Original title: Towards Machine Learning-based Quantitative Hyperspectral Image Guidance for Brain Tumor Resection
Authors: David Black, Declan Byrne, Anna Walke, Sidong Liu, Antonio Di leva, Sadahiro Kaneko, Walter Stummer, Septimiu Salcudean, Eric Suero Molina
This article delves into improving brain tumor resection by utilizing hyperspectral imaging and machine learning. It explores five different fluorophores’ emission spectra found in human brain tumors, aiming to classify tumors, grades, margins, and genetic mutations in 184 patients. Employing machine learning models, including random forests and multi-layer perceptrons, achieved significant accuracies ranging from 74% to 93%. Notably, the abundance of these fluorophores varied across tumor margins and grades, indicating potential as optical biomarkers. The study highlights the potential for hyperspectral imaging combined with machine learning to create intraoperative systems for better classification in neurosurgery, promising advancements in distinguishing tumor tissues and aiding in complete resections of brain tumors.
Original article: https://arxiv.org/abs/2311.10321