3D Visualization for Pre-operative Planning of Orthopedic Surgery

LIU-ITN-TEK-A-13/014-SE 3D Visualization for Pre-operative Planning of Orthopedic Surgery Alexander Steen Marcus Widegren 2013-05-22 Department of Science and Technology Linköping University SE-601 74 Norrköping, Sweden Institutionen för teknik och naturvetenskap Linköpings universitet 601 74 Norrköping

LIU-ITN-TEK-A-13/014-SE 3D Visualization for Pre-operative Planning of Orthopedic Surgery Examensarbete utfört i Medieteknik vid Tekniska högskolan vid Linköpings universitet Alexander Steen Marcus Widegren Handledare Timo Ropinski Examinator Sasan Gooran Norrköping 2013-05-22

Upphovsrätt Detta dokument hålls tillgängligt på Internet eller dess framtida ersättare under en längre tid från publiceringsdatum under förutsättning att inga extraordinära omständigheter uppstår. Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner, skriva ut enstaka kopior för enskilt bruk och att använda det oförändrat för ickekommersiell forskning och för undervisning. Överföring av upphovsrätten vid en senare tidpunkt kan inte upphäva detta tillstånd. All annan användning av dokumentet kräver upphovsmannens medgivande. För att garantera äktheten, säkerheten och tillgängligheten finns det lösningar av teknisk och administrativ art. Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman i den omfattning som god sed kräver vid användning av dokumentet på ovan beskrivna sätt samt skydd mot att dokumentet ändras eller presenteras i sådan form eller i sådant sammanhang som är kränkande för upphovsmannens litterära eller konstnärliga anseende eller egenart. För ytterligare information om Linköping University Electronic Press se förlagets hemsida http://www.ep.liu.se/ Copyright The publishers will keep this document online on the Internet - or its possible replacement - for a considerable time from the date of publication barring exceptional circumstances. The online availability of the document implies a permanent permission for anyone to read, to download, to print out single copies for your own use and to use it unchanged for any non-commercial research and educational purpose. Subsequent transfers of copyright cannot revoke this permission. All other uses of the document are conditional on the consent of the copyright owner. The publisher has taken technical and administrative measures to assure authenticity, security and accessibility. According to intellectual property law the author has the right to be mentioned when his/her work is accessed as described above and to be protected against infringement. For additional information about the Linköping University Electronic Press and its procedures for publication and for assurance of document integrity, please refer to its WWW home page: http://www.ep.liu.se/ Alexander Steen, Marcus Widegren

3D Visualization for Pre-operative Planning of Orthopedic Surgery Alexander Steen Marcus Widegren June 15, 2013

Abstract This report presents a master thesis on 3D visualization for pre-operation planning of orthopedic surgery done for Sectra Medical Systems AB. The focus is on visualizing clinically relevant data for planning a Total Hip Replacement (THR). The thesis includes a pre-study and the implementation of a prototype using the Sectra IDS7 workstation.

Contents 1 Introduction 4 2 Background 5 2.1 Sectra Medical Systems AB..................... 5 2.1.1 Sectra Orthopedics...................... 5 2.2 Purpose and Scope of the Thesis.................. 5 2.3 Medical Background......................... 6 2.3.1 Anatomy........................... 6 2.3.2 Osteoarthritis......................... 7 2.3.3 Total Hip Replacement................... 8 2.3.4 Total Knee Replacement................... 10 2.3.5 Pre-operative Planning................... 11 2.4 Medical Imaging........................... 12 2.4.1 Projectional Radiography.................. 12 2.4.2 X-ray Computed Tomography................ 13 2.4.3 Magnetic Resonance Imaging................ 13 2.4.4 Picture Archiving and Communication System...... 13 2.5 Related work............................. 13 2.5.1 Related work in 2D..................... 13 2.5.2 Related work in 3D..................... 14 3 Method 15 3.1 Pre-study............................... 15 3.1.1 Competitors.......................... 15 3.1.2 Comparison Between 2D and 3D.............. 15 3.1.3 Proposed Visualization Methods.............. 16 3.1.4 Proposed Interaction Methods............... 18 3.1.5 Coordinate Systems..................... 18 3.2 Prototype............................... 19 3.2.1 Basic Visualization and Control............... 20 3.2.2 Visualization of Relevant Data............... 20 3.2.3 Acetabular Cup........................ 23 3.2.4 Improved Control....................... 24 3.2.5 Measurements........................ 25 1

CONTENTS CONTENTS 4 Discussion 28 4.1 Current limitations.......................... 28 4.2 Radiation dose............................ 28 4.3 Widgets................................ 29 4.4 Acetabulum Cup........................... 29 4.5 Bone cement............................. 29 4.6 Future work.............................. 29 4.6.1 Metadata........................... 29 4.6.2 Usability testing....................... 30 4.6.3 Image analysis........................ 30 4.6.4 Repositioning......................... 30 5 Conclusion 31 2

Glossary Voxel Background radiation Statistical significance A volume element representing a value in a 3D grid. Analogous to a pixel which represents a value in a 2D grid. The combined ionizing radiation that the general population is exposed to. Includes natural and artificial sources. The probability that observations are not just obtained by chance. The observation is considered statistically significant if this probability is below some pre-defined value. Anterior-posterior (AP) Anatomical direction pointing from the front of the body to the back. Posterior-anterior (PA) Lateral Medial Anteversion Sievert (Sv) MPR Glyph Anatomical direction pointing from the back of the body to the front. Anatomical direction pointing out from the body to the side. Anatomical direction pointing from the side towards the body. Forward rotation of an anatomical structure. The SI-unit for radiation dose. Multi-planar Reconstruction. A simple 2D representation of the voxel values of one slice through the volume. A visible marker, such as an arrow, used to specify part of a visualization.

Chapter 1 Introduction Approximately one million Total Hip Replacements (THRs) are performed annually in the world[1]. It s a procedure commonly done because of osteoarthritis inthehipjoint. Whenthesurgerybeginsitisimportantthatthechosenimplant fits the patient well and does not lead to complications, like pain or unequal leg length. Pre-operative planning is an important step in predicting and preventing complications [2]. The conventional method for planning for a THR has been to use hard-copies of X-ray images and printed out templates for implants to find out which implant to use and how it should be inserted. Today it is more common to use software for pre-operative planning that is connected to the hospital s Picture archiving and communication system (PACS). Studies have shown that using a digital planning tool is as accurate as conventional planning [3, 4]. Most of the currently available planning tools are in 2D, using X-ray images and 2D templates of the implants. Some research has been done to investigate the value of using 3D planning tools instead of 2D with mostly positive results [5, 6]. The purpose of this thesis work is to research the usefulness of extending Sectra s orthopedic planning tool to visualizing the planning in 3D and to implement a prototype in Sectra Workstation IDS7. The main contribution of this work is a way of visualizing the fit of a hip implant while still giving the user contextual information about the bone anatomy. 4

Chapter 2 Background This chapter explains some key concepts, presents the general background of the medical procedures and the purpose of this thesis. 2.1 Sectra Medical Systems AB Sectra Medical Systems is part of Sectra AB, a company that is active in the business areas Medical Systems and Secure Communication Systems. Sectra is headquartered in Mjärdevi Linköping and has offices in twelve countries. The main products of Sectra Medical Systems are their Radiology information system (RIS) and their PACS. RIS is a journaling system for radiology and PACS is an imaging software for radiology. Sectra s RIS and PACS are usually sold together with Sectra management tools as one package called Sectra Diagnostic Imaging Suite. For more information about PACS, see section 2.4.4 at page 13. 2.1.1 Sectra Orthopedics Sectra orthopedic package is a software package for planning orthopedic surgery. It includes tools for hip, knee and fracture surgery. The hip module includes tools that help the user find an implant that will fit the patient well using image analysis. It can also predict and compensate for unequal leg length. The knee module includes similar functionality for the knee. 2.2 Purpose and Scope of the Thesis Sectra is interested in the idea of using 3D for visualization and operation planning. They are interested in seeing what possibilities that are introduced with 3D as well as potential drawbacks. The purpose of this thesis is therefore to investigate the usefulness of 3D as a tool for operation planning and demonstrate the findings with a prototype application. The thesis will discuss operations 5

2.3. MEDICAL BACKGROUND CHAPTER 2. BACKGROUND both for the hip and the knee but will focus mainly on the hip. The developed prototype was also specifically designed for the hip. Oneoftheconcernsofobtaining3Dvolumesofhumanbodiesisthatdepending on the method used, it can expose the person to a higher dose of radiation. As this is an important factor, it will be the topic of some discussion in the thesis. However, comparing the risks of higher radiation with the advantages of improved planning will not be the main focus of this thesis. 2.3 Medical Background 2.3.1 Anatomy This section will discuss the parts of the body that are within the scope of this thesis. These are the thigh bone (femur), the pelvis and the knee. The top part of the femur together with part of the pelvis can be seen in figure 2.1. The names of the different parts that are important for preoperative planning are also marked in the figure. The head of the femur (caput) is connected to the pelvis via a socket in the pelvis (acetabulum) and is held in place by tendons (not pictured in the figure). The surface of the caput is covered with a layer of cartilage that together with a lubricating fluid (called synovial fluid ) allows the caput to rotate with low friction. Figure 2.1: Top part of the thigh bone(femur) and part of the pelvis An illustration of the knee can be seen in figure 2.2. The knee connects 6

2.3. MEDICAL BACKGROUND CHAPTER 2. BACKGROUND the femur and the shinbone (tibia) and holds these together with a number of tendons. Between these bones there is cartilage as well as other materials that act as shock absorbers. 2.3.2 Osteoarthritis Figure 2.2: Illustration of a knee Both the knees and the hips are under practically constant stress daily. Because of this, most people run a risk of developing osteoarthritis with age. Osteoarthritis is the degeneration of cartilage in a joint. Two common areas where this may occur in the body are in the cartilage of the caput and in the knee. As the cartilage degrades, the body may try to compensate by producing more bone. Part of the smooth cartilage is then replaced by rough bone. This can eventually lead to bone rubbing against bone, causing friction and pain. Osteoarthritis can be diagnosed using X-rays. Even though cartilage itself does not show on an X-ray due to being too soft, the lack of cartilage can be identified using X-rays. There is a certain amount of space between caput and acetabulum and in the knee where the cartilage is. If this space (called joint space ) is smaller than it should be, that indicates a lack of cartilage. An example of this can be seen in figure 2.3. More information on osteoarthritis can be found in Ref. [7]. 7

2.3. MEDICAL BACKGROUND CHAPTER 2. BACKGROUND (a) Healthy amount of joint space. (b) Absence of joint space. Figure 2.3: Two X-ray images, (a) showing some space between caput and acetabulum in the top right part of the picture, indicating the presence of cartilage and (b) showing less space indicating a lesser amount of cartilage. If the symptoms of osteoarthritis are not too severe, they can be treated with medication. For severe osteoarthritis in the hip or knee, the solution is often a total hip/knee replacement. 2.3.3 Total Hip Replacement In a THR (figure 2.4) the caput is completely replaced by a metallic implant consisting of a stem and a head. While the outer shell of the femur (called cortical shell) is hard, the marrow canal inside is softer. To allow the implant stem to fit in the bone, parts of the marrow canal is removed using broaches of increasing size. A so called cup is inserted in the acetabulum which the head is then inserted into. 8

2.3. MEDICAL BACKGROUND CHAPTER 2. BACKGROUND Figure 2.4: AP X-ray image of a patient after a total hip replacement. The right hip joint has been replaced by an implant and fastened to the inside of the femur using bone cement. The implant can be either cemented or non-cemented. Cemented means that the implant is fastened using bone cement. After a cemented implant is inserted into the bone, the surgeons will have to wait about 15 minutes for it to set. A non-cemented implant has a rough surface which will allow the bone to grow into it over time, fusing the bone and implant together. After a hip replacement, the patient is usually able to walk with a walker within a day. The patient can usually be sent home after a couple of days but will have to take extra precautions to ensure a successful recovery. A full recovery usually takes a few months [8]. The different parts of the implants come in many different sizes and models, a sample of the different parts can be seen in figure 2.5. 9

2.3. MEDICAL BACKGROUND CHAPTER 2. BACKGROUND Figure 2.5: Photograph of a titanium stem, a ceramic head and a polyethylene acetabular cup. 2.3.4 Total Knee Replacement Total Knee Replacement (TKR) is a procedure where the load bearing parts of the knee joint are replaced by implants. Like THR, the most common cause for this procedure is osteoarthritis. Figure 2.6 shows two X-ray images of a knee after a TKR. As this thesis focuses mostly on THR, TKR will not be described in detail here. For detailed information about TKR we refer the reader to [9]. 10

2.3. MEDICAL BACKGROUND CHAPTER 2. BACKGROUND (a) AP radiograph of a replaced knee. (b) Lateral radiograph of a replaced knee. Figure 2.6: An AP (a) radiograph and a lateral (b) radiograph of a replaced knee. Images courtesy of Wikipedia user FpJacquot. 2.3.5 Pre-operative Planning Pre-operative planning helps the surgeon prepare for the operation. It helps decide what tools will be necessary, how the procedure will be performed and also helps anticipate possible problems that may occur during the procedure. One of the purposes of the pre-operative planning in a THR/TKR is to predict which implant model should be used as well as the sizes of the different parts of the implant. It is important that the implant fits well and therefore the surgical team will bring a number of implants in different sizes to be able to find a good match. A good prediction on the resulting implant will allow the surgical team to bring fewer implants into the operating room, lowering the number of tools that need sterilization. A good size estimate will also help decrease the time for the operation, and thereby lowering the infection risk [10, 11, 12]. Two other important factors regarding the acetabular cup are how much bone that will need to be removed to reach fresh (bone that bleeds) bone and at which angle the cup is to be inserted. If the angle of the cup is off, the stem might become dislocated if the patient bends their leg too much. 11

2.4. MEDICAL IMAGING CHAPTER 2. BACKGROUND Figure 2.7: Leg length discrepancy and compensation for it When suffering from osteoarthritis, the loss of cartilage may lead to one leg becoming shorter than the other. This is called leg length discrepancy. The removal of bone and insertion of an implant during a THR may also alter the leg length. An illustration can be seen in figure 2.7. The left figure shows leg length discrepancy and the right figure shows a compensation made by choosing an appropriate implant. Another important factor is the femoral offset (also illustrated in figure 2.7), or the perpendicular distance from the center of rotation of the head of femur to the long axis of femur. Increasing the femoral offset after a hip replacement has been shown to have positive effects on the range of abduction and abductor strength [13]. A good plan will help the surgeon to solve the discussed issues by choosing implants accordingly. 2.4 Medical Imaging Medical imaging is the process of obtaining images of the human body for use in medicine, such as diagnosing diseases and examining anatomy. This section will describe some areas of medical imaging relevant to this thesis. 2.4.1 Projectional Radiography In projectional radiography, images are obtained by having electromagnetic radiation such as X-rays hit a film after passing through an object (e.g. a human body). The varying densities of the object will absorb different amounts of radiation, resulting in an image where dense parts of the objects appear opaque and less dense parts appear transparent. This is the traditional method commonly referred to simply as X-rays. When performing an X-ray on the pelvis, the average effective dose of radiation is 0.6 msv [14]. As a reference, the annual background radiation exposed to the body is approximately 3 msv [14]. 12

2.5. RELATED WORK CHAPTER 2. BACKGROUND 2.4.2 X-ray Computed Tomography X-ray Computed Tomography (CT) is an imaging technique where X-rays are used to create tomographic images, or slices, of specific parts of the human body. These slices can be combined into a 3D volume of the body part that has been scanned. Approximately 62 million CT scans are performed annually in the United States [15]. CT is well suited for the detection of bone, which makes it suitable for orthopedic planning. However, a CT scan exposes the body to a higher dose of radiation than a normal X-ray image. A pelvic CT scan exposes the body to approximately 6 msv of radiation [14]. This is equivalent to two years of background radiation, or 10 pelvic X-ray images. 2.4.3 Magnetic Resonance Imaging Magnetic Resonance Imaging (MRI) is an imaging technique that uses strong magnetic fields and radio waves to acquire slices of a body part. MRI does not use ionizing radiation which means it does not introduce additional cancer risks the way planar radiography and CT do. MRI is best suited for soft tissue and is not very well suited for bone. Therefore visualization with MRI data sets will not be included in this thesis. 2.4.4 Picture Archiving and Communication System PACS is a technology used by hospitals for storing and accessing medical images from multiple locations, both within a hospital and between different hospitals. It replaces hard-copies and the need to store and archive them. A PACS typically consists of a server infrastructure and workstations where the radiologists can view images taken anywhere that is connected to the same PACS. A PACS can also include tools for pre-operative planning and 3D visualization of CT or MRI scans. 2.5 Related work The importance of planning before performing an operation is well-understood and documented [2]. Planning helps the surgeon to anticipate the correct implant size and can also help anticipate possible intra-operative difficulties (more than 80% of intra-operative difficulties were anticipated in the study performed in [2]). At the time of writing, pre-operative planning is primarily done in 2D using conventional X-ray films or digital X-ray images but there is research being done in using 3D in the planning process. 2.5.1 Related work in 2D There have been studies that compare traditional X-ray films and digital X-ray images [3, 4]. In their study, The et al. even concluded that digital plans slightly outperform analogue plans in their accuracy regarding the implant size. Today, 13

2.5. RELATED WORK CHAPTER 2. BACKGROUND there are a number of well-established companies that offer software that allows the surgeon to perform planning in 2D before going into the operation room. 2.5.2 Related work in 3D The accuracy of using 3D templates in planning has been studied. One study published by Kobayashi et al. in 2012 compared 2D analogue planning with digital 3D planning for TKR [5]. The results of the study indicated that 3D planning performs somewhat better than 2D. However, the results were not statistically significant. The study by Sariali et al. compared the accuracy of analogue 2D and 3D planning in THR and found that when counting both the stem and the cup in a THR, the implants that were used ended up being the same that was planned in 96% of the cases when using 3D and in 16% of the cases when using 2D [6]. Another study also found 3D planning to be accurate and repeatable, especially among less experienced surgeons [16]. In their article, Dick et al. discuss two methods for visualizing the 3D planning [17]. The article puts its main focus on the important distances present in operation planning. It presents two different approaches and investigates the advantages and disadvantages of the two. One of the methods used colored slices to visualize the distance and this method was used as a starting point for the visualization in this thesis. The article also discusses some of the inherent problems with 3D visualization such as visual cluttering and occlusion and presents ways of dealing with them. 14

Chapter 3 Method 3.1 Pre-study This section will go through the findings of the pre-study. The pre-study consists of a literature study, a look into what Sectra s competitors are doing and some proposals and general ideas of our own. 3.1.1 Competitors There are a number of companies that offer tools for pre-operative planning. Some of these tools are OrthoView, TraumaCAD and MediCAD. All of these and more were looked into in order to get an overview of what is available at the moment. Most of the studied programs only offered 2D planning but there were some that offered 3D to a certain extent. TraumaCAD offers the user the option to view the implant in a 3D view after using 2D views for the planning, but the planning itself is not done in 3D. Two programs that do offer 3D planning are KneeCAS/HipCAS and HipOp. HipOp is a free software for surgical planning conceived by Istituto Ortopedico Rizzoli and developed by the Bio-Computing Competence Centre. 3.1.2 Comparison Between 2D and 3D The obvious main advantage of using 3D data sets is that it contains more information. Instead of seeing the resulting attenuation along an X-ray you have values for each voxel. The sizes are more accurate and the structures are invariant to how the patient has rotated their limbs. There are however some drawbacks in using 3D instead of 2D. One issue is that it is generally harder to intuitively navigate in a 3D environment since the mouse and the computer screen only operate in two dimensions. Another disadvantage is that it is harder to present a general overview due to occlusion. Also specific to this is that the learning curve can be steep for people used to working with X-ray images. This might be more of a problem for radiologists than orthopedists. The study that 15

3.1. PRE-STUDY CHAPTER 3. METHOD was performed by Atsushi Kobayashi et al. showed that 3D templating had a slight advantage over 2D in accuracy [5]. However, the result was found to not be statistically significant. Viceconti et al. concluded in their study that their 3D orthopedic templating tool (HipOp) was comparable in usage to the conventional planning with radiographs [16], indicating that the learning curve might not be so bad. They also concluded that the sizes predicted by their tool were generally more accurate than the sizes predicted with conventional planning, especially for the acetabular cup and especially for less experienced surgeons. Otomaru et al. [18] presented a method for creating pre-operative plans for THRs using CT. They concluded that the plans created by their automatic software is roughly equivalent to the plan created by an experienced surgeon using the conventional method of radiographs and templates. They argue that this method will not only save time for experienced surgeons but also let less experienced surgeons create plans as good as if they were made by experienced surgeons. 3.1.3 Proposed Visualization Methods This section will present the ideas for visualization and interaction methods that were produced during the pre-study. The methods were developed by us in collaboration with developers and a medical expert at Sectra, and with the help of ideas and studies from the literature. Visualization of the Hip Implant Stem and Neck in Femur: By rendering the bone semi-transparent and the implant opaque, the implant will not occlude the bone. The level of transparency for the bone should fade when further away from the implant to provide some contextual information. Thedistancetothehardcorticalshelloffemurfromapointontheimplant could be visualized by shading the implant in different colors depending on the distance. Oriented glyphs on the implant could be used to visualize the distance to the nearest point on the cortical shell. To help visualize the structure of the bone, opaque color coded slices orthogonal to the anatomical axis of femur could be added. The color should depend on the distance to the implant surface and the distance to the cortical shell surface. If using few slices, contours should be shown at regular intervals so that the shape of femur can still be seen clearly even though it is semi-transparent. 16

3.1. PRE-STUDY CHAPTER 3. METHOD To be able to get detailed information of a specific part of the femur, a top-down view of a selected slice, separate from the 3D view could be added. This view should be added in a Multi-Planar Reformatting (MPR) view. The color coding will draw attention to the slice (due to pre-attentive processing) while the MPR view will give some contextual information in a format surgeons (or at least radiologists) are used to. A top-down view of a combination of a few slices, or a certain section of femur could be shown. This could visualize relevant information such as max/min values and variance for each pixel location. Cemented stems need some extra space for the bone cement. The cement coating could therefore be shown as a dotted line around the stem contour on the slice. The cemented coating could be displayed as a semi-transparent surface around the implant surface. It would have to be rendered in a color that enables the user to differentiate it from bone. Visualization of the Acetabular Cup: Combined anteversion could be visualized by a separate widget showing the angle (or the difference between the angle and the center of the safe zone ) with two lines, a bow between them and a number. How much bone that needs to be removed in order to get to fresh bone should be visualized. A red-green isoluminant color-map could be used, red meaning a lot of bone and green meaning none. The amount of bone that needs to be removed could be rendered as a transparent red volume. Render the whole volume transparent except for the bone that needs to be removed. In order to keep the contextual information, the transparency level should fade away. Visualizing the Tibial Tray: Since the cortical shell should carry most of the load, how well the tray covers the shell should be clearly visible. The cortical shell slice that is to be covered could be color-coded, green meaning that part has coverage and red meaning that part is not covered. Render everything except the bone that is to be removed transparent. 17

3.1. PRE-STUDY CHAPTER 3. METHOD Visualizing Femur for TKR: Emphasize the bone that is to be removed by for example rendering everything else transparent, or showing the bone that is to be removed as a semi-transparent red volume. 3.1.4 Proposed Interaction Methods Creating intuitive interaction in a 3D visualization is quite a challenge. We assume that the user will use a normal mouse and keyboard and a regular 2D computer screen. The slices should be selectable and movable. Hovering with the mouse over the top-down view of a slice should make the application present the distance between the point on the implant closest to the cursor and the cortical shell. Selecting one slice could toggle all other slices transparent. Ctrl + R should reset the view to a view that is familiar to the surgeon. For example it could reset to a regular AP view of femur. The implant should be selected when the user clicks on it. The user should be able to deselect the implant. This could be done for example by clicking on it again, or clicking somewhere else. An object is moved and rotated with two widgets. One for rotation and one for translation. Both allow manipulation in one dimension at a time. 3.1.5 Coordinate Systems In order for the system to be easy to use, as well as to let it calculate relevant distances, a coordinate system relevant to the anatomical parts of interest should be established. For the coordinate system in femur we are proposing that the z-axis is parallel to the marrow canal and pointing downwards. The x-axis is orthogonal to the z-axis and pointing medially. The y-axis is defined by the line that is orthogonal to the z-axis and the x-axis. For the acetabular cup we suggest that one axis is orthogonal to the surface at the center of the socket. One axis should point laterally, and the third should simply be orthogonal to those two. Since the anteversion needs to be controlled however, the rotation should not be around these axes but around the pelvic coordinate system. Both these coordinate systems can be established by letting the user click on a few anatomical landmarks. This should be done in the MPR views for increased accuracy. Once the coordinate systems are established the implant can be inserted into a reasonable spot automatically. This should make the interaction a lot easier since placing the implant manually in the 3D view would be quite hard. 18

3.2. PROTOTYPE CHAPTER 3. METHOD 3.2 Prototype This section will describe the resulting prototype that was developed. Weekly meetings were held with Sectra to assess the status of the development and decide in which direction it should continue. The prototype was implemented as a module in Sectra s already working framework for 3D visualization. The framework offers methods for both volume rendering and polygon rendering and has different interactions between components and the user already implemented. This removed some of the implementation problems and allowed for more focus to be put on the visualization problems and usability problems previously discussed. The prototype was developed using C.NET. Figure 3.1: Screenshot of Sectra s visualization software showing an AP view of a pelvis Figure 3.1 shows a screenshot of Sectra s visualization software without any of our implementations. The three windows to the left are MPR views that show cross-sections of the volume. The volume is rendered in 3D in the large window to the right. 19

3.2. PROTOTYPE CHAPTER 3. METHOD 3.2.1 Basic Visualization and Control The different implant parts are rendered as shaded polygon models. The models are semi-transparent to keep them from occluding bone. The distance shading of the implant which is later discussed can also be seen through the implant if it is semi-transparent. The user can switch to an implant with a different size by holding the mouse pointer over the implant and scrolling the mouse wheel. When an implant is selected, it becomes somewhat less transparent and a bounding-box is rendered around it to distinguish it from the other implant parts. Figure 3.2 shows a stem and an acetabular cup inserted in femur and acetabulum. The cup is selected as can be seen from the difference in transparency and the bounding box. Figure 3.2: Basic rendering of the implant parts with the acetabulum cup selected 3.2.2 Visualization of Relevant Data One important aspect when deciding what implant to use is that it fits well. A good visualization of the fit is therefore essential. This section will describe the features that was implemented to help visualize the fit. Most visualizations are implemented in HLSL pixel shaders. 20

3.2. PROTOTYPE CHAPTER 3. METHOD Bone Contours When positioning the implant model in the CT volume, it sometimes became difficult to see how close the implant was to the cortical shell without having to keep rotating the view. The implant was also often occluded by the bone. By allowing the contours of the bone to be visualized as well as the close proximity of the bone in the form of slices, the user can get a better understanding of how well the implant fits. The contours also help show the structure of the bone with less occluding of the implant. The user can change the orientation and spacing as well as the number of slices. An implementation of the bone contour visualization can be seen in figure 3.3. (a) A close up of femur in an X-ray rendering of (b) A close up of femur in an X-ray rendering of a CT volume, with bone highlights. a CT volume, without bone highlights. Figure 3.3: Two close-up views of femur in an X-ray rendering of a CT volume. In (a) the cortical shell is highlighted to show the contours of the bone, in (b) it is not. Implant Shading The shading of the implants was changed to visualize distance to the cortical shell of femur. An implant that is far away from any bone will be shaded using regular Phong shading but when being close to bone, it will interpolate between green and blue depending on the distance to bone. Parts of the implant that are close to bone will be blue. The implant should be close to but not intersect 21

3.2. PROTOTYPE CHAPTER 3. METHOD with bone and therefore if some part of the implant intersects with bone, that part will be rendered bright red as a warning. Yellow lines that are aligned with the bone contour highlights are also drawn on the implant to make it easier to see how the outlines relate to the implant. (a) A rendering of an implant that intersects femur, indicated by the red color. tersect femur along the (b) A rendering of an implant that does not in- stem. Figure 3.4: Two renderings showing how the shading indicates when the implant intersects bone. In (a) the implant intersects the cortical shell of femur, and in (b) it does not. Implant Contour in MPR View Contours of the implants are visualized in the MPR views. For each triangle of the mesh, each edge is checked for an intersection with the plane that the MPR view defines. This gives a good outline of the mesh for a thin slice. For thicker slices, we also need to take into account edges that are inside the slice. So in addition to checking intersections with the edges, we also check if each edge of the triangle is inside the slice. If they are we fill that triangle. So if the slice contains the whole model, a filled projection of it is rendered. 22

3.2. PROTOTYPE CHAPTER 3. METHOD (a) Implant overlay in MPR view for a thin slice. (b) Implant overlay in MPR view for a thick slice. Figure 3.5: The rendering of the implants in an MPR view for a thin slice (a) and a thick slice (b). 3.2.3 Acetabular Cup To be able to visualize the cup s position relative to the volume, the cup had to be clipped by the clip plane as having the whole cup visible made it occlude the volume. The parts of the cup that are close to the clip plane are rendered yellow without any diffuse or other lighting. This is done to highlight the edges more clearly. The cup is not shaded differently with regards to intersections with bone, it only takes into consideration the distance from the clip plane. The resulting visualization can be seen in figure 3.6. 23

3.2. PROTOTYPE CHAPTER 3. METHOD (a) Cross section of acetabulum. (b) Cross section of acetabulum with a cup inserted deep into the pelvis. Figure 3.6: Cross section of acetabulum before (a) and after positioning the acetabulum cup (b). 3.2.4 Improved Control Widgets Widgets for translation and rotation in the 3D view were implemented. The coordinate system for the rotation and translation is defined by the femur coordinate system (which is set up with the planning guide), and the widgets translate and rotate in the coordinate system of the currently selected view. Planning Guide A simple planning guide was implemented to help the user find implant parts with a good fit and put these in approximately the right place. When the guide is started, the user is asked to click on a number of landmarks in the MPR views. After the landmarks positions are collected the program goes through a list of different implants to find the implant parts with the best fit and puts these into place. The camera is then moved to put focus on femur, a clip-plane is activated to give a cross-section of the bone and a number of bone contour slices are rendered. These operations make it easier to see how well the implant fits. Figure 3.7 shows the result of using the guide, with an implant automatically inserted into femur. The goal with the guide is that with good accuracy when selecting the landmarks, the guide will create an operation plan that will give the user a good starting point and will only need some tweaking instead of having to manually 24

3.2. PROTOTYPE CHAPTER 3. METHOD move the implant parts into place and going through all possible implant sizes to find one with a good fit. (a) The standard AP view for the stem in femur. (b) The standard PA view for the stem in femur. Figure 3.7: Two standard views for visualizing how the stem fits in femur. Part of the clip-plane representation can be seen as a green out-line 3.2.5 Measurements Leg Length Discrepancy To help the user measure possible leg length discrepancy, a tool for measuring and visualizing this was implemented. The tool allows the user to position and reposition four dots onto the volume. When all four dots are placed, the distance from the two lower dots to the line connecting the top two dots are calculated and displayed. By placing the dots in the same position on both the left and right femur and the same position on both sides of the pelvis, the user can get a good measurement of a possible leg length discrepancy. The result from using the tool can be seen in figure 3.8. 25

3.2. PROTOTYPE CHAPTER 3. METHOD Figure 3.8: Measurement of leg length discrepancy Femoral Offset After using the planning guide, the user will have provided the locations necessary to calculate and visualize the femoral offset. A line from the femur to the center of acetabulum will then be drawn, as well as a line through femur. Dots are placed at the end-points to make them more distinct and the femoral offset is displayed. The result can be seen in figure 3.9. 26

3.2. PROTOTYPE CHAPTER 3. METHOD Figure 3.9: Cross-section of femur with the visualization of the femoral offset 27

Chapter 4 Discussion 4.1 Current limitations Some limitations existed in the 3D engine we used to implement our prototype. The volume and the polygon models are rendered separately, which lead to transparency not always working as intended. In the case of visualizing the stem in femur, in the rendering pipeline the implant is always completely behind the volume or completely in front of the volume. This lead to the implementation of an opaque bone rendering with clip planes. After some discussion with a medical expert at Sectra we feel that this is overall a better solution, since it is much easier to see contextual information about the anatomy with this method. Repositioning of different parts of the volume is not possible in the current engine. This means you currently can not visualize how the patient would look after the surgery is complete. 4.2 Radiation dose The radiation dose from a CT scan increases the risk of cancer by approximately 1 in 2000. This is a noticeable increase from the 1 in 16000 increase by a pelvic X-ray. However the numbers are small compared to the 2 in 5 average risk we all have of developing cancer [19]. The additional risk of cancer introduced by a CT scan would also have to be weighted against the improved results obtained by using 3D planning reported from the discussed studies and the potential benefits from improving the 3D tools further. As concluded in the studies, a good plan will reduce the risk for intraoperative complications [2] as well as the risk for the need for revisional surgery (which in turn would require additional X-rays to be performed). Lowering these risks may be enough to warrant the higher dose of radiation obtained from a CT scan. 28

4.3. WIDGETS CHAPTER 4. DISCUSSION 4.3 Widgets The widget implementation differed significantly from the proposed method. In the implementation, the widgets translate and rotate in the current clip plane, which is defined by the coordinate system set up in femur. This is generally the coordinate system you would want to move the implant in. However, there is currently no way to translate the implant in the direction orthogonal to the clip plane. 4.4 Acetabulum Cup Unlike when positioning the stem in femur, it became hard to define a good or bad position for the acetabulum cup. Hard cortical shell in femur should not be removed but when inserting the cup it is necessary to remove some hard bone from the pelvis in order to insert the cup. Different orthopedists also prefer to insert the cup at different depths. These factors led to the visualization of the cup to focus on position and visibility instead of trying to visualize the fit. 4.5 Bone cement The current implementation assumes that the implant chosen is uncemented. A cemented implant needs a bit more space from the cortical shell for the cement coating to fit. Since most hip replacements in Sweden use cemented implants it is important that a production implementation of this includes support for cemented implants. 4.6 Future work This section will present some features and possible next steps for improvement that would help the prototype move closer towards a commercial product. 4.6.1 Metadata In Sectra s 2D operation planning software, implants have metadata added to them containing information such as were the neck of the stem is among other information. This data allows for features not possible with just a simple 2D representation of the implant. The implants in this thesis just contains vertex and triangle data, adding more information about the implant could allow for the adding of additional features. This metadata would probably have to be added by hand for all the different models of implants. This assumption is based on how metadata is currently added to the 2D implant models. 29

4.6. FUTURE WORK CHAPTER 4. DISCUSSION 4.6.2 Usability testing Usability testing would be a good method to assess how well the current interaction methods work in practice. The testing should ideally be performed with orthopedists familiar with using Sectra s current planning tools so that the feel and work flow of the program is consistent with Sectra s 2D planning tools. 4.6.3 Image analysis Image analysis should be used together with the planning guide to calculate which implant fits best and how it should be placed. It could also assess the current fit, as well as find anatomical landmarks to calculate leg length discrepancy. 4.6.4 Repositioning The ability to be able to visualize in 3D how the leg will look after the surgery would be a useful aid. It could help visualize leg length discrepancy and femoral offset and show how different implants would affect these issues. 30

Chapter 5 Conclusion 3D visualization and planning seems to have potential for improving the performance of pre-operative planning, as indicated by both our implemented prototype and the presented studies. The prototype was able to show relevant data and also visualize elements that are not as easily visualized in 2D. However, one of the biggest challenges for future work would be to make the interactions as intuitive as they are in 2D, as people working in the medical field has a long history of working in 2D and may not be as familiar with 3D. Furthermore, if the potential improvement of using 3D is large enough it might outweigh the extra risks from CT scans. 31

Bibliography [1] G. C. Uwe Holzwarth, Total Hip Arthroplasty, Institute for Health and Consumer Protection, European Union Joint Research Center, 2012. [2] S Eggli, M Pisan, M.E. Müller, The value of preoperative planning for total hip arthroplasty, The Journal of Bone Joint Surgery (British Volume), vol. 80, no. 3, pp. 382 390, 1998. [3] Bertram The et al., Digital Versus Analogue Preoperative Planning of Total Hip Arthroplasties, The Journal of Arthroplasty, vol. 22, no. 6, 2007. [4] Yona Kosashvili et al., Digital versus conventional templating techniques in preoperative planning for total hip arthroplasty, Canadian Journal of Surgery, vol. 52, no. 1, 2009. [5] Atsushi Kobayashi et al., Comparison of analog 2D and digital 3D preoperative templating for predicting implant size in total knee arthroplasty, Computer Aided Surgery, vol. 17, no. 2, pp. 96 101, 2012. [6] E. Sariali et al., Accuracy of the preoperative planning for cementless total hip arthroplasty. a randomised comparison between three -dimensional computerised planning and conventional templating, Orthopaedics Traumatology: Surgery Research, vol. 98, pp. 151 158, 2012. http://www. ncbi.nlm.nih.gov/pubmed/22381566. [7] A.D.A.M Medical Encyclopedia, Osteoarthritis, http://www.ncbi.nlm. nih.gov/pubmedhealth/pmh0001460/, Accessed January 23rd 2013. [8] Zimmer, Frequently asked questions, http://www.zimmer.com/ctl?op= global&action=1&id=8140&template=pc, Accessed January 23rd 2013. [9] S. H. Palmer and M. J. Cross, Total Knee Arthroplasty, http: //emedicine.medscape.com/article/1250275-overview#showall, Accessed January 23rd 2013. [10] R. G. G. Zanetti and R. Platt, Cefoxitin versus erythromycin, neomycin, and cefazolin in colorectal operations. Importance of the duration of the surgical procedure., Emerging Infectious Disease Journal, vol. 7, no. 5, pp. 828 831, 2001. 32

BIBLIOGRAPHY BIBLIOGRAPHY [11] S. et al., Risk factors for infection at the operative site after abdominal or vaginal hysterectomy, New England Journal of Medicine, vol. 307, no. 27, pp. 1661 1666, 1982. [12] C. PE and F. R, A five-year prospective study of 23,649 surgical wounds, Archives of Surgery, vol. 107, no. 2, pp. 206 210, 1973. [13] B.J.McGrory,B.F.Morrey,T.D.Cahalan,K.-N.An,andM.E.Cabanela, Effect of femoral offset on range of motion and abductor muscle strength after total hip arthroplasy, British Editorial Society of Bone and Joint Surgery, vol. 77, 1995. [14] Fred A. Mattler Jr et al., Effective Doses in Radiology and Diagnostic Nuclear Medicine, Radiology, vol. 248, no. 1, pp. 254 263, 2008. [15] David J. Brenner et al., Computed Tomography An Increasing Source of Radiation Exposure, New England Journal of Medicine, vol. 357, pp. 2277 2284, 2007. http://www.columbia.edu/~djb3/papers/nejm1. pdf. [16] M. Viceconti, R. Lattanzi, B. Antonietti, S. Paderni, R. Olmi, A. Sudanese, and A. Toni, Ct-based surgical planning software improves the accuracy of total hip replacement preoperative planning, Medical Engineering Physics, vol. 25, no. 5, pp. 371 377, 2003. [17] Christian Dick, Rainer Burgkart, Rüdiger Westermann, Distance Visualization for Interactive 3D Implant Planning, IEEE Transactions on Visualization and Computer Graphics, vol. 17, no. 12, 2011. [18] I. Otomaru, Automated Preoperative Planning of Femoral Component for Total Hip Arthoplasty (THA) from 3D CT Images, Journal of Biomechanical Science and Engineering, vol. 3, no. 4, pp. 478 489, 2008. [19] H. N, N. AM, K. M, N. N, A. R, W. W, A. SF, K. CL, R. J, T. Z, C. H, M. A, E. MP, L. DR, C. HS, F. EJ, and C. K. (eds), SEER Cancer Statistics Review, 1975-2009 (Vintage 2009 Populations), 2012. http: //seer.cancer.gov/csr/1975_2009_pops09/, Accessed May 2nd 2013. 33