Detection and histo-scoring of HER2/CEN17 biomarkers in SISH images

Abstract

This article presents a method for assessing HER2/CEN17 status from genetic biomarkers using SISH (Silver-enhanced in situ hybridization) images, that are helpful in the diagnosis procedure of breast cancer. SISH staining does not require the use of expensive or advanced microscopes, it can be performed using ordinary bright field microscopes. A combination of image analysis and traditional machine learning approach is used for detection of the biomarkers. The method used Gabor filter bank and K-means clustering for the detection of red, and blue signals. The proposed method's quantitative and visual results show that it is effective in diagnosing breast cancer. When evaluating the HER2 status of primary breast cancer, we conclude that SISH and bright field microscopy are excellent alternatives to FISH and CISH.