ICMFKC with optimize XGBoost classification for breast cancer image screening and detection

Nowadays most vicious disease is cancer, the cure of which must be the main argument through scientific investigation. The prior detection of cancer could assist in curing the disease completely. A cancerous tumor in the breast comprises of a mass cancer cells which develop in an abnormal, uncontrolled way. Breast mammogram images can be enhanced using digital image processing tools to assist physicians in detecting breast tumors at the initial stage. A lot of researchers have worked on early detection and classification of mammogram images. The main aim of this research work is, to develop a highly accurate automatic approach to analyze and segment the breast tumor and classify it into benign or malignant images. Here, in this research article, the experimental analysis work mammography images are taken from the both Public Digital Database of Screening Mammography (DDSM) and Mammographic Image Analysis Society (MIAS), the in-house clinical dataset from Metro scans and laboratories. The proposed work's first stage is to remove noise from the input image and boost the contrast of the image's anomalous region by using Anisotropic Diffusion with Median Filter (ADWMF). In the second phase, the denoise image was segmented, and Identifying the accurate breast tumor position using Improved Centroid-based Macqueen's K-Means Clustering (ICMFKC) method was adopted. From the segmented ROI image, the GLCM features are extracted in the third phase. Finally, images of benign and malignant breast cancer are classified. The classification is carried out depending on the extracted features from the ROIs using the Light Gradient Boosting Machine (LGBM) Classifier and Optimize Extreme Gradient Boosting (OXGBoost) Classifier. Using the LGBM classifier an accuracy of 85.4% is obtained; with the OXGBoost classifier, an accuracy of 96.6% is obtained. Hence, the proposed work is most helpful for radiologists in the diagnosis of breast cancer.