Contents I. Introduction
One symptom of cancer is an uncontrollable increase in the number of cells in a particular bodily part. It has been posing a threat to people for many years. Cancerous cells have the ability to infiltrate and destroy the body's healthy cells. It often starts in one area of the body and can spread throughout the entire body through a process termed metastasis. [1]. Breast cancer develops in breast tissue. It happens at the time of mutation of breast cells and proliferation uncontrollably, which results in a mass of tissue called tumor. Breast cancer, like other cancers, can invade and grow into the tissue surrounding the breast. It can also spread to other parts of women body and cause new tumours to form. The most frequent kind of cancer in women is breast cancer (BC). According to the World Health Organization (WHO), 2.1 million women have breast cancer that poses a serious threat to their lives [2]. During the year 2020, there have been expected to be 1,392,179 cancer patients in India, with the mouth, the cervix uteri, the breast, the lung, and the tongue being the common 5 top sites. With breast cancer, 56.0% of patients had received their diagnosis at a locally advanced stage [3]. The American Cancer Society recently released a report [4] that estimates that 31% of new cancer diagnoses in women are due to breast cancer alone. It has an extremely high fatality rate and is the most prevalent cancer in women worldwide. [5] Over the past 26 years, the age-standardized incidence rate of BC in females has increased across the board, increasing 39.1% (95% confidence interval, 5.1 to 85.5). [6]. According to reports, there is a 13% chance that a woman in the US may have breast cancer at some time in her life. [7]. In the United States, 43,600 deaths from breast cancer were anticipated in 2021. Males and females alike can develop breast cancer. It is anticipated that 2,650 new cases of invasive breast cancer in men would be discovered in 2021. Around 1 in 833 men may develop breast cancer in their lifetime [8]. The diagnosis process is highly laborious and susceptible to observer variations due to the accumulation of digital pathology images, which are frequently in the form of whole slide images (WSIs). On the other hand, there is growing interest in using machine learning (ML) approaches to automate this process and reduce human error while improving accuracy. To aid in early detection for computer-aided diagnosis systems (CADs), numerous ML-based techniques [9], have been presented and shown to be reliable and successful [10]. Most Machine Learning-based methods for classifying breast cancer involve the three steps of feature extraction, pre-processing, and classification. The feature extractor used can be improved with specialized human expertise. The reliability of Machine Learning-based strategies for the categorization of breast cancer is somewhat constrained due to the limitations of manual features and the sensitivity of feature extractors. Convolution neural networks (CNNs), in particular, have recently been successfully used in this sector. Deep learning (DL) techniques can be divided into two types, namely region of interest (ROI)-level methods and image-level approaches, based on the input of the network. Despite deep learning's obvious advantages and the apparent advancements in breast cancer categorization stated therein, there are still few practical uses for it. Most publicly available datasets of breast pathology images for breast cancer classification are of a modest scale, making under-constraint deep learning models likely overfit due to the difficulty of manually labelling pathological images.
