A Hybridized Early Detection, Classification and Diagnosis of Breast Cancer Using Deep Learning Algorithm

Abstract

There is a rise in the cases of breast cancer in low income population like northern Nigeria. Breast
cancer is considered as one the major killer diseases among women of child bearing age (WCBA)
(Aslan et al., 2018). There have been many researches on the identification and diagnosis of breast
cancer disease for decades however, some of these researches are manual based and are inefficient due
to their time consumption. With the recent advancement in ICT, machine learning algorithms are used
to classify images (Kaji & Kida, 2019). A great achievement was made in classifying and detecting
breast cancer using traditional machine learning algorithms such as Decision Tree and Artificial Neural
Network (ANN) (Higa, 2018).
Despite the performance of these traditional machine learning algorithms in cancer prediction and
diagnosis, there are common limitations that need to be addressed. These limitations include manual
feature selection, fewer number of classes in classifying tumour (usually being classified into two
classes) and their inability to classify larger dataset on time (Yari & Nguyen, 2020). This research is
aimed at improving the performance of traditional machine learning algorithms by using a Deep
Learning algorithms. Deep learning algorithm gives a promising result in image classification and can
therefore extend the number of classes to more than the usual two (2) and solve the problem of fewer
classes.