This Small Business Innovation Research Phase I project will develop computer software that automatically analyzes medical images to diagnose brain diseases affecting the elderly (such as Alzheimer's disease) that cause difficulty with memory, thinking, judgment, and other normal functions. Such diseases affecting the elderly share similar symptoms, so they can be very difficult to diagnose and it is even more difficult to predict whether the patient will worsen. The software we will develop will use a patient's brain scan to determine which disease the patient actually has, and if the disease is Alzheimer's, the software will determine the stage of progression of the disease, and will aim to predict future decline.
The broader/commercial impacts of this research are numerous. This project will create new jobs by launching a new business that will lead to the first product of its kind. The market for the proposed product is very large, and the applicant company is well placed for commercialization, as its parent companies are already in the business of providing brain-image analysis for the pharmaceutical industry, and have already been providing preliminary diagnostic reports on patients for select clinicians on a research basis. During the project, input and feedback will be sought from a major drug company, a major imaging manufacturer, and leading physicians, to ensure that the product meets real-world needs of patient and health-care providers, and carefully considers critical issues for commercialization.
This project concerns the development of a software product for diagnosis of Alzheimer’s disease (AD) and various other forms of dementia based on the application of statistical pattern recognition (machine learning) algorithms to medical images of the brain. The ultimate objective will be a software-based service called ADMdxTM (www.admdx.com) that we will use to provide diagnostic evaluations of patients, yielding analytical reports similar in style to reports produced from blood tests or other diagnostic assays. In Phase I, our goal was to study the feasibility of producing an accurate and user-friendly diagnostic using glucose metabolism images obtained by positron emission tomography (PET) imaging. In Phase II, we will expand our diagnostic to be capable of analyzing amyloid PET images and structural magnetic resonance images (MRI), and we will develop multi-modality classifiers that can use one or more types of brain scans to form the patient’s diagnosis and prognosis for future decline. The Phase I project was highly successful: We achieved our main technical feasibility milestones and obtained our first commercial revenue based on the proposed approach. We successfully developed one new classifier that differentiates various types of dementia, and another classifier that places AD patients along a severity scale (the so-called AD "cascade"). A great deal of research and development remains, but we have successfully created an initial working version of an automated software system for FDG-PET dementia evaluation that produces excellent performance.
