A machine vision system for diagnostic histopathology offers five modules: 1) for the automated detection of regions of abnormality in histopathologic sections; 2) for fully automated image segmentation and diagnostic information extraction by a knowledge-guided procedure; 3) for the derivation of histometric indices, such as a progression index or grade for a lesion; 4) for diagnostic evidence evaluation by Bayesian inference networks; and 5) for individual patient targeted prognosis based on a case-based reasoning process. The system has been in operation for several years. Correct segmentation for even complex scenes such as cribriform glands has been achieved with a high success rate for histopathologic sections from prostate, colon, and breast. The lesion search module and prognostic module have passed feasibility testing and are still undergoing development.
No takes yet. Share an insight, caveat, or question.
Bartels et al. (1997) studied this question.