Prediction of Breast Cancer Risk Using Microarrays and Deep Learning

Abstract

More than 1.15 million new instances of breast cancer are identified each year. In the clinic, only a few reliable prognostic and predictive indicators are utilized to make decisions about the treatment of breast cancer patients. The mortality rate of breast cancer patients may be lowered and their survival time extended by early identification. Analysis and processing of Microarray images, the principal test used for screening and early diagnosis, are the keys to improving breast cancer prognosis and are at the heart of this study. The Fuzzy C-means (FCM) approach is used for image segmentation in microarray for the detection of breast cancer. After features are retrieved from the segmented areas and the system is fully trained, the efficient classifier is used to assign microarrays to their respective classes. Techniques such as Multi-level Discrete Wavelet Transform, Principal Component Analysis (PCA), and Gray-level Co-occurrence Matrix (GLCM) are used to extract texture information. Differentiating masses and microarray image calcifications from the surrounding tissue is achieved with the aid of morphological operators and the classification of these features is handled by the Deep Convolution Neural Network (DCNN) algorithm. In a microarray, the tumor’s borders are highlighted and exhibited to the doctor, who may then assess the extent of the growth.