Abstract
Introduction: Breast Cancer (BC) is a worldwide prevalent carcinoma with varying growth rates. It is also a heterogeneous condition with several clinical manifestations. There is a guideline for effective screening of BC, but the mammograms that are obtained for the same often return false negatives or may not effectively diagnose BC, especially at the early stages. The mortality rates due to BC is also significant and efficient screening may reduce more than 30% of mortality. The assessment and accurate determination of tumour growth is difficult due to the substantial failures of screening BC cases.
Aims and Objectives: This current study intends to introduce methods of screening BC using an Artificial Intelligence capable machine learning system, mainly for accurate and efficient detection of BC and assessing its prognostic outcomes.
Materials and Methods: The study comprises a cohort of 120 BC patients who visited Veronica Gynae Clinic aged between 50 and 65 years old. The study involves the development of serial mammograms for detection of in-vivo growth rate and further assessment of immunohistochemical staining. The study also facilitates the development of a machine learning model for detecting BC cases from serial mammograms and several other factors. Statistical analyses have been conducted for effective significance tests between the newly developed and existing methods.
Results: The newly developed method proved to be successful in accurately assessing several aspects of the tumour, including size, stage, grade, mitotic score, type of histology, Nottinham Prognostic Index (NPI) and many more. The newly developed method is more significant than the existing one in assessing clinico-pathological characteristics. The newly developed method has shown significantly better prognostic ability than the existing one (p = 0.0284). Kaplan-Meier Survival Curve has been plotted to analyze the growth of the tumour with its prognosis.
Conclusion: The study has effectively presented the development of a prognostic model for Breast Cancer patients with the application of machine learning algorithms, which can be utilized in the screening of breast cancers and proper management of the same to improve its prognosis.