Multidimensional MRI-based Big Data Analytics to Study Osteoarthritis

点击翻译按钮获取中文摘要

ABSTRACT This project outlines technical medical image processing and machine learning developments to study the pathogenesis and natural history of osteoarthritis (OA). In the past few years, the availability of public datasets that collect data such as plain radiographs, MRI genomics and patients reported outcomes has allowed the study of disease etiology, potential treatment pathways and predictors of long-range outcomes, showing an increasingly important role of the MRI. Moreover, recent advances in quantitative MRI and medical image processing allow for the extraction of extraordinarily rich arrays of heterogeneous information on the musculoskeletal system, including cartilage and bone morphology, bone shape features, biomechanics, and cartilage biochemical composition. Osteoarthritis, being a polygenic and multifactorial disease characterized by several phenotypes, seems the perfect candidate for multidimensional analysis and precision medicine. However, accomplish this ambitious task, will require complex analytics and multifactorial data-integration from diverse assessments spanning morphological, biochemical, and biomechanical features. In this project, we propose to fill this gap developing automatic post-processing algorithms to examine cartilage biochemical compositional and morphological features and to apply new multidimensional machine learning to study OA This “Pathway to Independence” award application includes a mentored career development plan to transition the candidate, Dr. Valentina Pedoia, into an independent investigator position, as well as an accompanying research plan describing the proposed technical developments for the application of big data analytics to the study of OA. The primary mentor, Dr. Sharmila Majumdar, is a leading expert in the field of quantitative MRI for the study of OA, and the co-mentors, Dr. Adam Ferguson and Dr. Ramakrishna Akella, have extensive experience in the application of machine learning and topological data analysis to big data. The diversified plan of training and the complementary background of these mentors will allow the candidate to develop a unique interdisciplinary profile in the field of musculoskeletal imaging. The candidate, Dr. Valentina Pedoia, is currently in a post-doctoral level position (Associated Specialist) at the University of California at San Francisco (UCSF), developing MR image post-processing algorithms. The mentoring and career development plan will supplement her image processing background with valuable exposure to machine learning, big data analysis, epidemiological study design, and interdisciplinary collaboration to facilitate her transition to a medical imaging and data scientist independent investigator position. Ultimately, she aims to become a faculty member in a radiology or bioengineering institute, where she can further research technical biomedical imaging and machine learning developments applied to the musculoskeletal system.