NIMG-102. RAPNO-DEFINED SEGMENTATION AND VOLUMETRIC ASSESSMENT OF PEDIATRIC BRAIN TUMORS ON MULTI-PARAMETRIC MRI SCANS USING DEEP LEARNING; A ROBUST TOOL WITH POTENTIAL APPLICATION IN TUMOR RESPONSE ASSESSMENT

Abstract Volumetric measurements of whole tumor and its components on MRI scans, facilitated by automatic segmentation tools, are essential to reduce inter-observer variability in monitoring progression response assessment for pediatric brain tumors. Here, we present a fully model based deep learning that reliably delineates the recommended Response Assessment Pediatric Neuro-Oncology (RAPNO) working group evaluation treatment response. Multi-parametric (mpMRI) scans (T1-pre, T1-post, T2, T2-FLAIR), acquired multiple scanners with different field strengths vendors, cohort 218 patients variety histologically confirmed subtypes were collected. The mpMRI co-registered manually segmented experienced neuroradiologists consensus identify subregions including enhancing (ET), non-enhancing (NET), cystic (CC), peritumoral edema (ED) regions. A convolutional neural network DeepMedic architecture was trained using as inputs subregions. showed excellent performance tumor, suggested median dice 0.90/0.85 validation (n = 44)/independent test 22) sets. ET (union NET, CC, ED) scores 0.78/0.84 0.76/0.74 validation/test sets, respectively. automated manual segmentations demonstrated strong agreement estimating VASARI (Visually AcceSAble Rembrandt Images) features Pearson’s correlation coefficient R > 0.75 (p < 0.0001) ET, ED components. Our proposed method developed protocols, equipment, from subtypes, shows potential application reliable generalizable volumetric which can be used clinical trials.