Imaging by numbers - The story of nuclear medicine physics research in Southampton by Dr John Fleming, November 2013

Transcription

1 Imaging by numbers - The story of nuclear medicine physics research in Southampton by Dr John Fleming, November 2013 Introduction Key milestones in technical development Clinical applications References Staff of the nuclear medicine physics group Introduction The past 50 years has brought about a revolution in medical imaging. Patients going into a hospital in 1970 for an image would most probably be going in for a planar x-ray. In the early 21 st century, we now have many different ways of imaging the body providing complementary information, which is useful in diagnosis and treatment planning. This has been brought about by the application of new physics techniques for imaging, combined with the development of computers. Digital technology not only made these novel techniques practically feasible, but also gave the exciting possibility of processing and analysing the images that were produced. Nuclear medicine imaging, the ability to trace the body s handling of a radioactive substance, was one of the techniques that emerged over this period. The gamma camera was first invented in the 1950s but it was not until the late 1960s that they began to appear in hospitals in the UK. They enabled the exciting possibility of assessing function of the body to complement anatomical detail available from x- rays. The Department of Medical Physics in Southampton wanted to see this new technology applied locally and in the late 1960s set up one of the first autonomous Departments of Nuclear Medicine in the UK. It was led by a specialist nuclear medicine physician supported by a team consisting of a biochemist, medical physicists, technical staff and radiographers. Over the next 40 years, the Department was to have a key role in the development of new techniques and applications of nuclear medicine imaging, which were to have both national and international impact. Figure 1 Left: The first gamma camera built by Hal Anger in 1957 and Right: an early commercial gamma camera. The Nuclear Medicine Physics Group played an important part in these developments. It consisted for most of the period between 1970 and the present day of four medical physicists supported by technical staff and supplemented from time to time by students and postgraduate and postdoctoral research assistants on fixed term placements. Key milestones in technical development

2 The 1970s saw the growth of nuclear medicine imaging to investigate diseases of the lung, thyroid, liver, kidneys and bones. Radiopharmaceuticals were developed to enable the function of these organs to be studied. These were biochemicals that when administered to the body were processed by the organ being studied. They were labelled with a suitable gamma emitting radionuclide, enabling the uptake to be imaged with a gamma camera. This provided information on the function of organs which was complementary to the structural detail provided by conventional x-rays. Figure 2 Example of early lung image obtained from a gamma camera As with other imaging modalities, most interpretation was performed subjectively by skilled nuclear medicine physicians or radiologists, comparing the observed scan with the expected normal appearance based on experience. The availability of digital nuclear medicine images enabled the possibility of quantitative analysis. We realised that this would give the opportunity of calculating quantitative parameters from the images which might lead to more objective and therefore more robust interpretation. However it was clear, even in these early days, that getting reliable numbers from the images was not going to be easy. Although there might indeed be safety in numbers, this would only be useful for clinical interpretation if the numbers were safe. The usefulness in quantification of nuclear medicine images lay in measuring the amount of radioactivity in organs of the body, as it was this that reflected the organ s function. However, the images detected the number of gamma ray counts detected, and this did not bear a simple relationship to the amount of activity. It was confounded by attenuation and scattering of the gamma rays, the limited resolution of the imaging, and noise. We worked on the philosophy that it was important as far as possible to correct for these factors and measure absolute amounts of activity rather than relying on empirical indices based on the number of counts detected. Planar image quantification The early gamma cameras produced two dimensional images of the body, rather analagous to the way that conventional plane x-rays also produce a 2D projection of the underlying 3D structure. Each pixel in a nuclear medicine image detected counts from a rod of tissue through the patient at right angles to the camera face. Activity in this rod did not contribute uniformly to the counts observed in the gamma camera image. Sections of the rod more distant from the camera were severely affected by attenuation with counts being reduced by factors of 5 to 10 compared to activity close to the camera.

3 Figure 3. Illustration of the rod of tissue contributing to each pixel in a planar gamma camera image with the parameters required to calculate the effect of attenuation on the counts [1] The idea of summing two opposing views to average out the effect of attenuation had been applied to whole body counting, and we felt that this principle could also be used in gamma camera imaging. In 1979 we described the first use of using geometric mean imaging of opposed views for absolute quantification of nuclear medicine images [1]. Since that time the technique has been commonly used in numerous applications both in clinical routine and research. Figure 4 This shows an example of the value of geometric mean analysis in obtaining correct quantification in a study assessing renal function from the uptake of Technetium-99m labelled DMSA. In reality the two kidneys have almost equal uptake and therefore almost equal function. Taking a single posterior view alone would seriously underestimate the quantification of activity in the right kidney. By taking the opposed anterior view and calculating the geometric mean counts, an accurate estimate of relative renal function is possible. [11] Time series analysis The ability to capture images on computer meant that it was possible to record a time series of images representing the variation of distribution of radioactivity with time. This data could give further information on the function of organs. Physiological models of the body s handling of the radiopharmaceuticals were developed, which were then used to analyse the digital image data. Parameters of function could be estimated, which were of interest in both clinical routine and research. The Southampton group developed applications for measuring renal [2] and hepatic [3] function from dynamic imaging.

4 Figure 5 The left panel shows time activity curves obtained over the kidneys following intravenous injection of iodine-123 labelled hippuran, illustrating the uptake and passage of the radiopharmaceutical through the kidney. The right panel shows curves obtained for each kidney following deconvolution analysis. This gives the response of the kidney to a simulated bolus in the renal artery which enables the transit of the radiopharmaceutical through the kidney to be accurately determined. [2] Imaging Software With the growing applications of image analysis, the need for creation of an efficient software platform for development became apparent. Commercial manufactures were beginning to provide image processing systems, but for groups wishing to implement new ideas for analysis, the ability to develop software rapidly was required. As so often, necessity proved the mother of invention, and we started work on our own image processing software system, Portable Imaging Computer Software in 1982 (PICS 1991, Southampton University Hospital NHS Trust) [4]. The idea was to create a system for general medical image analysis which was portable between computer systems. It was built in a modular structure so that lower level general routines could be easily put together for specific applications. The later development of other commercial systems providing similar facilities more elegantly meant that it was never used on a wide scale. However it provided the basis of the group s image processing software development for around 20 years or so and was used for the Nuclear Medicine Department s routine analysis and in many research projects. Multimodality Imaging The other vision for PICS was that it would not be specific for nuclear medicine images. We realised the value of the complementary contributions of different imaging modalities and the potential of combining them synergistically. Therefore PICS was designed to allow analysis of different types of image and projects involving ultrasound [5] and CT imaging [6] were undertaken. SPECT-CT Three dimensional imaging using a gamma camera became widely available during the 1980s. This was labelled Single Photon Emission Computed Tomography (SPECT). We felt that there would be advantages in developing our own reconstruction software and incorporated this in PICS. This enabled us to implement novel reconstruction approaches and in 1989, we published the first description of using CT imaging for attenuation correction in SPECT. The same method is now used worldwide on most gamma cameras with CT attachments [7].

5 Figure 6. Left: A modern gamma camera capable of rotating to perform three dimensional SPECT studies together with an associated CT system. Right: The bi-linear graph developed in Southampton allowing estimation of linear attenuation coefficients for gamma rays from CT number [7]. Lower: an image of aerosol deposition on the lung overlaid with the corresponding CT image, allowing the deposition pattern from SPECT to be described relative to its position in the airway tree derived from CT. Computer Simulation As new techniques for quantification of images were developed, it became clear that methods of evaluation were vital. While physical phantom measurements could provide validation of some aspects of the measurement, the experimental conditions did not match the complexity of clinical measurements in humans. By contrast the development of computing technology meant that detailed simulation of experimental conditions could be mimicked quite accurately on computer. We developed a system of simulation of gamma camera imaging [8], which was extremely useful in developing [9] and evaluating [10] quantitative measurement techniques over the subsequent 20 years.

6 Figure 7 Right: model of the segmental structure of the lung used to produce, Left: a realistic computer simulated image of lung perfusion. Images produced in this way were used in a national audit of quantification of lung perfusion pattern from imaging [10] Audit and Guidelines During the 1980s and 1990s, a number of reports appeared in the literature pointing out potential variability in the values obtained from quantitative analysis of nuclear medicine images. In the mid 1990s, the Nuclear Medicine Special Interest Group of the Institute of Medical Physics and Engineering in Medicine (IPEM) created a Nuclear Medicine Software Working Party (NMSWP) to investigate this issue and look for ways to ensure reliability of nuclear medicine image analysis. Prof John Fleming from the Southampton Department took a leading role in establishing this group. A system of audit of quantitative analysis of nuclear medicine investigations was established in the UK and audits of the most commonly applied analyses have been carried out [11]. In one case, that of measurement of glomerular filtration rate (GFR), the variation between centres was so marked that it led to the British Nuclear Medicine Society requesting the working party to produce guidelines for analysis to avoid such variability in the future [12]. The Southampton group was also involved in producing guidelines for imaging measurements of inhaled aerosol deposition. Reports of variability of measurement technique between centres for this type of investigation led the International Society of Aerosols in Medicine to call for recommendations for a standardised technique. Given the group s extensive experience in this area, we were able to take a leading role in the development of these guidelines. [13] Some interesting diversions Not all our projects led to direct clinical applications. The most notable of these was the Compton Camera, the idea for which had come from the Department of Electrical Engineering at the University of Southampton, following a visit to the hospital in the early 1970s. This was a completely novel idea to use measurement of gamma rays scattered by the Compton Effect to produce images. In conventional imaging, Compton scatter degrades image quality, but with this new concept, they could actually be used to position the initial direction of the gamma rays without the need for the collimator used in most gamma cameras [14]. Although a wonderful idea, the technical difficulties of implementation meant that it has never been developed into a clinical device. Nevertheless the concept is still being worked on by other groups some 40 years later. Other interesting areas of work which were never practically implemented were the use of coded apertures [15] and maximum entropy [16].

7 Figure 8. The principle of the Compton Effect Gamma Camera introduced in the 1970s by the Southampton group. The first two interactions of the gamma ray in the detector are recorded and from this the trajectory of the incident photon is determined to be somewhere on a cone with its apex at the position of the first interaction. Clinical Applications Renal medicine Investigation of renal function had been pioneered in the UK in the late 1960s using probe systems to monitor the kidney s handling of intravenously injected radiopharmaceuticals. The availability of gamma cameras in the early 1970s gave the opportunity for improved data, due to ability to image the kidneys. The Southampton department were quick to realise this potential, and we were one of the earliest groups to implement gamma camera renography [17]. This clinical application was enhanced by the work of the physics group, who developed novel methods of analysis, in particular the use of deconvolution to obtain measures of absolute transit through the kidney. Later in the 1980s, the group collaborated with Dr Derek Waller in the Department of Clinical Pharmacology to implement a non-imaging radionuclide technique for measuring GFR. We made important contributions to the development of this technology [18] and its clinical application [12]. Hepatic imaging Static imaging of the liver using radiolabelled colloids taken up by the reticuloendothelial system was one of the most common nuclear medicine investigations of the 1970s and 1980s, with applications in cirrhosis and liver cancer. The group initially became aware of the potential of dynamic imaging of colloid uptake to measure liver blood flow via involvement in research project with the Dr Stephen Karran of the University Department of Surgery. His group were investigating the phenomenon of liver regeneration following resection. Compartment models of the body s handling of colloid were produced and enabled development of techniques for measuring both total hepatic blood flow [19] and its arterial and portal components [20]. In 1987 we hosted an international meeting on the topic in Southampton [21].

8 Figure 9. Compartmental model of the body s handling of injected colloid, which allowed quantitative measurements of liver blood flow to be obtained [20] Lung In the early 1970s the group began its interest in the measurement of lung function using a gamma camera [22]. Although that particular line of research came to an end, we later linked up with Professor Stephen Holgate, a respiratory physician in the Faculty of Medicine at the University of Southampton. This led to a very fruitful collaboration investigating the use of nuclear medicine imaging to measure the fate of inhaled aerosol in the body. It gave the opportunity for a practical application of our SPECT-CT technology which enabled quantitative measurements of the 3D distribution of aerosol deposition from SPECT to be obtained and related to lung anatomy from CT. This led via a number of key collaborations to the development of our shell analysis, describing the regional distribution of aerosol in the lung in 3D [23], anatomical models of the airway tree [24] and techniques for estimating the deposition by airway generation [25]. These methods now form part of the international guidelines on this methodology [13]. Our clinical projects in this area were led by Professor Joy Conway of the Faculty of Health Sciences. This involvement in lung imaging research led to us becoming part of the Southampton Respiratory Biomedical Research Unit in 2008, funded by the National Institute of Health Research. The physics group still has strong links with the Southampton Respiratory Imaging Group which is part of the Respiratory BRU. 25 Percentage per shell Shell number Shell number Figure 10 Left: diagram showing the principle of shell analysis which divides the lung into zones approximating to airway generation. Right: the results of applying shell analysis to quantify the distribution pattern of aerosol in the lung [23]. This information is used in optimising design of devices used for inhalation therapy for asthma and other lung diseases. Targeted Therapy

9 The clinical head of Nuclear Medicine in the 1970s and 1980s, Professor Duncan Ackery and his successor Dr Val Lewington had their principal research interest in targeted radionuclide therapy. The Department of Nuclear Medicine was in the forefront of the developments in this field, making important contributions to the development of mibg for the treatment of bone neuroendocrine tumours and strontium-89 for the palliation of bone metastases. The physics group had an important role in developing techniques for measuring doses delivered by these techniques [26-27], another important application of gamma camera image quantification. Figure 11. Left: transaxial and Right: coronal views of the uptake of indium-111 labelled antibody in the bone marrow in a subject with myeloma. The uptake pattern is superimposed on aligned CT images showing the bone structure. This imaging enables calculation of the dose delivered in subsequent treatment using Yttrium-90 labelled antibody. The Brain The arrival in Southampton of nuclear medicine physician, Dr Paul Kemp, during the late 1990s resulted in a step change in our involvement in brain imaging. Technetium 99m labelled HMPAO was being used in the diagnosis of Alzheimer s dementia, but conventional visual interpretation of the scans was difficult. Dr Kemp was convinced that computer analysis of the images would help solve the problem by providing more objective interpretation. This gave the opportunity for the physics team to develop the use of Statistical Parametric Mapping for this purpose [28]. With the emergence of dopamine transporter imaging in the diagnosis of Parkinson s disease (DaTSCAN), quantification was again considered important to improve objectivity of diagnosis and the physics team developed a technique for quantitative analysis that was taken up commercially [29]. The group is still currently involved in both national and international initiatives for standardizing brain image analysis techniques. Figure 12. Left: images of the uptake of technetium-99m labelled HMPAO in the brain, illustrating the pattern of cerebral perfusion. Right: an example of analysis using statistical parametric mapping. Areas of abnormal perfusion in a subject with Alzheimer s dementia are shown in colour superimposed on a structural image of the brain surface obtained from a computer model.

10 References 1 Fleming, J.S. A technique for the absolute measurement of activity using a gamma camera and computer. Physics in Medicine and Biology 24(1), 1979, Fleming, J.S., Goddard, B.A. A technique for the deconvolution of the renogram. Physics in Medicine and Biology, 19(4), 1974, Fleming, J.S., Ackery, D.M., Walmsley, B.H., Karran, S.J. Scintigraphic estimation of the arterial and portal blood supplies to the liver. Journal of Nuclear Medicine 24, 1983, Fleming, J.S., Britten, A.J., Perring, S., Keen, A.C., Howlett, P.J. A general software system for the handling of medical images. Journal of Medical Engineering Technology 15 4/5, 1991, Fine, D., Perring, S., Herbetko, J., Hacking, C.N., Fleming, J.S., Dewbury, K. Three dimensional ultrasound of the gallbladder and the dilated biliary tree: reconstruction from real time scans. British Journal of Radiology 64, 1991, Perring, S., Hunt, T.M., Fleming, J.S., Blaquiere, R.M., Taylor, I. Automated measurement of tumour extent in patients with colorectal liver metastases from X-ray computed tomography. British Journal of Radiology 64, 1991, Fleming, J.S. A technique for using CT images in attenuation correction and quantification in SPECT. Nuclear Medicine Communications, 10, 1989, Fleming, J.S., Simpson, D.E. A technique for simulation of the point spread function of a gamma camera. Physics in Medicine and Biology 39, 1994, Fleming, J.S., Alaamer, A.S. A rule based method for context sensitive threshold segmentation in SPECT using simulation. Physics in Medicine and Biology, 43(8), 1998, Fleming, J.S., Whalley, D.R., Skrypniuk, J.V., Jarritt, P.H., Houston, A.S., Cosgriff, P.S, Bailey, D UK audit of relative lung function measurement from planar radionuclide imaging Nuclear Medicine Communications 25, Fleming, J.S, Cosgriff, P.S., Houston, A.S., Jarritt, P.H., Skrypniuk, J.V., Whalley, D.R., United Kingdom audit of relative renal function measurement using DMSA scintigraphy. Nuclear Medicine Communications 19, 1998, Fleming, J.S., Zivanovic, M.A., Blake, G.M. Burniston, M., Cosgriff, P.S Guidelines for the measurement of glomerular filtration rate using plasma sampling Nuclear Medicine Communications 25, Fleming J, Bailey DL, Chan H-K, Conway J, Kuehl PJ, Laube BL, Newman S. Standardization of techniques for using Single-Photon Emission Computed Tomography (SPECT) for aerosol deposition assessment of orally inhaled products. Journal of Aerosol Medicine and Pulmonary Drug Delivery, 2012, 25 (S1), S29-S Everett,D.B., Fleming, J.S., Todd, R.W., Nightingale, J.M. Gamma radiation imaging system based on the compton effect. Proc IEE 124(11) , (Awarded the Ambrose Fleming Premium by the IEE in 1977). 15. Fleming, J.S., Goddard, B.A. A comparison of techniques for stationary coded aperture imaging in nuclear medicine. Medical and Biological Engineering and Computing 20, 1982, Simpson, D.E., Fleming, J.S., Aldous, A.J., Daniell, G.J. Deconvolution of planar scintigrams by maximum entropy. Physics in Medicine and Biology, 40, 1995, Kenny, R.W., Ackery, D.M., Fleming, J.S., Goddard, B.A., Grant, R.W. Deconvolution of the scintillation camera renogram. British Journal of Radiology 48, 1975, Waller, D.G., Keast, C.M., Fleming, J.S., Ackery, D.M. Measurement of glomerular filtration rate with Tc-99m DTPA - a comparison of plasma clearance techniques. Journal of Nuclear Medicine, 28, 1987, Karran, S.J., Eagles, C.J., Fleming, J.S., Ackery, D.M. In-vivo measurement of liver perfusion in the normal and partially hepatectomised rat using Tc-99m sulphur colloid. Journal of Nuclear Medicine 20(1), 1979, Fleming, J.S., Ackery, D.M., Walmsley, B.H., Karran, S.J. Scintigraphic estimation of the arterial and portal blood supplies to the liver. Journal of Nuclear Medicine 24, 1983, Fleming, J.S., Britten, A.J. The role of radionuclides in assessing hepatic haemodynamics and arterial therapy. Nuclear Medicine Communications 8, 1987, Fleming, J.S., Goddard, B.A. Regional ventilation assessment by transmission scintigraphy. Acta Radiologica, 12(5), 1973,

11 23. Perring, S., Summers, Q., Fleming, J.S., Nassim, M.A., Holgate, S.T. A new method of quantification of the pulmonary regional distribution of aerosols using combined CT and SPECT and its application to nedocromil sodium administered by metered dose inhaler. British Journal of Radiology 67, 1994, Fleming, J.S., Nassim, M.A., Hashish, A.H., Bailey, A.G., Conway, J.H., Holgate, S.T., Halson, P., Moore, E.A., Martonen, T.B. Description of pulmonary deposition of radiolabeled aerosol by airway generation using a conceptual three dimensional model of lung morphology. Journal of Aerosol Medicine, 8(4), 1995, Fleming, J.S., Conway, J.H., Holgate, S.T., Bailey, A.G., Martonen, T.B. Comparison of methods for deriving aerosol deposition by airway generation from three-dimensional radionuclide imaging Journal of Aerosol Science. 31(10), 2000, Tristam, M., Alaamer, A., Fleming, J.S., Lewington, V.J., Zivanovic, M.A. Dosimetry of I-131 metaiodobenzylguanidine in cancer therapy. Journal of Nuclear Medicine, 37, 1996, Blake GM, Zivanovic MA, McEwan AJ, Batty VB, Ackery DM. 89Sr radionuclide therapy: dosimetry and haematological toxicity in two patients with metastasising prostatic carcinoma. Eur J Nucl Med. 1987;13(1): Kemp, P.M., Holmes, C., Hoffmann, S.M.A., Bolt, L., Holmes, R., Rowden, J, Fleming, J.S. Alzheimer s disease: differences in Tc-99m HMPAO SPECT scan findings between early and late onset dementia. Journal of Neurology, Neurosurgery and Psychiatry Tossici-Bolt L, Hoffmann SM, Kemp SM, Mehta RL, Fleming JS Quantification of [(123)I]FP-CIT SPECT brain images: an accurate technique for measurement of the specific binding ratio. European Journal of Nuclear Medicine and Molecular Imaging. 33: Staff of the Nuclear Medicine Physics Group Mr Alan Aldous ( ) (now in Nuclear Medicine in Ipswich) Dr Glen Blake ( ) (now senior lecturer at Guys and Thomas Hospital, London) Dr Alan Britten ( ) (now Head of Medical Physics at St Georges Hospital, London) Dr Kevin Byard ( ) Professor Barrie Condon ( ) (now Professor in Magnetic Resonance at Southern General Hospital, Glasgow) Professor John Fleming (1971-present) (Head of Group from ) (now Consultant Physicist with the Respiratory NIHR Biomedical Research Unit in Southampton) Dr Keith Goatman ( ) (now in Medical Physics in Aberdeen) Dr Tony Goddard ( ) (Head of Group ) (went on to be Head of Medical Physics in Southampton, now retired) Mr John Gray ( ) Ms Gemma Lewis (2009-present) Dr Matthew Guy (2010-present) (Head of Group 2010-present) Mr Peter Halson ( ) (now working with General Electric) Prof Dave Hawkes ( ) (now Head of the Centre for Medical Image Computing at the University of London) Dr Alex Hoffmann ( ) (now working for Nuclear Medicine at Salford) Dr Peter Howlett (Head of group ) (went on to be Head of Medical Physics and then Finance Director at Portsmouth Hospitals) Mr Bob Kenny ( ) (now Head of Link Medical) Mr Andrew Keen ( ) Dr Alessandra Malaroda ( ) (moved to Australia) Dr Ray Pope ( ) (went on to be Head of Radiotherapy Physics in Southampton, died in 2013) Dr Steve Perring ( ) (now at Bournemouth and Poole Hospital) Dr Dave Taylor ( ) (went on to be Head of Nuclear Medicine Physics at Coventry) Dr Livia Tossici-Bolt (1992-present) Dr Maria Tristam ( ) (retired) Mr Efstathios Varzarkis (2011-present)