Fig. 2: Overview of the device and method.
From: Towards machine learning-based quantitative hyperspectral image guidance for brain tumor resection
This shows the steps of the data processing, from top right to bottom left. The imaging device first captures a hyperspectral data cube from which the ex vivo tumor tissue is segmented and the corresponding spectra are extracted, processed, and unmixed into abundance vectors or dimensionally reduced using PCA. The abundances can be used, for example, to create a PpIX overlay plot or perform statistical analyses. In this study, they are used as inputs to machine learning models to classify tumor type, margins, WHO grades, and IDH mutation.
