Utilization of Deep Learning and Machine Learning Algorithms in Detection, Segmentation, and Classification of Brain Metastasis: A Systematic Literature Review

One-tenth to around a quarter of cancer patients lose their lives to Brain Metastasis (BM) globally. Accurate neuroimaging of BM is essential to encourage early detection, decide treatment options, and increase survival rates. This systematic literature review (SLR) examines the utilization of machine/deep learning algorithms and imaging modalities in the detection, segmentation, and classification of BM. The study adheres to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis Statement) and meticulously evaluates 32 relevant articles spanning from 2018 to 2023. The investigation explores scholarly databases including Google Scholar, SCOPUS, Web of Science, PubMed, SCISPACE, IEEE, Science Direct, Hindawi, and EBSCOhost to pull out relevant studies. The four aspects viz. employment of various imaging modalities, utilization of machine/deep learning algorithms, evaluation of performance metrics, and investigation of software tools for the identification of BM considered in this study. The review identifies Convolutional Neural Networks (CNN) as the most frequently utilized algorithm with the highest AUC-ROC (0.98) and demonstrates its efficacy for BM detection. Furthermore, out of 32, 15 articles highlight the significance of contrast-enhanced T1weighted images (T1CE/ T1c). This SLR unearths the transformative impact of machine/deep learning algorithms and emphasizes the importance of imaging modalities for enhancing the accuracy of diagnostics.