A biological tissue sample is scanned to produce an image and corresponding optical-density data. A computerized algorithm is used to identify, segregate, and produce images of nuclei contained in the image. The OD values corresponding to nuclear chromatin are used to identify numerical patterns known to have statistical significance in relation to the health condition of the biological tissue. These patterns are analyzed through discriminant analysis and a non-supervised learning algorithm to predict changes that suggest a risk for the recurrence of a cancer event, such as a malignant lesion.