The objective of the proposed research is to relate ultrasound attenuation and backscatter coefficients, and their frequency dependencies,and their sound speed to the pathological state of breast tissue. These data are required for adequate assessment of the potential of ultrasound transmission imaging and ultrasound computed tomography when combined with ultrasound reflection imaging in breast cancer detection. Methodology: Freshly excised biopsy specimens will be interrogated by finely focused ultrasound beams. Translation of the sample holder under computer control permits interrogation at selected locations in the specimen. Data will be obtained over the frequency range 3 to 9 MHz and stored for subsequent analysis. Time-of-flight data will yield ultrasound speed at each location; insertion loss data will yield attenuation coefficients. Transmitted ultrasound power will be detected by a phase-insensitive acousto-electric receiver, thereby avoiding phase-cancellation artifacts; backscattered ultrasound will be detected by a focused piezoelectric transmitter-receiver. Improvements to the apparatus will be undertaken to increase speed of data acquisition, maximize spatial resolution, and increase sensitivity of signal detection. Detailed histological examination and coding will permit analysis to correlate ultrasonic data with pathological state. In particular, we seek to discriminate benign from malignant states. Discriminant analysis using the jackknifing technique, classification and regression trees, and neural net analyses will be applied.