Enhancing Prostate Cancer Prognosis through Digital Pathology and Machine Learning: A Systematic Review and Meta-Analysis
Chidiebere Ogbonna, F. Y. Ayankoya, Shade Kuyoro · Asian Journal of Engineering and Applied Technology · 2024
Analyzing digital pathology scans with machine learning can significantly improve prostate cancer prognosis. In recent years, machine learning (ML) algorithms have shown impressive capabilities in automating Gleason grading and prostate cancer prognostication, addressing challenges such as inter-observer variability among pathologists. This study investigates the development and validation of machine learning models specifically designed for prostate cancer prognostication. The study was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The review includes studies conducted between 2010 and 2024 that applied machine learning techniques to analyze digital pathology scans for prostate cancer prognosis. Inclusion criteria were studies focused on machine learning, prostate cancer, and digital pathology or whole-slide scans in the context of prostate cancer prognosis. Exclusion criteria included studies not involving machine learning, digital pathology, or prostate cancer, as well as studies not published in English or outside the specified time frame. Convolutional Neural Networks (CNNs) were the most commonly used machine learning approach in the reviewed studies, with brief mentions of other techniques. Meta-analysis was conducted using GraphPad Prism to create graphical representations of machine learning techniques employed in prostate cancer prognosis. The findings underscore the transformative potential of combining machine learning with digital pathology in revolutionizing prostate cancer prognosis. The integration of deep learning algorithms with digital pathology scans offers more accurate and efficient prognostication, significantly improving patient outcomes in the fight against prostate cancer.