Cascaded Mutual Enhancing Networks for Brain Tumor Subregion Segmentation in Multiparametric MRI.

Researchers

Journal

Modalities

Models

Abstract

Accurate segmentation of glioma and its subregions plays an important role in radiotherapy treatment planning. Due to a very populated multiparameter magnetic resonance imaging (mpMRI) image, manual segmentation task can be very time-consuming, meticulous, and prone to subjective errors. Here, we propose a novel deep learning (DL) framework based on mutual enhancing networks (MENs) to automatically segment brain tumor subregions. Proposed framework is suitable for segmentation of brain tumor subregions owing to contribution of retina U-Net followed by implementation of mutual enhancing strategy between classification module and segmentation module. Retina U-Net is trained to accurately identify view-of-interest (VOIs) and feature maps of whole tumor (WT), which are then transferred to classification module and segmentation module. Subsequently, classification localization map (CLM) generated by classification module is integrated with segmentation module to bring forth mutual enhancing strategy. In this way, our proposed framework first focuses on WT through retina U-Net, and since WT consists subregions, mutual enhancing strategy then further aims to classify and segment subregions embedded within WT. We implemented and evaluated our proposed framework on brain tumor segmentation challenge (BraTS) 2020 dataset consisting of 369 cases. We performed 5-fold cross validation on 200 datasets and hold-out test on remaining 169 cases. To demonstrate the effectiveness of our network design, we compared our method against the networks without retina U-Net, mutual enhancing strategy, and a recently published Cascaded U-Net architecture. Results of all four methods were compared to the ground truth for segmentation and localization accuracies. Our method yielded significantly (P < 0.01) better values of dice-similarity-coefficient, center-of-mass-distance, and volume difference compared to all three competing methods across all tumor labels on both validation and hold-out dataset. Overall quantitative and statistical results of this work demonstrate ability of our method to both accurately and automatically segment brain tumor subregions.© 2022 Institute of Physics and Engineering in Medicine.