Figure 5 from Predicting Molecular Subtype and Survival of Rhabdomyosarcoma Patients Using Deep Learning of H&E Images: A Report from the Children's Oncology Group

Abstract

<p><i>MYOD1</i> gene mutation prediction from H&E images. <b>A,</b> Workflow for deep learning of <i>MYOD1</i> mutations from FN-RMS WSIs. <b>B</b> and <b>C,</b> Representative (<b>B</b>) H&E images and (<b>C</b>) class activation maps of a MYOD1 wild-type tumor and a tumor with a MYOD1 p.L122R mutation (VAF = 0.919). <b>D,</b> Confusion matrix for predictions on a test dataset. Micro F1, Macro F1, and Matthew's correlation coefficient shown below. <b>E,</b> Performance statistics for <i>MYOD1</i> mutation prediction. <b>F,</b> Average ROC curve for <i>MYOD1</i> mutation prediction using validation data. <b>G,</b> Plot of <i>MYOD1</i> mutation positive ratio thresholds with test datasets (blue = known <i>MYOD1</i> wild-type; red = known <i>MYOD1</i> mutant) and independent dataset of known <i>MYOD1</i>-mutant tumors <i>n</i> = 10 (green) shown as geometric mean ± 1 geometric SD.</p>

Ähnliche Arbeiten