Figure 1 | Scientific Reports

Figure 1

From: Comparative analysis of machine learning approaches to classify tumor mutation burden in lung adenocarcinoma using histopathology images

Figure 1

Overview of modeling approaches and inference. Overview of the histologic-subtype approach (top) and the weakly supervised approach (bottom). In the histologic-subtype approach, feature maps of histologic patterns produced by LungCNN-Histo in stage 1 and are combined with clinical features to predict TMB-status in stage 2. The weakly supervised approach also utilizes a two stage approach, whereby a weakly supervised model (WS-S1) is developed to produce patch-level TMB predictions at inference. The patch-level data are aggregated over the entire tissue area in stage 1 and combined with clinical features to provide TMB classification in stage 2.

Back to article page