Cancer detection using image processing techniques based on cell counting, cell area measurement and clump detection

Abstract

We proposed cancer cell detection using image processing techniques based on cancer cell counting, cell area measurement and clump detection. The proposed system autonomously detects traits of cancer in microscopy images of biopsy samples. Previous similar attempts lack flexibility in terms of features and the level of accuracy is not consistent on respective type of cancer. The system pre-processes the input image by means of gray scaling, binarization, inversion, median filtering and flood fill operation. The pre-processed image then undergoes "cell counting", "area measurement" or "clump detection" depending on the type of trait to be detected. Several sets of images were processed using this methodology and the system was fine-tuned using the feedback from these test runs. The proposed method can be effectively used for autonomous cancer cell detection, which will significantly accelerate cancer cell researches and open up new dimensions.