Development of mammogram computer-aided diagnosis systems using optical processing technology

Abstract

Mammogram computer-aided diagnosis (CAD) systems based on digital image processing algorithms are under development. These systems act as an aid to the radiologist, providing image cues of the locations of likely cancer lesions, and they have been shown to increase lesion detection performance. The use of state-of-the-art hybrid optical and digital processing technologies affords a reduction in processing time, and enables further improvements in sensitivity and specificity. Many CAD algorithm components can be mapped into linear filtering operations and implemented in a Fourier transform coherent optical processor. With the near-term promise of performing 512/spl times/512-sample convolutions at 1,000 frames-per-second rates, optical processors will greatly increase the CAD system speed. A test-bed hybrid optical/electronic processor is under development following the successful completion of studies that verified the feasibility of mapping algorithms into optically-implemented filtering operations.