Scale Invariant Deep Neural Multiple Feature Learning Based Boosted Support Vector Entropy Classification for Breast Cancer Diagnosis using Mammograms
Sai Krishna K, Grace Kanmani Prince P · International Journal of Electrical and Electronics Engineering · 2023
An earlier diagnosis of cancer rises the living days of patients. The evolution of medical systems supports to identify the presence of cancer. In the state-of-the-art works, few methods were implemented for diagnosing breast cancer with different Deep Neural Network (DNN) support. However, the traditional DNN techniques were computationally expensive as it does provide better feature extraction performance with minimal time usage. Besides that, the misclassification error observed during the diagnosis process was higher, impacting the accuracy of early breast cancer prediction. Therefore, a novel Scale Invariant Deep Feature Learning Based Boosted Support Vector Entropy Classification (SIDFL-BSVEC) method is introduced. The SIDFL-BSVEC method is proposed to get better accuracy for the early identification of breast cancer with lesser time. The SIDFL-BSVEC method initially designs Scale Invariant Deep Neural Multiple Feature Extraction (SIDNMFE) algorithms to discover critical features in input mammograms with minimal time by working as robust to illumination variations, noise, partial occlusion, and minor viewpoint changes in mammograms images. These traits are significant for early cancer disease prediction when cells in the mammogram image have dissimilar sizes and orientations. After that, Support Vector Entropy Boosted Cancer Classifier (SVEBCC) algorithm is designed to minimize the error rate determined during the early analysis of breast cancer. The performance of the proposed CWDCMFE-MBRC technique is verified by taking the parameters such as disease diagnosis accuracy, disease diagnosis time, sensitivity, specificity and error rate along with various numbers of input images.