Deep Ensemble approach for Enhancing Brain Tumor Segmentation in Resource-Limited Settings
Jeremiah Fadugba, Isabel Lieberman, Olabode Ajayi, Mansour Osman, Solomon Oluwole Akinola, Tinashe Mustvangwa, Dong Zhang, Udunna C Anazondo, Raymond Confidence
TL;DR
The paper tackles automated brain tumor segmentation in Sub-Saharan Africa by addressing generalization challenges across heterogeneous data. It introduces a deep ensemble of UNet3D, V-Net, and MSA-VNet trained on BraTS-GLI and fine-tuned on BraTS-SSA, fused with STAPLE, and followed by post-processing. The ensemble achieves competitive Dice scores across Tumor Core, Whole Tumor, and Enhancing Tumor (0.8358, 0.8521, 0.8167 respectively on BraTS-SSA) and demonstrates improvements over individual models in cross-validation and online evaluations, highlighting potential for scalable, automated segmentation in low-resource settings. Overall, the approach offers a robust, clinically relevant solution for brain tumor delineation in SSA, contributing to timely diagnosis and treatment planning where radiology resources are limited.
Abstract
Segmentation of brain tumors is a critical step in treatment planning, yet manual segmentation is both time-consuming and subjective, relying heavily on the expertise of radiologists. In Sub-Saharan Africa, this challenge is magnified by overburdened medical systems and limited access to advanced imaging modalities and expert radiologists. Automating brain tumor segmentation using deep learning offers a promising solution. Convolutional Neural Networks (CNNs), especially the U-Net architecture, have shown significant potential. However, a major challenge remains: achieving generalizability across different datasets. This study addresses this gap by developing a deep learning ensemble that integrates UNet3D, V-Net, and MSA-VNet models for the semantic segmentation of gliomas. By initially training on the BraTS-GLI dataset and fine-tuning with the BraTS-SSA dataset, we enhance model performance. Our ensemble approach significantly outperforms individual models, achieving DICE scores of 0.8358 for Tumor Core, 0.8521 for Whole Tumor, and 0.8167 for Enhancing Tumor. These results underscore the potential of ensemble methods in improving the accuracy and reliability of automated brain tumor segmentation, particularly in resource-limited settings.