Advancements in Breast Cancer Detection: Harnessing Artificial Neural Networks for Improved Accuracy

Abstract

Female breast malignancy is the exceedingly prevalent reason for the demise of women around the world. Women who are revealed to have breast cancer earlier in life get a lower death rate from the disease and increase the life expectancy of patients. Mammography screening is one of the effortless, efficient, and affordable ways to identify breast cancer in advance. The early investigators pioneered many methods based on statistical measurements and textural traits for the earliest identification of carcinoma of the breast. Due to artefacts, noise, pectoral muscles, and irregular illumination, the accuracy of cancer prediction in these works is relatively low.   The accuracy of predictions made by employing textural characteristics for forecasting breast cancer in earlier work is 83.33%. The research proposal processes of mammograms to remove noise, artefacts, pectoralis, and inconsistent illumination in an endeavor to increase forecast accuracy. The proposed research uses an Artificial Neural Network (ANN) to classify breast masses as benign or malignant based on geometric pattern features. Its prediction accuracy is 86.67%, which is superior to research studies based on textural and statistical characteristics of breast mammograms.