Early prognostication of overall survival for pediatric diffuse midline gliomas using MRI radiomics and machine learning: a two-center study

Abstract

ABSTRACTBackgroundDiffuse midline gliomas (DMG) are aggressive pediatric brain tumors that are diagnosed and monitored through MRI. We developed an automatic pipeline to segment subregions of DMG and select radiomic features that predict patient overall survival (OS).MethodsWe acquired diagnostic and post-radiation therapy (RT) multisequence MRI (T1, T1ce, T2, T2 FLAIR) and manual segmentations from two centers of 53 (internal cohort) and 16 (external cohort) DMG patients. We pretrained a deep learning model on a public adult brain tumor dataset, and finetuned it to automatically segment tumor core (TC) and whole tumor (WT) volumes. PyRadiomics and sequential feature selection were used for feature extraction and selection based on the segmented volumes. Two machine learning models were trained on our internal cohort to predict patient 1-year survival from diagnosis. One model used only diagnostic tumor features and the other used both diagnostic and post-RT features.ResultsFor segmentation, Dice score (mean [median]±SD) was 0.91 (0.94)±0.12 and 0.74 (0.83)±0.32 for TC, and 0.88 (0.91)±0.07 and 0.86 (0.89)±0.06 for WT for internal and external cohorts, respectively. For OS prediction, accuracy was 77% and 81% at time of diagnosis, and 85% and 78% post-RT for internal and external cohorts, respectively. Homogeneous WT intensity in baseline T2 FLAIR and larger post-RT TC/WT volume ratio indicate shorter OS.ConclusionsMachine learning analysis of MRI radiomics has potential to accurately and non-invasively predict which pediatric patients with DMG will survive less than one year from the time of diagnosis to provide patient stratification and guide therapy.KEY POINTSAutomatic machine learning approach accurately predicts DMG survival from MRIHomogeneous whole tumor intensity in baseline T2 FLAIR indicates worse prognosisLarger post-RT tumor core/whole tumor volume ratio indicates worse prognosisIMPORTANCE OF STUDYStudies of pediatric DMG prognostication have relied on manual tumor segmentation from MRI, which is impractical and variable in busy clinics. We present an automatic imaging tool based on machine learning to segment subregions of DMG and select radiomic features that predict overall survival. We trained and evaluated our tool on multisequence, two-center MRIs acquired at the time of diagnosis and post-radiation therapy. Our methods achieved 77-85% accuracy for DMG survival prediction. The data-driven study identified that homogeneous whole tumor intensity in baseline T2 FLAIR and larger post-therapy tumor core/whole tumor volume ratio indicates worse prognosis. Our tool can increase the utility of MRI for predicting clinical outcome, stratifying patients into risk-groups for improved therapeutic management, monitoring therapeutic response with greater accuracy, and creating opportunities to adapt treatment. This automated tool has potential to be easily incorporated in multi-institutional clinical trials to provide consistent and repeatable tumor evaluation.