The following is a summary of “Hyperspectral imaging benchmark based on machine learning for intraoperative brain tumour detection,” published in the November 2023 issue of Oncology by Leon et al.

Brain surgery stands as a primary treatment for brain tumors, yet neurosurgeons encounter the challenge of precisely delineating tumor boundaries for optimal resection while avoiding harm to healthy tissue and potential neurological complications. Hyperspectral Imaging (HSI) has been promised in various medical applications for tumor detection. In this study, the researchers present a robust k-fold cross-validation approach demonstrating the potential of HSI combined with a proposed processing framework as an effective intraoperative tool for identifying and delineating brain tumors. 

Their analysis, conducted on an in-vivo brain database comprising 61 HS images from 34 patients, includes primary (high-grade and low-grade) and secondary tumors. Through the fusion of spectral and spatial information, their approach achieved the highest median macro F1-Score result of 70.2 ± 7.9% on the test set. This research aims to establish a benchmark using machine learning methodologies to advance in-vivo brain tumor detection and delineation with hyperspectral imaging. The findings suggest the potential for using HSI as a real-time decision-support tool during neurosurgical procedures, providing a foundation for further developments in this critical area of study.

This benchmark, built on machine learning techniques, signifies a significant step towards leveraging hyperspectral imaging for precise intraoperative tumor detection and delineation in brain surgery. The results highlight the potential for further refinements and practical applications of HSI as an invaluable tool in neurosurgical workflows.

Source: nature.com/articles/s41698-023-00475-9