Meddetect was formed as a start-up firm by Rose Health Enterprises (Denver, CO) and Lockheed Martin (Denver, CO) to use BMDO-funded technology to analyze medical images. M e d D e t e c t âs optical system is projected to quickly identify 75 percent of screening mammograms that are negative, allowing doctors more time to examine potential cancer cases closely. In addition, the technology makes screening a more powerful and accurate tool by automatically identifying abnormal image attributes. In preliminary tests on an archive of mammographic images, it has already detected a cancerous breast lesion, that, using conventional mammography, did not appear for another year. MedDetectâs initial work in the medical arena has been focused on integrating optical processors with complex algorithms to improve mammographic images. The company expects to use these methods in other medical applications as well, such as to impro v e cancer detection processes in chest x-rays and Pap smears. It is using much of the technology, including target scene generation software and optical components for rapid data processing, that Lockheed Martin developed with BMDO funding for advanced target acquisition and recognition . MedDetectâs optical system will be compatible with âfilmlessâ digital mammography, which several companies and re s e a rch groups are developing. Using MedDetectâs technology, digital x-ray images can be analyzed in less than a minute. The images can then be transmitted
to another radiologist for a second opinion. In addition, digitally storing images at a central location allows physicians to quickly access re c o rds for baseline image analysis and comparison. Both Rose and Lockheed Martin are providing seed money for MedDetect, with plans to raise additional private capital and have a prototype available within 18 to 24 months. Technology impro ements achieved by MedDetect will also be re t u rned to Lockheed Mart i n âs defense technologies. MedDetect âs system is a hybrid of optical and digital processing. An optical correlator uses lenses and a low-power laser to examine the mammogram. The optical correlator, with programmable spatial light modulators and Fourier transform lens pairs, uses photons instead of electrons to perfom the calculations to detect an abnormal feature. This information is then transmitted to a computer
that uses neural network software to âlearnâ the specific attributes of breast abnormalities. The leaned information is stored and ap