Deep learning of rhabdomyosarcoma pathology images for classification and survival outcome prediction.

Researchers

Journal

Modalities

Models

Abstract

Rhabdomyosarcoma (RMS), the most common malignant soft tissue tumor in children, has several histologic subtypes that influence treatment and predict patient outcomes. Assistance of histological classification for pathologists, as well as optimized predictive biomarkers, are in a strong need. A convolutional neural network (CNN) for RMS histology subtype classification was developed using digitized pathology images from 80 patients collected at time of diagnosis. A subsequent embryonal rhabdomyosarcoma (eRMS) prognostic model was also developed in a cohort of 60 eRMS patients. The RMS classification model reached a performance of area under the receiver operating curve (AUC) of 0.94 for alveolar rhabdomyosarcoma (aRMS) and AUC of 0.92 for eRMS at slide-level in the test dataset (n = 192). The eRMS prognosis model separated the patients into predicted high- and low-risk groups with significantly different event-free survival (EFS) outcome (likelihood ratio test, p value = 0.02) in the test dataset (n = 136). The predicted risk group is significantly associated with patient EFS outcome after adjusting for patient age and sex (predicted high vs. low-risk group HR = 4.64, 95% CI = 1.05-20.57, p value = 0.04). This is the first comprehensive study to develop computational algorithms for subtype classification and prognosis prediction for RMS histopathology images. Such models can be an aid to pathology evaluation and provide additional parameters for risk stratification.Copyright © 2022. Published by Elsevier Inc.