Classification of Breast Microscopic Imaging using Hybrid CLAHE-CNN Deep Architecture
Ankit Vidyarthi, Jatin Shad, Shubham Sharma, Paridhi Agarwal · 2019
Breast cancer is the most prevalent form of cancer and can occur in both men and women, although it is most common among women. The problem with manual pathology examination of breast cancer is that it is time-consuming as it requires scanning through images of tissue under various distinct magnification levels to obtain accurate diagnoses. The advancement in computer assisted diagnosis system help to improve the early diagnosis process using various machine learning algorithms. This paper proposes the hybrid architecture of CLAHE and deep convolutional network for the classification of the breast microscopic imaging. The proposed architecture is experimented with BreakHis dataset that comprises of 7909 images in total having 2480 images of benign class and rest is of malignant class. Proposed method is experimented with two level of test i.e., with and without pre-processing of images with CNN architecture. Initially, proposed method uses the traditional CNN architecture to classify the images into benign and malignant. Later, the experiment is performed with hybrid architecture of Contrast Limited Adaptive Histogram Equalization (CLAHE) followed by the watershed algorithm, as a pre-processing for image enhancement followed by CNN architecture for image classification. The experimental result suggest that the proposed hybrid architecture outperforms traditional methods in terms of accuracy gained around 3%.