in one revoution of the inac gantry prior to treatment. Static and dynamic images are possibe, identica in operation to the standard EPID. Imaging isocentre IS the treatment isocentre 55 Uses a fat pane imaging system in pace of the image intensifier on the simuator. Thus can be identica to the treatment machine (ie, with no need for movement of the X-ray tube head). Can use cone beam technoogy and software for acquiring fu voume cone beam CT data as the reference for treatment verification A modaity of X-ray radiation therapy which combines the use of a computer-controed inear acceerator, mutieaf coimator (MLC) and CT detector subsystem, a mounted on a rotating gantry. Intensity-moduated therapy is deivered in a heica fashion as the patient is moved on the couch through the rotating gantry 100,101 Uses the principe of photogrammetry to measure opticay the position of markers paced on the patient s skin, or the entire skin surface itsef, reative to a reference point (such as the machine isocentre in the treatment room). Light of optica or infrared waveengths may be used. As a set-up or geometric verification device, it can register pretreatment and treatment images to determine associated patient set-up errors. This may be performed continuousy throughout the patient set-up and treatment deivery process, and has no associated concomitant radiation exposures 102,103 Beacon (wireess) transponders which may be impanted in the tissue voume whose position is to be verified, in the same way as god fiducia markers for prostate ocaisation. The ocation of the markers reative to the treatment isocentre is determined post-panning, and verified in the treatment room using a magnetic source and receiver coi. The detector array itsef is tracked in rea-time (with respect to the machine isocentre) using an infrared optica tracking system 104 A concept to utiise the very high-quaity soft tissue imaging capabiity of MR imaging for image guidance during treatment deivery. Designs use a combination of diagnostic MR imaging equipment with an added inac mounted on a ring structure (simiar to tomotherapy) in the midtransverse pane of the MR magnet. An aternative is, instead of the inac, to mount three cobat sources (with MLC) on the rotating structure 16,105 A method of addressing respiratory motion during the deivery of radiotherapy. Uses optica (visibe ight or infrared) or mechanica (pressure transducers) methods to correate movement of the chest or diaphragm (externa surrogates) with interna movement of the target voume, during initia CT scanning. During treatment deivery, the movement of the externa surrogates is used to indicate when the target voume is centred on the treatment isocentre and is correated with the beam-on timing of the treatment machine 106 On target: ensuring geometric accuracy in radiotherapy 41

41 7 Site-specific protocos for geometric verification 7.1 Brain Suggested protoco for brain verification Fraction 1 (images acquired & actioned before treatment deivery) Fractions 2 & 3 Action before Fraction 4 Acquire orthogona image set, minimising dose to critica structures (where possibe) If fied edge verification is needed, where possibe acquire images of a treatment fieds Assess for and correct gross errors immediatey Acquire orthogona image set Assess each image against toerance eves set and correct gross errors for each fraction where necessary Cacuate the overa systematic error (average of the isocentric set-up error) in each orthogona direction If set-up error is greater than the action eve, appy the systematic set-up error correction (NAL recommended) Fractions 4 & 5 If the set-up has been corrected, confirm by repeat imaging (typicay two or more fractions) Weeky & first day of each phase of treatment pan If practica, cacuate the new overa systematic set-up error and correct vaues greater than action eves Acquire orthogona image set each week Assess each image and correct gross errors for each fraction where necessary If set-up error is greater than the toerance vaue, check by repeat imaging (typicay two or more fractions) Appy any systematic set-up error correction Daiy verification may be required for treating tumours panned with very sma margins such as those treated stereotacticay Intrafractiona verification is unnecessary Specia consideration shoud be given for reducing concomitant exposure when treating benign tumours such as pituitary adenomas Toerances and action eves to use wi vary, particuary with the immobiisation used and compiance of the patient and shoud be chosen accordingy Anatomica match structures It is recommended that at east three structures visibe within the fied be outined. The anatomies identified beow indicate the most stabe features. It is recommended that these structures shoud be used for comparison, wherever they fa within the fied arrangement chosen. Stabe radioopaque structures/ surgica defects Orbita ridges Nasa septum Inner border of sku vaut Fronta sinuses Zygoma Occiput Pituitary fossa Fronta sinuses Orbita ridges 42 On target: ensuring geometric accuracy in radiotherapy

42 7.1.2 Evidence for brain verification guideines A particuar issue with intracrania radiotherapy is the proximity of structures that are particuary sensitive to radiation and where damage with functiona oss can have major consequences. Some treatments require radiotherapy to be panned with steep dose gradients to avoid these critica structures. Effective immobiisation and accurate radiation deivery methods are therefore crucia to provide the higher degree of set-up accuracy required. Immobiisation and patient positioning Achievabe accuracy of treatment deivery varies with: Immobiisation used the materia type, fixation method, area of materia in contact with the patient and supporting technique a affect the achievabe reproducibiity Compiance of the patient accuracy may be compromised by the inabiity of the patient to remain sti for the treatment duration. This can be due to probems such as neuroogica deficit, where the patient is physicay unabe to keep sti, nausea from raised intracrania pressure or anxiety. Immobiisation for brain treatments may often require whoe face shes or masks to be used; this can be probematic in patients with caustrophobia. These patients shoud be identified pretreatment and the probem resoved or the appropriate margins panned, and an individua toerance set rather than conforming to the standard for the technique The uncertainty resuting from the mutipe imaging modaities used in panning with image registration. This arises from each imaging stage carrying its own eve of uncertainty, as we as the many opportunities for transfer errors to occur. Set-up reproducibiity Reproducibiity wi vary with immobiisation used. Studies using stereotactic frames have reported set-up errors in the region of and 2 mm. 108,105 The set-up error using immobiisation masks has been measured as ranging from approximatey 3 mm using high meting point thermopastic (acryic) systems 110 to mm using ow meting point thermopastic systems 111 and 3.27 mm using thermopastics in combination with bite bock. 112 Interna organ motion The brain can move very itte inside the cranium and the contribution to set-up accuracy from interna organ motion is very sma in this group of patients. Thus intrafractiona anaysis is not required. Imaging and radiotherapy fieds to image The timing and frequency of imaging for verification of radiotherapy to the brain is currenty undefined. The structure of the head is such that effective immobiisation may resut in ess patient positiona variation than in other anatomica sites and the anatomy of the brain is not subject to arge interna motions. If immobiisation has been previousy evauated and the department is confident with the process, first-day ony images may be sufficient to identify gross and systematic data preparation errors. However, there are other factors that may introduce arge systematic and random uncertainties and daiy images for a minimum of the first three days woud be required to identify these. 113 Imaging for a minimum of the first three days is therefore recommended. Additiona uncertainties may occur if the fit of the immobiisation device changes over time; for exampe, with steroid use. Weeky imaging is therefore recommended. 39 The proximity of critica structures in the brain means that for a radica treatments, images shoud be taken of a treatment fieds wherever possibe, so that the fied edges can be reviewed as we as the isocentre. Where insufficient anatomy is seen in the image, doube exposures shoud be used, taking care to use asymmetrica fieds to avoid exposing critica structures. It may not be possibe to image some panned beams such as vertex fieds. If the margins are critica, visua verification of the inac ight fied compared to anatomy may be made. The structure of stereotactic immobiisation devices and the non-copanar arrangement of treatment fieds may resut in the inabiity to gain cear images, therefore measurements of fiducia surrogates can give good information. 108 Specia consideration shoud be given for reducing concomitant exposure when treating benign tumours such as pituitary adenomas. For other tumours, where doube exposures are required, consideration shoud be given to where the anatomica information can be gained without imaging through critica structures. On target: ensuring geometric accuracy in radiotherapy 43

43 7.2 Head and neck Suggested protoco for head and neck verification Fraction 1 (images acquired & actioned before treatment deivery) Acquire orthogona image set, minimising dose to critica structures (where possibe) If fied edge verification is needed, where possibe acquire images of a treatment fieds Assess for and correct gross errors immediatey Fractions 2 & 3 Acquire orthogona image set Action before Fraction 4 Assess each image against toerance eves set and correct gross errors for each fraction where necessary Cacuate the overa systematic error (average of the isocentric set-up error) in each orthogona direction If set-up error is greater than the action eve, appy the systematic set-up error correction (NAL recommended) Fractions 4 & 5 If the set-up has been corrected, confirm by repeat imaging (typicay two or more fractions) Weeky & first day of each phase of treatment pan If practica, cacuate the new overa systematic set-up error and correct vaues greater than action eves Acquire orthogona image set each week Assess each image and correct gross errors for each fraction where necessary If set-up error is greater than the toerance vaue, check by repeat imaging (typicay two or more fractions) Appy any systematic set-up error correction Daiy verification may be required for treating tumours panned with very sma margins or hypofractionated techniques Immobiisation is mandatory Effect of organ movement of arynx and tongue shoud be considered Intrafractiona verification may be necessary Obique images are difficut to interpret and it is recommended that the isocentre is verified using anterior/posterior and atera views The images acquired must be of sufficient size to ensure bony anatomy is visibe Toerances and action eves to use wi vary, particuary with the immobiisation and treatment technique used as we as compiance of the patient and shoud be chosen accordingy Anatomica match structures It is recommended that at east three structures visibe within the fied be outined. The anatomies identified beow indicate the most stabe features. It is recommended that these structures shoud be used for comparison, wherever they fa within the fied arrangement chosen. Nasa septum Stabe radiopaque structures Vertebra bodies Sinuses Maxia Cavices Posterior wa of trachea Pituitary fossa Base of sku Vertebra bodies (anterior borders) 44 On target: ensuring geometric accuracy in radiotherapy

44 7.2.2 Evidence for head and neck verification guideines Immobiisation and patient positioning Immobiisation devices are routiney used in head and neck cancer and shoud now be considered mandatory for other tumour types invoving this region, such as ymphoma. There are three main types of device: Low meting point thermopastic masks High meting point thermopastic masks, such as Perspex or acryic Bite bock using denta casts. The bite bock immobiisation offers a sma improvement in positiona accuracy but is not suitabe for a patients (since many are edentuous) and it has to be used with skin marks or other means of identifying the isocentre. Change in shape of anatomy is potentiay important in head and neck cancer and may ead to i-fitting immobiisation masks and subsequent patient movement. This is due to both tumour shrinkage and weight oss. Where there are arge noda masses at the start of treatment, gross tumour voume (GTV) has been shown to decrease by a median of 1.8% per day. 114 Seected patients with buky disease shoud be considered for re-masking and re-panning in the third or fourth week of treatment. Set-up reproducibiity Overa, the iterature shows good concordance with a of the immobiisation methods stated above. Reported set-up errors vary between 1.6 and 4.6 mm. 36 Materia type, fixation method and headrest used aso affect the achievabe reproducibiity; the more rigid the materia, the greater the number of fixation points and the more the shape of the headrest matches the anatomy of the patient, the more effective the immobiisation. Thermopastic masks shoud not be used immediatey after being made since there may be shrinkage of up to 2 mm in the first 24 hours. Set-up error is most marked in the shouder region. This can be minimised by ensuring there are additiona fixation points on each shouder as we as the standard points over the head. 115 This is particuary important where the fied covers the ower neck and supracavicuar fossae. Interna organ motion The tongue and arynx may show some movement on swaowing which is of particuar reevance when seecting sites suitabe for anatomica matching. Imaging and radiotherapy fieds to image Imaging shoud be on Days 1 3 of treatment, and simiary at the beginning of each new phase of treatment. It shoud be weeky thereafter using site-specific toerances determined ocay but which is ikey to be in the order of 2 3 mm Mutipe fieds are routiney used in head and neck cancer. Obique images are difficut to interpret and it is difficut to determine couch corrections from this, therefore it is recommended that the isocentre is verified using anterior/ posterior and atera views with a NAL protoco. The images must be of sufficient size to ensure bony anatomy is visibe. This is of particuar importance where sma fieds are used such as in gottic cancer or boost voumes. Dose-sensitive structures such as the eyes are to be avoided when imaging. On target: ensuring geometric accuracy in radiotherapy 45

45 7.3 Thorax and mediastinum Suggested protoco for thoracic and mediastinum verification Fraction 1 (images acquired & actioned before treatment deivery) Acquire orthogona image set, minimising dose to critica structures (where possibe) If fied edge verification is needed, where possibe image a treatment fieds where fied shaping is used (manua shieding or MLCs) Assess for and correct gross errors immediatey Fractions 2 & 3 Acquire orthogona image set Action before Fraction 4 Assess each image against toerance eves set and correct gross errors for each fraction where necessary If avaiabe, acquire a movie oop to assess intrafractiona movement Cacuate the overa systematic error (average of the isocentric set-up error) in each orthogona direction If set-up error is greater than the action eve, appy the systematic set-up error correction (NAL recommended) Fractions 4 & 5 If the set-up has been corrected, confirm by repeat imaging (typicay two or more fractions) Weeky & first day of each phase of treatment pan If practica, cacuate the new overa systematic set-up error and correct vaues greater than action eves Acquire orthogona image set each week Assess each image and correct gross errors for each fraction where necessary If set-up error is greater than the toerance vaue, check by repeat imaging (typicay two or more fractions) Appy any systematic set-up error correction Daiy verification may be required for treating tumours panned with very sma margins or hypofractionated techniques Patient immobiisation devices to hep maintain treatment position is essentia Compensation for organ movement due to respiration and cardiac contractions shoud be considered Obique images are difficut to interpret and it is recommended that the isocentre is verified using anterior/posterior and atera views Toerances and action eves to use wi vary, particuary with the immobiisation and treatment technique used as we as compiance of the patient and shoud be chosen accordingy Anatomica match structures It is recommended that at east three structures visibe within the fied be outined. The anatomies identified beow indicate the most stabe features. It is recommended that these structures shoud be used for comparison, wherever they fa within the fied arrangement chosen. Media edge of rib Cavices Carina and trachea Anterior chest wa Sternum Latera extent of chest wa Latera edge of vertebra bodies Anterior edge of vertebra bodies 46 On target: ensuring geometric accuracy in radiotherapy

46 7.3.2 Evidence for thoracic and mediastinum verification guideines Immobiisation and patient positioning Most radica treatments wi have the patient supine with their arms raised, uness a superior tumour is being treated. Immobiisation with a T bar or simiar is necessary. Patient set-up errors can be more important than interna organ motion. 113 Simpe devices to improve patient comfort for this position therefore may have a significant effect on reproducibiity. Compex extra crania stereotactic devices are currenty being evauated for this site. Set-up reproducibiity For radica ung treatments, set-up errors have been reported between 1 to 5 mm. 36 For ymphoma treatments, greater systematic errors are reported with arge mante type fieds ranging from 3 to 10 mm. Interna organ motion The thorax and mediastinum presents a particuar probem because of interna organ movement during respiration and the cardiac cyce. Greatest movement is seen with ower obe tumours which have no attachment to adjacent structures when average cranio-cauda movement has been measured at 12 mm. 117 Other studies have reported a range of movement due to respiration of up to 22 mm. 118 Movement of tumours cose to the heart due to cardiac function may be 1 4 mm. These movements due to respiration, or the cardiac cyce, must be addressed for radica radiotherapy by one of the foowing approaches. Screening or preferaby video imaging or sow CT to measure the extent of movement and an appropriate increase in the PTV expansion made. This is the east satisfactory since athough it wi ensure coverage of the PTV it wi increase the voume of norma tissue irradiated, but is the simpest and most readiy adopted in departments which do not have the additiona equipment or expertise for more compex approaches. Typica expansions wi be of 15 mm in the cranio cauda direction and 10 mm ateray. 119 Breathing contro during radiation exposure. This may use either vountary breath 120 or active breathing contro (ABC) in which temporary breath hoding at a chosen point in the respiratory cyce is used. 121 However, the breath-hod eve and patient training can affect reproducibiity and the interfraction and intrafraction reproducibiity of the breath hods shoud be estabished. 122,123 Gated radiation exposure using one of the systems which puse inear acceerator output with a specific phase in the respiratory cyce, usuay inspiration determined by impanted markers or externa markers as a surrogate for ung position. 124 Impanted markers are an extremey invasive procedure. However, for externa markers to be effective, the reationship between the externa marker and tumour position must be constant. Erratic breathing patterns affect the reproducibiity, 125,126 athough audio breathing instructions have been found to make breathing more reguar. 127,128 Imaging and radiotherapy fieds to image It may be possibe to visuaise some ung tumours on EPI, but for the majority of patients it wi ony be possibe to verify the patient position using bony anatomy. The frequency of the imaging wi depend on the treatment technique used. Techniques to compensate for breathing motion may require additiona imaging to verify the reproducibiity of the technique. Radica treatments wi require three or four fied pans; obique views of the thorax are difficut to interpret; the recommendation therefore is for antero-posterior and atera views to ocate the isocentre with adoption of the no action eve protoco as described earier. Lymphoma treatments are usuay parae opposed fieds but aso often incorporate compex shieding. Verification fieds shoud incude a shieded areas together with reference anatomy to enabe effective vaidation of the shieding accuracy as we as the fied set-up. To ensure a the information is acquired, two EPIs or port fims may be required. On target: ensuring geometric accuracy in radiotherapy 47

47 7.4 Breast Suggested protoco for breast verification Fraction 1 (images acquired and actioned before treatment deivery) For standard whoe-breast radiotherapy, acquire images of either or both treatment fieds and a noda fieds, avoiding dose critica structures (where possibe) Doube exposures not necessary Set-up error shoud be determined as a measure of centra ung distance and skin coverage rather than isocentre dispacement (critica aspects of whoe-breast RT) For partia fied irradiation or dynamic IMRT deiveries, use other pretreatment checks to verify the MLC pattern and use open fied images to verify anatomy Opposed fieds cannot resove set-up errors into a three orthogona directions. Based on the panar image, correction is made by either: Estimating changes to isocentre and re-imaging to check correction Determining isocentre shift using simuation equipment Fractions 2 & 3 Further images ony appicabe where isocentre dispacements are cacuabe and actionabe A more accurate evauation of the systematic component of the set-up error in each panar image wi be obtained by repeat imaging over a number of fractions Cacuation of corrective couch shifts is difficut with tangentia images. Accurate correction of a systematic set-up error wi generay require re-simuation Any systematic set-up correction shoud be appied before Fraction 4 and imaging repeated (typicay 2 or more fractions). Any corrections appied to the treatment set-up must be verified. Weeky Acquire images of tangentia fieds each week to assess shape change or trends Assess each image and correct gross errors for each fraction where necessary if set-up error is greater than the toerance vaue, check by repeat imaging (typicay two or more fractions) A more accurate evauation of the intrafractiona variabiity of set-up wi be obtained by using mutipe intrafractiona imaging such as cine acquisition or movie oops. Intrafractiona verification to be considered where possibe to inform panning margins Daiy verification may be required for tumours panned for IMRT or partia organ irradiation Patient immobiisation devices to hep maintain treatment position are essentia Contro of systematic and random set-up errors is more effectivey achieved on a treatment popuation basis by improved immobiisation, adherence to protocos and training Toerances and action eves to use wi vary, particuary with the immobiisation and treatment technique used as we as compiance of the patient and shoud be chosen accordingy Anatomica match structures It is recommended that at east three structures visibe within the fied be outined. The anatomies identified beow indicate the most stabe features. It is recommended that these structures shoud be used for comparison, wherever they fa within the fied arrangement chosen. Breast outine Lung area/ant. border Centra fash distance Centra ung distance Inferior centra margin Shieding Cervica spine Lung area 48 On target: ensuring geometric accuracy in radiotherapy

48 7.4.2 Evidence for breast verification guideines Immobiisation and patient positioning Verification is an important component of breast radiotherapy. Uncorrected gross and arge systematic set-up errors may resut in reduced treatment efficacy and increased compications. 129,130 Daiy variations arise from two main sources, the reproducibiity of the patient positioning and the interna movement of the breast tissue and nodes caused by respiratory and cardiac motion. Patient positioning can be improved by using effective immobiisation; breast position and shape are atered by the arm position and anguation of the immobiisation device. Footrests, knee and bottom supports may minimise patient sippage. Set-up reproducibiity The immobiisation used wi affect the set-up reproducibiity. Set-up errors of 2.1 mm and 6.5 mm have been reported, varying between immobiisations such as arm poes and tited boards and apha crade. 66,131 From a comprehensive survey of set-up errors reported in the iterature, 70 median vaues for the standard immobiisation method of anged board, arm support and various combinations of foot, knee and buttock supports were derived and are reproduced beow. Anatomica parameter Σ set-up (mm) σ set-up (mm) Centra ung distance (CLD) Centra fash distance (CFD) Inferior centra margin (ICM) Mean Therefore 3 mm may be used as suitabe starting vaue for both Σ set-up and σ set-up and action eves shoud be set appropriatey. A shrinking action eve or fixed action eve is therefore most appropriate. 40 For exampe, a fixed action eve of twice the random set-up error, σ (typicay around 6 mm) coud be empoyed. Where possibe, oca vaues of σ shoud be ascertained from a porta imaging study. Breast sweing during radiotherapy may affect the reative accuracy of the measurements used for assessment. Interna organ motion Breathing and cardiac motion are chaenging and verification protocos shoud expore the use of 4D detection and correction strategies. 69 The contribution to set-up accuracy from respiration can be measured by taking mutipe images through the treatment fraction. 132 Imaging and radiotherapy fieds to image Where the panned treatment fied covers the whoe breast, sufficient anatomica information shoud be avaiabe with which to make a decision and doube exposures wi not be necessary. For non-imrt techniques, centra ung distance is the most reiabe andmark for anatomy matching. 131 Both tangentia fieds shoud be imaged where there is a need to verify non-rectanguar fied shaping, or where the variation in ung voume treated throughout the entire fraction needs to be assessed. 132 Correction of systematic errors is difficut for breast radiotherapy as orthogona images are not usuay acquired. Tangentia opposed fieds aone cannot resove set-up errors into a three orthogona directions. Accurate correction of a systematic set-up error wi generay require re-simuation. Contro of systematic and random set-up error is more effectivey achieved on a treatment popuation basis by improved immobiisation, adherence to protocos and training. It has been demonstrated that breast voume changes occur between the 5 8 fractions, 133 which may have dosimetric consequences. 134 Weeky imaging is recommended for detecting surface outine changes and systematic trends. 39 For standard whoe-breast radiotherapy, there is ony sma benefit in using daiy imaging protocos over weeky after the initia assessment of the first few fractions. 132,135 A nationa study aims to assess the imaging protocos needed for partia breast irradiation and IMRT (IMPORT). 136 For partia fied irradiation or dynamic IMRT deiveries, other pretreatment checks can be used to verify the MLC position or pattern and an open fied image used to verify anatomy position. 137 On target: ensuring geometric accuracy in radiotherapy 49

49 7.5 Pevis (prostate, badder, gynaecoogica and coorecta cancers) Suggested protoco for pevis verification Fraction 1 (images acquired & actioned before treatment deivery) Image a treatment fieds where possibe Fractions 2 & 3 Image orthogona set Action before Fraction 4 Use open fieds or doube exposures, where necessary, to ensure sufficient stabe anatomy can be seen in the images Avoid exposing dose critica structures by reducing imaging borders where possibe Assess for and correct gross errors immediatey Assess each image against toerance eves set and correct gross errors for each fraction Cacuate the overa average of the isocentric set-up error in each orthogona direction If set-up error is greater than the action eve vaue, appy correction to isocentre Fractions 4 & 5 If the set-up has been corrected, confirm by repeat imaging (typicay two or more fractions), cacuate the new overa average set-up error and correct vaues greater than action eves Weeky & first day of each phase of treatment pan Image orthogona set each week If set-up error is greater than the toerance vaue, check by repeat imaging (typicay two or more fractions) Assess and correct set-up based on new average set-up errors Daiy onine verification may be required for treating tumours panned with very sma margins or those ikey to have arge interna movements Intrafractiona verification may be necessary when using imaging to cacuate the impact of interna movement for popuation panning margins Anatomica match structures It is recommended that at east three structures visibe within the fied be outined. The anatomies identified beow indicate the most stabe features. It is recommended that these structures shoud be used for comparison, wherever they fa within the fied arrangement chosen. Iiac crest Superior pubic ramus Vertebra bodies Sacrum Obturator foramen Acetabuum Pubic symphysis 50 On target: ensuring geometric accuracy in radiotherapy

50 7.5.2 Evidence for pevis verification guideines Using bony anatomy with EPI for verification of the patient s position provides no information regarding the soft tissue organs; the effect of interna organ movement on the position of the target is therefore not gained. Where verification using bony anatomy aone is used, the ack of information shoud be taken into consideration in panning treatment margins. Methods exist to overcome this, some can image soft tissue and some use a radio-opaque surrogate that is impanted in the soft tissue target and moves with it. 1. Impanted markers It is possibe to use impanted god markers 81,83,138 to enabe soft tissue imaging using EPI when treating prostate patients and is under investigation for other sites. 82 This is an invasive procedure and the possibiity of marker migration has to be eiminated before cinica use. 2. Utrasound The accuracy of using utrasound techniques is currenty under debate. One study shows the B-mode acquisition and targeting (BAT) system is equivaent to within 3 mm when compared with daiy CT scans for prostate ocaisation. 85 However, other studies have shown arger systematic errors when compared to CT or impanted marker verification methods. 88,139 Additionay, there may be significant inter-user variation of the contour aignment process either as a resut of misidentification of the target structures on the utrasound image, or due to prostate movement at the time of BAT acquisition. 140, Cone beam CT X-ray voumetric imaging aows a 3D image to be taken prior to treatment with the patient remaining in the treatment position. 54,99,142 Initia studies have focused on prostate and badder cancer patients but this technoogy has potentia to be usefu for a the sites mentioned above. 4. Tomotherapy Heica tomotherapy is another method of achieving voume imaging in pevic patients. This is under investigation at present and there are a few pubications describing eary experience. 143,144 Genera pevic toerances The toerances used depend on the site treated, the motion observed and the margin used, as discussed previousy. Hurkmans et a 36 have summarised motion for these sites. This information can be used as a guideine to hep set toerances but it is essentia that the toerance is investigated in each individua department. Toerances for pevic treatments tend to be between 3 mm and 5 mm depending on site and size of treatment fied. The mean dispacement using bony anatomy in the pevis and no correction can be expected to be in the range of 2 5 mm. This can be reduced to <2.5 mm by the use of correction protocos in a directions. To achieve the same reduction in prone patients requires more corrections and/or onine imaging. On target: ensuring geometric accuracy in radiotherapy 51

51 7.5.3 Evidence for prostate verification guideines Immobiisation and patient positioning Patients treated supine have been associated with ess interna prostate movement and patient movement than patients treated prone. 89,145 However the prone position has been reated to a reduction of the dose to the rectum and bowe. 19,146 Leg movement and pevic tensing can affect the position of the prostate when in the supine position. 147,148 Foot and/or anke supports can be used which maintain the ange between the ankes and so a stabe position. Aso avoiding over-fu badders and ensuring the patient is in a reaxed state. Setting the isocentre height from the couch top is a more accurate method of setting the isocentre than using skin marks since it is not infuenced by changes in patient separation. 146 Set-up reproducibiity The toerance used for EPI depends on the site treated, the motion observed and the margin used, as discussed above. Hurkmans et a 36 have summarised motion for these sites. This information can be used as a guideine to hep set toerances but it is essentia that the toerance is investigated in each individua department. Toerances for pevic treatments tend to be between 3 mm and 5 mm depending on site and size of treatment fied. The mean dispacement using bony anatomy in the pevis and no correction can be expected to be in the range of 2 5 mm. This can be reduced to <2.5 mm by the use of correction protocos in a directions. To achieve the same reduction in prone patients requires more corrections and/or onine imaging. Interna organ motion Changes in recta size and shape are the main factors which affect prostate movement, independent of bony anatomy, 147, which can be mm in the anterior posterior direction and up to 13.1 mm (mean 0.8 mm; SD 2.3 mm) superiory. 147, Methods to reduce the variation in recta distension incude the use of enemas or recta baoons. 155,156 The use of a recta baoon has the added advantage of significanty decreasing the dose to the recta wa by distending the rectum so that more of it is outside the high-dose radiation target voume. 157 The use of enemas in reducing recta distension is being investigated as are changes in diet, but at the time of writing, evidence of their efficacy does not exist. It is a matter of debate whether badder fiing has an impact on prostate movement, 150,151 athough it wi affect the dose voume histograms. Imaging The frequency of imaging wi vary according to the size of the voume being treated and the interna movement of the prostate. If the prostate itsef can be imaged and the voume to be treated or margins panned are reativey sma, daiy onine verification may be needed. Radiotherapy fieds to image Sufficient stabe anatomy needs to be visibe in the verification images to ensure accurate matching can be made. For sma treatment beams or beams covering soft tissue targets ony, open fieds or doube exposures are necessary. The verification images shoud be taken avoiding exposing dose critica structures where possibe, by reducing the imaging borders. The pevic rim is routiney used on the anterior image to verify the position in the superior inferior direction. However, this must be used with the knowedge that pevic rotation may occur around the atera axis and so affect the readings. In atera fieds the femora, athough visibe, shoud not be used as the position varies too easiy. The acetabuum and symphysis or sacrum are more stabe and better visuaised. 52 On target: ensuring geometric accuracy in radiotherapy

52 7.5.4 Evidence for badder verification guideines Immobiisation and patient positioning Badder patients tend to be treated supine which is stabe in terms of patient position but interna organ movement must be considered. Setting the isocentre height from the couch top is a more accurate method of setting the isocentre than using skin marks. 149 Set-up reproducibiity The toerance used for EPI depends on site treated, the motion observed and the margin used, as discussed above. Hurkmans et a 36 have summarised motion for these sites. This information can be used as a guideine to hep set toerances but it is essentia that the toerance is investigated in each individua department. Toerances for pevic treatments tend to be between 3 mm and 5 mm depending on site and size of treatment fied. The mean dispacement using bony anatomy in the pevis and no correction can be expected to be in the range of 2 5 mm. This can be reduced to <2.5 mm by the use of correction protocos in a directions. To achieve the same reduction in prone patients requires more corrections and/or onine imaging. Interna organ motion Conventionay patients treated for badder cancer are treated with an empty badder. Despite advising patients to empty their badder immediatey prior to treatment, the badder voumes during a treatment schedue can vary. The most evident motion is in the superior inferior direction and the anterior posterior direction. 142,158 This trend is refected when treating patients with a fu badder. 156 Treating patients for badder cancer can require margins of up to 23 mm posteriory and superiory. Imaging The frequency of imaging wi vary according to the size of the voume being treated and the interna movement of the badder. If the badder itsef can be imaged and the voume to be treated or margins panned are reativey sma, daiy onine verification may be needed. Radiotherapy fieds to image Sufficient stabe anatomy needs to be visibe in the verification images to ensure accurate matching can be made. For sma treatment beams or beams covering soft tissue targets ony, open fieds or doube exposures are necessary. The verification images shoud be taken avoiding exposing dose critica structures where possibe, by reducing the imaging borders. The pevic rim is routiney used on the anterior image to verify the position in the superior inferior direction. However, this must be used with the knowedge that pevic rotation may occur around the atera axis and so affect the readings. In atera fieds the femora, athough visibe, shoud not be used as the position too easiy varies. The acetabuum and symphysis or sacrum are more stabe and better visuaised. On target: ensuring geometric accuracy in radiotherapy 53

53 7.5.5 Evidence for coorecta verification guideines Immobiisation and patient positioning Recta voumes are treated prone which as mentioned previousy, can be unstabe depending on the immobiisation used. There is potentia for the pevic musce tension to affect ana verge dispacements and affect the anterior posterior and superior inferior position of the fied. 160 A coorecta immobiisation device in which the patient ies prone, caed a bey board, is often used to reduce the amount of sma bowe in the PTV. Care must be taken to standardise the position of the patient within a bey board. 161,162 Using no immobiisation in seven patients, interfraction movement was most evident in the anterior posterior direction (23% movements were >10 mm) compared to the right eft and superior inferior (3% and 16% respectivey). The frequency of interfraction dispacements >5 mm was more simiar for each direction. 102 Set-up reproducibiity The toerance used for EPI depends on site treated, the motion observed and the margin used, as discussed above. Hurkmans et a 36 have summarised motion for these sites. This information can be used as a guideine to hep set toerances but it is essentia that the toerance is investigated in each individua department. Toerances for pevic treatments tend to be between 3 mm and 5 mm depending on site and size of treatment fied. The mean dispacement using bony anatomy in the pevis and no correction can be expected to be in the range of 2 5 mm. This can be reduced to <2.5 mm by the use of correction protocos in a directions. To achieve the same reduction in prone patients requires more corrections and/or onine imaging. Interna organ motion There is aso a possibiity of interna organ motion occurring in these patients. Athough there are no data avaiabe for coorecta patients, there is extensive data on recta movement in prostate patients mentioned previousy. Imaging The frequency of imaging wi vary according to the size of the voume being treated and the interna movement of the target. If the target itsef can be imaged and the voume to be treated or margins panned are reativey sma, daiy onine verification may be needed. Radiotherapy fieds to image Sufficient stabe anatomy needs to be visibe in the verification images to ensure accurate matching can be made. For sma treatment beams or beams covering soft tissue targets ony, open fieds or doube exposures are necessary. The verification images shoud be taken avoiding exposing dose critica structures where possibe, by reducing the imaging borders. The pevic rim is routiney used on the anterior image to verify the position in the superior inferior direction. However, this must be used with the knowedge that pevic rotation may occur around the atera axis and so affect the readings. In atera fieds the femora, athough visibe, shoud not be used as the position too easiy varies. The acetabuum and symphysis or sacrum are more stabe and better visuaised. 54 On target: ensuring geometric accuracy in radiotherapy

54 7.5.6 Evidence for gynaecoogica verification guideines Immobiisation and patient positioning The data avaiabe on the treatment set-up and movement of gynaecoogica patients are much more imited than that avaiabe for prostate and badder cancer patients. Bey boards with the patient prone have aso been used to reduce the amount of sma bowe in the treatment fied. 163 Set-up reproducibiity The toerance used for EPI depends on site treated, the motion observed and the margin used, as discussed above. Hurkmans et a 36 have summarised motion for these sites. This information can be used as a guideine to hep set toerances but it is essentia that the toerance is investigated in each individua department. Toerances for pevic treatments tend to be between 3 mm and 5 mm depending on site and size of treatment fied. The mean dispacement using bony anatomy in the pevis and no correction can be expected to be in the range of 2 5 mm. This can be reduced to <2.5 mm by the use of correction protocos in a directions. To achieve the same reduction in prone patients requires more corrections and/or onine imaging. Interna organ motion Interna motion is aso an important factor in gynaecoogica patients though ess we documented. Movement of the cervix and uterus has significant impact on margins. 163 The uterus has been found to move with respect to badder fiing and the argest effect was in the SI direction. 164 Imaging The median movement for the corpus uteri was 7 mm and (95% CI 3 15 mm) and the cervix 4 mm (1 6 mm). The frequency of imaging wi vary according to the size of the voume being treated and the interna movement of the target. If the target itsef can be imaged and the voume to be treated or margins panned are reativey sma, daiy onine verification may be needed. Radiotherapy fieds to image Sufficient stabe anatomy needs to be visibe in the verification images to ensure accurate matching can be made. For sma treatment beams or beams covering soft tissue targets ony, open fieds or doube exposures are necessary. The verification images shoud be taken avoiding exposing dose critica structures where possibe, by reducing the imaging borders. The pevic rim is routiney used on the anterior image to verify the position in the superior inferior direction. However, this must be used with the knowedge that pevic rotation may occur around the atera axis and so affect the readings. This has been shown to be true in cervix patients. The position of the umbar vertebrae may be more accurate for the ongitudina direction in arge pevic fieds. On target: ensuring geometric accuracy in radiotherapy 55

55 7.6 Spine Suggested protoco for spina verification Fractions 1 3 (images acquired & actioned before treatment deivery) Action before Fraction 4 Image a treatment fieds where possibe, avoid exposing dose-critica structures Assess each image against toerance eves set and correct gross errors for each fraction Cacuate the overa systematic error (average of the isocentric set-up error) in each orthogona direction If set-up error is greater than the action eve, appy the systematic set-up error correction (NAL recommended) Fractions 4 & 5 If the set-up has been corrected, confirm by repeat imaging (typicay two or more fractions) Weeky & first day of each phase of treatment pan If practica, cacuate the new overa systematic set-up error and correct vaues greater than action eves Image a treatment fieds weeky Assess each image and correct gross errors for each fraction where necessary If set-up error is greater than the toerance vaue, check by repeat imaging (typicay two or more fractions) Appy any systematic set-up error correction Daiy verification may be required for treating tumours panned with very sma margins or to identify ength changes from the variabiity of daiy spina compression Concomitant exposures shoud be especiay considered in chidren and adoescents Imaging fied width shoud be increased to provide sufficient additiona anatomy to enabe spine identification, where necessary Particuar attention shoud be paid to fied junctions Length of some spina fieds may cause probems with image acquisition and may need to be acquired in sections Toerances and action eves to use wi vary, particuary with the immobiisation used and compiance of the patient and shoud be chosen accordingy Anatomica match structures It is recommended that at east three structures visibe within the fied be outined. The anatomies identified beow indicate the most stabe features. It is recommended that these structures shoud be used for comparison, wherever they fa within the fied arrangement chosen. Posterior edges of vertebra bodies Latera edge of vertebra bodies cervica, thoracic, umbar Anterior edge of vertebra bodies Spinous processes Superior/inferior edges vertebra bodies 56 On target: ensuring geometric accuracy in radiotherapy

56 7.6.2 Evidence for spina verification guideines Immobiisation and patient positioning Achievabe accuracy of treatment deivery varies with: Immobiisation used the patient s body may be immobiised by using restraining accessories such as vacuum bags, apha crades and stereotactic body frames. The patient may be immobiised in a supine or prone position. Some patients may not be immobiised but variations in their position may be minimised by papating vertebra anatomy and correcting accordingy. Compiance of the patient accuracy may be compromised by the inabiity of the patient to remain sti for the treatment duration, due to neuroogica deficit or pain. The reproducibiity of fit can aso be affected if the patient is unabe to fee their position inside the immobiisation. These patients shoud be identified pretreatment and the probem soved or the appropriate margins panned, and an individua toerance set rather than conforming to the standard for the technique. 166 Set-up reproducibiity Very few studies have measured set-up reproducibiity for spina treatments. The use of apha crades has been shown to give Σ set-up of 0.5 mm and σ set-up of 5.39 mm. 165 Set-up reproducibiity with vacuum bags has been measured to be in the region of 4 mm 167 and sterotactic frames a mean vaue of 3.6 mm. 168 Interna organ motion Spina treatments have the same disadvantages as other extracrania sites and are therefore subject to positiona and interna movement variations. The reative position of the individua vertebrae may change due to compression or rotation. Imaging and radiotherapy fieds to image Imaging frequency and distribution are as for the genera guideines; first 3 5 fractions then weeky. 39 The entire spina ength that is treated shoud be imaged, paying particuar attention to fied junctions. Where stepping or feathering of matched fied edges occur, each step shoud be verified. The identification of vertebra eve is often difficut on sma fied spina images. Where it is deemed that eve identification is not possibe (assess using the reference images), imaging fied width shoud be increased to provide sufficient additiona anatomica information. The ength of some spina fieds may cause probems with image acquisition on some simuators and eectronic porta imagers the superior and inferior aspects of the fied may need to be imaged by separate port fims. Many of the patients treated for spina tumours wi be chidren and the impact of a sma increase in overa integra dose from imaging and the associated risks of secondary maignancies are unknown. Long-term foow-up is needed to answer this question. On target: ensuring geometric accuracy in radiotherapy 57

57 7.7 Limb Suggested protoco for imb verification Fraction 1 (images acquired & actioned before treatment deivery) Fractions 2 & 3 (not necessary for a patients) Fraction 4 (not necessary for a patients) Before Fraction 5 (not necessary for a patients) Weeky & first day of each phase of treatment pan Image treatment fieds (usuay anterior or direct atera fieds) Take care to avoid exposing dose-critica structures where possibe (testes) For sma fieds, use open imaging fieds or doube exposures, where necessary, to ensure sufficient stabe anatomy can be seen in the images Assess for and correct gross errors immediatey Image treatment fieds Assess each image against toerance eves set and correct gross errors for each fraction where necessary Cacuate the overa systematic error (average of the isocentric set-up error) in each orthogona direction If set-up error is greater than the action eve, appy the systematic set-up error correction (NAL recommended) If the set-up has been corrected, confirm by repeat imaging (typicay two or more fractions) If practica, cacuate the new overa systematic set-up error and correct vaues greater than action eves Image treatment fieds each week Assess each image and correct gross errors for each fraction where necessary If set-up error is greater than the toerance vaue, check by repeat imaging (typicay two or more fractions) Appy any systematic set-up error correction Daiy verification may be required for treating tumours panned with very sma margins, prosthesis boosts or IMRT Immobiisation and positioning is critica Daiy visua check of radiation ight fied required Particuar attention shoud be paid to fied junctions Length of some imb fieds may cause probems with image acquisition and may need to be acquired in sections Concomitant exposures shoud be especiay considered in chidren and adoescents Anatomica match structures It is recommended that at east three structures visibe within the fied be outined. The anatomies identified beow indicate the most stabe features. It is recommended that these structures shoud be used for comparison, wherever they fa within the fied arrangement chosen. Edges of ong bones Joint spaces Surgica defects or impants Tuberosities and protuberances Media edges of ong bones Edges of ong bones 58 On target: ensuring geometric accuracy in radiotherapy