Computerized image analysis of nuclear shape as a prognostic factor for prostatic cancer.

Computer assisted image analysis was used to determine quantitative and reproducible nuclear morphology of 17 totally excised pathological specimens from stage B-2 prostatic cancer in an attempt to correlate histopathological analysis with clinical prognosis for the individual patient. We describe a nuclear roundness factor that correlates with the clinical outcome of the individual patient. This nuclear shape factor appears to identify prostatic tumors with a high metastatic potential from tumors that are less aggressive. Quantitative nuclear image analysis, its accuracy, and reproducibility are described and discussed. This technique may provide the pathologist with a valuable tool for analyzing prostatic cancer cells in a quantitative manner, thereby contributing precise information to the urologist relevant to the prognosis of the individual patient.