Much progress has been made in the fight against breast cancer through early detection with screening mammography. Unfortunately, breast cancer can be missed on mammograms due to the difficulty of interpretation and inter-reader variability. For this reason, MedDetect, has worked to build a hybrid optical and digital processor for the identification of potentially cancerous lesions on mammograms. The processor is comprised of an optical correlator (OC) combined with a neural network (NN). It is designed to mimic the radiologist - the OC being the eyes rapidly scanning for areas of interest, and the NN being the brain making a recommendation that the case is normal or has suspicious areas. MedDetect's proposed hybrid processor takes advantage of the best of both worlds -- the best elements of optical processing and digital computing to create a complete Computer-aided Diagnosis (CAD) system. Supporting MedDetect in this effort are radiologists Drs. Brenner, Sadowsky, and Levy, University of South Florida (Dr. L. Clarke) algorithm experts, and optics experts at Lockheed Martin. If the proposed hypotheses are proven, this innovative technology will be ready for rapid transition into the clinical setting where it can assist radiologists in the early detection of breast cancer. PROPOSED COMMERCIAL APPLICATIONS: The proposed technology is responsive to a significant market. 25 million mammograms are performed each year and the number is growing each year. A similar volume of studies exists internationally. The ability to find breast cancer more consistently and perhaps earlier is of great interest to patients, radiologists, and payors. Thus, success with this research should lead to significant business opportunities.
Thesaurus Terms:artificial intelligence, computer assisted diagnosis, computer system design /evaluation, image processing, mammography, optics computer data analysis, computer processing of laboratory data human dataNATIONAL CANCER INSTITUTE
