Early Detection of Breast Cancer on Mammograms Using: Perceptual Feedback, Computer Processed Images and Ultrasound.
Abstract
Several approaches for improving the detection of small tumors in the breast were explored. These included: (1) Perceptual feedback to decrease errors in missing tumors that are actually visible on the initial screening mammogram. The observers head-eye-position was recorded while viewing mammograms. It was found that false negative decisions in identifying a lesion was associated with prolong dwell time. Visual feedback, obtained by placing a small circle on the image surrounding the suspicious region (as predicted by the gaze duration), enhanced the detectability of masses. (2) Computer processing of screening mammograms for detection of clusters of microcalcifications and parenchyma patterns associated with developing lesions. Preserving the high spatial resolution and wide latitude in the digitized mammogram was found important for computer aided detection of subtle microcalcifications. Elevated mammography density has been associated with an increased risk of developing breast cancer. An algorithm was developed to perform quantification of the elevated parenchyma density on digitized mammograms. (3) High resolution ultrasound. The ultrasound distortion in the female breast image was identified and compensating algorithm developed to reduce the distortions. Design consideration was performed for two dimensional large acoustic transducers for high resolution breast imaging. The analysis indicates that using a 2D array would significantly increase the resolution of ultrasound mammography.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1996
- Accession Number
- ADA307820
Entities
People
- Peter Bloch
Organizations
- University of Pennsylvania