Deep learning-based automatic tumor burden assessment of pediatric high-grade gliomas, medulloblastomas, and other leptomeningeal seeding tumors

Abstract

Background Longitudinal measurement of tumor burden with MRI is an essential component of response assessment in pediatric brain tumors. We developed a fully automated pipeline for the segmentation of tumors in pediatric high-grade gliomas, medulloblastomas, and leptomeningeal seeding tumors. We further developed an algorithm for automatic 2D and volumetric size measurement of tumors. Methods A preoperative and postoperative cohort were randomly split into training and testing sets in a 4:1 ratio. A 3D U-Net neural network was trained to automatically segment the tumor on T1 contrast-enhanced and T2/FLAIR images. The product of the maximum bidimensional diameters according to the RAPNO criteria (AutoRAPNO) was determined. Performance was compared to that of two expert human raters who performed assessments independently. Volumetric measurements of predicted and expert segmentations were computationally derived and compared. Results A total of 794 pre-operative MRIs from 794 patients and 1,003 post-operative MRIs from 122 patients were included. There was excellent agreement of volumes between preoperative and postoperative predicted and manual segmentations, with ICCs of 0.912 and 0.960 for the two preoperative and 0.947 and 0.896 for the two postoperative models. There was high agreement between AutoRAPNO scores on predicted segmentations and manually calculated scores based on manual segmentations (Rater 2 ICC=0.909; Rater 3 ICC=0.851). Lastly, the performance of AutoRAPNO was superior in repeatability to that of human raters for MRIs with multiple lesions. Conclusions Our automated deep learning pipeline demonstrates potential utility for response assessment in pediatric brain tumors. The tool should be further validated in prospective studies.

Automated Tumor Segmentation and Brain Tissue Extraction from Multiparametric MRI of Pediatric Brain Tumors: A Multi-Institutional Study

“…Such models do not generalize well to PBTs 15 due to the different radiological appearance of tumors compared to adult brain tumors, and the anatomical differences as a result of the developing brain in children 16 . A few studies have proposed different DL solutions to the PBT segmentation problem, with whole tumor Dice scores ranging between 0.68-0.76 [17][18][19][20] . These approaches show lower performance than the models proposed for adult brain tumor segmentation, are often only designed for a particular histology, only segment the whole tumor without subregions, or segment the tumors based on one or two MRI sequences.…”

Direct evaluation of progression or regression of disease burden in brain metastatic disease with Deep Neuroevolution

“…Recent work has used supervised deep learning to measure the volumes of primary brain tumors [19,20], comparing their volumes across successive scans as a surrogate to RANO [11]. In contrast to primary brain tumor patients who typically have just one target, patients with brain metastases often have many targets.…”

Artificial intelligence applications in pediatric oncology diagnosis

“…AI applications in the identification of diagnostic signature AI-based principles have been used for the detection and segmentation of pediatric malignant tumors. For example, Wu et al [64] used a residual fusion network to detect osteosarcomas on MRI scans; Peng et al [65] used a CNN for automated pediatric brain tumor detection and segmentation on MRI scans with automatic two-dimensional (2D) and volumetric size measurement of tumors; Strijbis et al [66] used a CNN for automated eye and tumor segmentation on MRI in retinoblastoma patients; and Bouget et al [67] used three-dimensional neural network architectures to automatically detect meningioma on MRI scans.…”