Neural networks applied to the detection and diagnosis of Breast Cancer, a systematic review of the scientific literature of the last 5 years

Abstract

One of the fatal diseases that occurs in women is breast cancer and is associated with late diagnosis and poor access to medical care according to the patient's needs, therefore neural networks play a relevant role in detection of breast cancer and aims to be a support to guarantee its accuracy and reliability in cancer results. Therefore, the aim of the present systematic review is to learn how neural networks help to improve accuracy in breast cancer diagnosis through image recognition. For this, the formula generated with the PICO methodology was used; Likewise, the first result was 203 investigations related to the topic and based on the established inclusion and exclusion criteria, 20 final free access scientific articles were selected from the Scopus database. In relation to the results, it was found that the use of neural networks in the diagnosis of breast cancer, especially convolutional neural networks (CNN), has proven to be a promising tool to improve the accuracy and early detection of the disease, reaching achieve an accuracy of 98 % in the recognition of clinical images, which means a big difference compared to traditional methods. On the other hand, although there are challenges such as the limited availability of high-quality data sets and bias in training data, it is suggested to investigate the development of methods that integrate multiple sources of information and the use of deep learning techniques.