Development of imaging biomarkers for knee osteoarthritis; Radiographs, DXA and MRI Active Shape and Appearance Models
Lead Research Organisation:
University of Aberdeen
Department Name: Division of Applied Medicine
Abstract
This project aims to find better ways to look at osteoarthritis by looking at the shape and appearance of the knee joint.
The shape of the joint tells us a lot about its health; Osteoarthritis makes joints flattened and deformed, which causes pain when people put weight on them, but certain shapes may reflect early signs of osteoarthritis, or be a sign that someone is at risk of getting osteoarthritis in the future. However it is difficult to fully describe the shape of complicated structures, such as a knee joint using a few lengths, widths and angles. To build a computer model of the shape and appearance of the knee, we will use image analysis techniques used in other applications such as face recognition.
Currently, the only effective treatment for osteoarthritis is joint replacement. There are no effective drugs available, other than for pain relief. Unfortunately, because osteoarthritis is a complicated disease that is hard to identify in its early stages and often progresses quite slowly from year to year, running drug trials is often considered too expensive and time consuming by pharmaceutical companies. We hope to find that the shape and appearance of the knee can be used to measure smaller changes in disease severity than currently possible and whether it could be used to identify a ?high-risk? group of patients, so that drug trials would become more economical to run.
We will compare two and three dimensional models of the knee and use different types of medical image to build the model, including MRI scans and x-rays. Each method has different advantages and disadvantages so by comparing all three we hope to find the best way to model the osteoarthritic knee.
The shape of the joint tells us a lot about its health; Osteoarthritis makes joints flattened and deformed, which causes pain when people put weight on them, but certain shapes may reflect early signs of osteoarthritis, or be a sign that someone is at risk of getting osteoarthritis in the future. However it is difficult to fully describe the shape of complicated structures, such as a knee joint using a few lengths, widths and angles. To build a computer model of the shape and appearance of the knee, we will use image analysis techniques used in other applications such as face recognition.
Currently, the only effective treatment for osteoarthritis is joint replacement. There are no effective drugs available, other than for pain relief. Unfortunately, because osteoarthritis is a complicated disease that is hard to identify in its early stages and often progresses quite slowly from year to year, running drug trials is often considered too expensive and time consuming by pharmaceutical companies. We hope to find that the shape and appearance of the knee can be used to measure smaller changes in disease severity than currently possible and whether it could be used to identify a ?high-risk? group of patients, so that drug trials would become more economical to run.
We will compare two and three dimensional models of the knee and use different types of medical image to build the model, including MRI scans and x-rays. Each method has different advantages and disadvantages so by comparing all three we hope to find the best way to model the osteoarthritic knee.
Technical Summary
The aim of this project is to develop, test and compare imaging biomarkers for knee osteoarthritis. Osteoarthritis (OA) is a common, painful and debilitating disease, yet it is still poorly understood. OA of weight bearing joints is particularly problematic as it restricts mobility especially knee OA which is present in approximately 30% of people over 75,. There are no disease modifying drugs (DMOADS) available and the only proven treatment is joint replacement. My hypothesis is that the shape and appearance of the knee in Radiographs, DXA and MRI scans can be used as imaging biomarkers, to reduce the size and duration of clinical trials of DMOADS and improve our understanding of the disease process.
The shape of a joint is an important factor in the progression and incidence of OA. However it is difficult to fully describe the shape of the knee joint using basic geometry. To investigate the knee, I will develop Active Shape and Active Appearance Models. Three longitudinal cohorts will be assessed, a subset of the Melbourne Collaborative Cohort Study, the Osteoarthritis Initiative (OAI) and a recent Aberdeen study, comprising matched radiographs, DXA and MRI images. This cohort will be extended longitudinally during this project with annual DXA images.
The first step is to validate the pilot study results by applying the existing model to baseline radiographs from OAI and comparing measures of disease severity (Kellgren Lawrence grade and Joint Space). Comparison of the different datasets will also allow me to develop and optimise the model, maximising the information content, whilst minimising noise and variation not associated with OA. Tests for differences in the knee model between subgroups (incidence vs. progression or presence, absence and development of bone marrow lesions) will be performed and links between with anthropometric data will also be explored, including measures of knee alignment from full limb radiographs in the OAI. Using MRI scans, comparisons between two and three dimensional models will be performed in collaboration with iMorphics Ltd. Finally changes over time will be explored using the OAI and Aberdeen data to test how long is required before significant differences in knee shape can be observed and whether the model can identify those at greatest risk of becoming ?fast-progressors?.
The study will result in a valuable tool to help identify those at greatest risk of OA, with the potential to aid DMOAD development in the long term.
The shape of a joint is an important factor in the progression and incidence of OA. However it is difficult to fully describe the shape of the knee joint using basic geometry. To investigate the knee, I will develop Active Shape and Active Appearance Models. Three longitudinal cohorts will be assessed, a subset of the Melbourne Collaborative Cohort Study, the Osteoarthritis Initiative (OAI) and a recent Aberdeen study, comprising matched radiographs, DXA and MRI images. This cohort will be extended longitudinally during this project with annual DXA images.
The first step is to validate the pilot study results by applying the existing model to baseline radiographs from OAI and comparing measures of disease severity (Kellgren Lawrence grade and Joint Space). Comparison of the different datasets will also allow me to develop and optimise the model, maximising the information content, whilst minimising noise and variation not associated with OA. Tests for differences in the knee model between subgroups (incidence vs. progression or presence, absence and development of bone marrow lesions) will be performed and links between with anthropometric data will also be explored, including measures of knee alignment from full limb radiographs in the OAI. Using MRI scans, comparisons between two and three dimensional models will be performed in collaboration with iMorphics Ltd. Finally changes over time will be explored using the OAI and Aberdeen data to test how long is required before significant differences in knee shape can be observed and whether the model can identify those at greatest risk of becoming ?fast-progressors?.
The study will result in a valuable tool to help identify those at greatest risk of OA, with the potential to aid DMOAD development in the long term.
