Biomedical Computing and Visualization Tools for Computer-integrated Diagnostic and Therapeutic Data Science - Project Summary / Abstract Computer-integrated diagnostic and interventional data science encompasses the processing, analysis, and visualization of images and signals, with the goal to better diagnose disease, better plan, deliver and monitor therapy, and advance training and simulation for medical education. The emergence of multi-modality medical imaging and its ability to “see” inside the human body has transformed disease diagnosis and enables less invasive therapies. Moreover, artificial intelligence (AI) has led to significant improvements in biomedical image computing and modeling. However, the adoption of large-scale data science techniques into diagnostic and especially interventional workflows has been hampered by the limited availability of tools that can robustly handle the size, diversity, and dimensionality of the that must be manipulated, often in real time. To mitigate these barriers, my vision has been to establish and grow a multi-disciplinary research program that integrates image computing, modeling, visualization, and validation to help elucidate biological complexity by developing intelligent imaging informatics tools to interact with biomedical data. The program’s long-term goal has been guided by the premise that effective utilization of biomedical informatics to develop versatile computing and visualization tools will lead to solutions that enable more accurate and timely disease diagnosis and less invasive therapies. Projects to date have focused on the development and validation of image computing, modeling, and visualization techniques that 1) quantify and track imaging biomarkers to diagnose and monitor disease progression, 2) identify and plan optimal therapeutic routes, and 3) guide, monitor, and deliver therapy under less invasive conditions. These tools have been developed and demonstrated in the context of cardiac, orthopedic, lung, brain, and spine applications, in collaborations with clinicians and industry partners. The work heretofore builds on a decade of research in computer-assisted diagnosis and therapy and will help answer the key questions in biomedical computing and visualization. We will research novel AI-based image computing algorithms that robustly handle label noise, out of distribution detection, and class imbalance while yielding high performance on limited training data; build and validate hybrid AI/physics-based computational models for interactive biomedical simulation; research uncertainty modeling, propagation, and visualization techniques; objectively validate the developed tools; and evaluate different display paradigms for information presentation. This research will yield innovative imaging informatics tools that have the potential to advance computer-integrated diagnosis and interventional data science by catering to a broad range of diseases, organ systems, and minimally invasive therapy applications. We will make the developed techniques available to the biomedical research community and to clinician scientists to promote their clinical translation to impact a larger patient population. Lastly, this research program will continue to foster opportunities to train and mentor skilled workforce for biomedical research from diverse backgrounds and education streams. 1
