Combined molecular subtyping, grading, and segmentation of glioma using multi-task deep learning

Abstract

Background Accurate characterization of glioma is crucial for clinical decision making. A delineation of the tumor is also desirable in the initial decision stages but is time-consuming. Previously, deep learning methods have been developed that can either non-invasively predict the genetic or histological features of glioma, or that can automatically delineate the tumor, but not both tasks at the same time. Here, we present our method that can predict the molecular subtype and grade, while simultaneously providing a delineation of the tumor. Methods We developed a single multi-task convolutional neural network that uses the full 3D, structural, preoperative MRI scans to predict the IDH mutation status, the 1p/19q co-deletion status, and the grade of a tumor, while simultaneously segmenting the tumor. We trained our method using a patient cohort containing 1508 glioma patients from 16 institutes. We tested our method on an independent dataset of 240 patients from 13 different institutes. Results In the independent test set, we achieved an IDH-AUC of 0.90, an 1p/19q co-deletion AUC of 0.85, and a grade AUC of 0.81 (grade II/III/IV). For the tumor delineation, we achieved a mean whole tumor Dice score of 0.84. Conclusions We developed a method that non-invasively predicts multiple, clinically relevant features of glioma. Evaluation in an independent dataset shows that the method achieves a high performance and that it generalizes well to the broader clinical population. This first-of-its-kind method opens the door to more generalizable, instead of hyper-specialized, AI methods.

AdvancedMRTechniques for Preoperative Glioma Characterization: Part 2

“…Van der Voort et al developed an algorithm with data from 1,508 patients which simultaneously segments, grades, and genotypes of glioma in terms of IDH mutation and 1p/19q codeletion. The model reached accuracies of 80%–90% in an entirely independent dataset of 240 patients, which were similar to those reached in the development set, evidencing the robustness of the algorithm against scanner, site, and protocol variations 106 …”

“…The model reached accuracies of 80%-90% in an entirely independent dataset of 240 patients, which were similar to those reached in the development set, evidencing the robustness of the algorithm against scanner, site, and protocol variations. 106 As the contrast-agent injection is part of clinical protocols and DSC is highly validated, research protocols commonly include DSC. This makes DSC datasets fairly available in glioma datasets.…”

White matter injury detection based on preterm infant cranial ultrasound images

“…In order to find the most suitable multi-task deep learning network (MTDL-Net), we compared Fast R-CNN ( 25 ), Faster R-CNN ( 26 ), Mask R-CNN ( 27 ) and RS-Net ( 28 ) for white matter segmentations and WMI status prediction. Based on the result of this ablation experiment (detailed in Supplementary Material ), we used Mask R-CNN as our MTDL-Net.…”

Computer-aided diagnosis and prediction in brain disorders

“…For differential diagnosis in brain cancer, numerous MRI-based machine learning approaches have been presented. These developments have partly been facilitated by the availability of several valuable public datasets, see for example the overviews in [141,92]. Most literature is dedicated to glioma characterisation, which is therefore discussed in more detail here.…”