Grading of Glandular Tissue Cancer by Detailed Image Analysis of Stained Tissue Slices

摘要

An improved histopathological score is obtained by identifying objects in images of glandular tissue from cancer patients. The objects are identified based on staining by a biomarker. The score predicts that a cancer patient will have a recurrence of cancer of the glandular tissue based on a geometric characteristic of individual identified objects but not on any pattern formed by the identified objects. First objects are generated from the image of glandular tissue which has been stained with a single biomarker that stains epithelial cells. Second objects are then generated using the first objects. A geometric feature of each of the second objects is measured. A shape index is then calculated for each of the second objects based on the geometric feature, and an average shape index is calculated. Based on the average shape index, a score is determined that indicates a level of cancer malignancy of the glandular tissue.

主权项

1. A method comprising:
identifying objects in a digital image of glandular tissue from a cancer patient, wherein the objects are identified based on staining by a biomarker of the glandular tissue; and predicting that the cancer patient will have a recurrence of cancer of the glandular tissue based on a geometric characteristic of individual identified objects but not on any pattern formed by a plurality of the identified objects.