Skip to main content
. 2019 Jan 11;12:1046. doi: 10.3389/fnins.2018.01046

FIGURE 1.

FIGURE 1

Schematic flowchart of the automated grading framework. We first automatically selected the representative regions of interest (ROIs) from the H&E images. Based on these ROIs, we extracted and selected important visual, sub-visual, and immunohistochemical features. We established automated machine learning models with these features for glioma grading. The grading results output from the model were further explained with the LIME algorithm.