Histological Grading of Breast Cancer Malignancy Using Automated Image Analysis and Subsequent Machine Learning
Aim: The objective of this study was to determine the histological degree of breast cancer malignancy using the automated principle of machine learning with the free access computer programs CellProfiler and Tanagra. Methods and results: Digital photographs of neoplastic tissue histological slides were obtained from 224 women with breast cancer. The digitized images were transferred to the CellProfiler software and treated according to a predetermined algorithm, resulting in a database exported to the Tanagra software for further automated classification of the histological degree of malignancy. The Kappa index of agreement between the medical pathologist and the automated analysis performed in the Tanagra software was 0.91 for the tubular score, 0.55 for the nuclear score, and 0.49 for the mitotic index score.
