Raman Spectroscopy and Machine Learning in the Diagnosis of Breast Cancer.

Abstract

The most prevalent cancer in women worldwide, breast cancer, greatly benefits from early identification for better prognoses. But traditional diagnostic techniques, like biopsies and mammograms, can require invasive procedures and lack accuracy. The non-invasive, quick, and accurate nature of machine learning (ML) and Raman spectroscopy (RS) in breast cancer diagnoses are examined in this review. Combining machine learning's capacity to analyse intricate spectrum datasets with Raman spectroscopy's ability to produce molecular fingerprints of biochemical alterations linked to cancer improves diagnostic precision. Using the PRISMA methodology, studies published from 2017 to 2024 were examined, with an emphasis on those that reported sensitivity and specificity values greater than 80%. With sensitivity and specificity frequently over 90%, the nine included studies show that Raman spectroscopy combined with machine learning methods such as support vector machines, convolutional neural networks, and linear discriminant analysis yields good diagnostic metrics. The investigation highlights Raman spectroscopy's adaptability in analysing biological material, such as tissues and serum, with prospective uses extending to intraoperative, real-time evaluations. Although encouraging, there are still issues that need to be resolved, like the requirement for common frameworks, multi-centre validation, and affordable technology. A thorough assessment of RS-ML applications is given by this study, which also offers insights into its therapeutic potential and directs future studies in breast cancer detection. CLINICAL TRIAL NUMBER: Not applicable.