Machine learning applications to diffuse reflectance spectroscopy in optical diagnosis: a systematic review: a systematic review

Diffuse Reflectance Spectroscopy has a strong aptitude for characterizing biological tissues. However, the signals’ broadband, smooth nature requires algorithmic processing, as they are difficult to distinguish visually. The implementation of machine learning models for the processing of these signals has demonstrated high accuracy and led to a range of proposed applications in the medical domain. In this systematic review, we summarize the state of the art of the applications of machine learning for the classification and regression of diffuse reflectance spectroscopy signals, highlight current gaps in research and identify future directions. This review was conducted following the PRISMA guidelines. Seventy-seven studies published between 2005 and May 2024 were retrieved, and an in-depth analysis was conducted. It is concluded that diffuse reflectance spectroscopy and machine learning have strong potential for tissue differentiation in clinical applications, but more rigorous sample stratification, in-vivo validation and explainable algorithm development are required going forward.