This Small Business Innovation Research (SBIR) Phase II project proposes to develop robust and effective imaging techniques for assessment of atherosclerotic disease severity for prognostic and longitudinal use. In the United States alone, approximately 5 million patients suffer pre-stroke symptoms of which 795,000 go on to a stroke annually. About 610,000 of these are first or new strokes, while the remainder are recurrent strokes. Despite these statistics, there is no effective test to tell who will or will not suffer acute events or to measure whether medical therapies are effective at reducing the risk. In this project, multivariate quantitative descriptors are developed using data from controlled outcome studies on a specialized model to discover and validate prognostic signatures that, in composite, perform well in both cross-sectional prognostic and longitudinal applications with high predictive value. Phase I results are extended to support histologically verifiable tissue types, expanding the functional and performance attributes of the product with a tie to localized ground truth maps made possible with co-registration of histology with MRI. The plan is to iteratively validate the developed capability under commercially accepted design controls.
The broader impact/commercial potential of this project will be the development of effective means for computer-aided prognostics using quantitative imaging phenotypes. Physicians face a complex and heterogeneous series of clinical manifestations of disease. Because disease arises through a complex interaction of multiple molecular signals and pathways often confounding the eventual effect, tools and approaches are needed to identify key pathways that reflect the underlying pathological processes. Functional imaging modalities have recently emerged for characterization of these disease processes and to obtain a better mechanistic understanding of the underlying biologic processes to distinguish more aggressive from less aggressive disease phenotypes. Computer-aided prognosis (CAP) of disease is a new and exciting complement to the field of computer-aided diagnosis (CAD). Since CAP approaches distinguish between different subtypes of a particular disease (as opposed to CAD schemes trying to distinguish diseased from benign processes), there is a need for more sophisticated image analysis, computer vision, and machine learning methods to identify subtle disease signatures that can separate unstable from stable disease. The chosen application in this project is a use case that has the potential to radically increase the power of applications to support clinicians in pursuit of personalized medicine.
