2- and 3-D differentiation of cancerous from benign prostate tissue for diagnosis, monitoring, and planning based on spectrum analysis

Spectrum analysis of echo signals from the prostate continues to show promise for differentiating cancerous from non-cancerous tissue. Data acquired from 44 patients, including 9 with biopsy-proven cancer, shows this method to be superior to conventional B-mode methods for identifying cancer. Based on 227 physically independent scans, the ROC-curve area for spectrum analysis was 0.77/spl plusmn/0.09; the area for B-mode interpretation was 0.65/spl plusmn/0.06. Tissue-type "stained" images based on spectral parameters may improve biopsy guidance, reduce true- and false-negative biopsies, and improve the efficacy of detection by over 50%. 3-D images based on such staining show promise for volumetric evaluations of cancer, which may improve clinical staging, treatment planning and treatment monitoring.