A deep learning algorithm to diagnose breast cancer among female

Abdul Rahman, Siti Suhailah (2020) A deep learning algorithm to diagnose breast cancer among female. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (Full Text) A deep learning algorithm to diagnose breast cancer among female.pdf - Submitted Version Download (1MB)

Abstract

Breast cancer is an infection in which cancer is found and develops in the tissue of the breast. Presently, approximately 1 in 30 females in Malaysia has started to have breast cancer in their lifetime. Hence, the early detection of breast cancer by the radiologist can stop these health issues from becoming chronic and may facilitate the patient to treat adequately based on the histopathology image (image of tissues and cells). Breast cancer categorized into two types, which are malignant and benign. A mammogram is one of the essential criteria that could detect abnormality if exist areas of white, high-density tissue depending on the size, shape, and edges of the breast cancer. However, the doctors have a difficult time detecting breast cancer due to the complex structure of mammogram images, and the ability of a digital screening mammogram is limited in the extremely dense breast with 60-70% detection accuracy. With the advent of the Computer-Aided Diagnosis (CAD) system, recent researches have shown that the deep learning techniques applied to classify various cancers based on histopathology images. Besides, the growth of the Convolutional Neural Network (CNN) is helpful in systems for detecting relatively whether the tumor is malignant or benign. The classification result may guide the doctors to develop a more realistic treatment plan that provides an early detection result and intervention besides can aid the patients in obtaining an accurate diagnosis. By using deep learning techniques, it expects to create an algorithm that can classify between benign and malignant factual. This project has proven that CNN architecture has successful obtained higher accuracy which for VGG-19 has accuracy of 91% while for VGG-16 has the accuracy of 94.34%.

Actions (login required)

View Item