A Unified Benchmark of Deep Learning Models for Multi-task 3D Brain Tumor Segmentation from Magnetic Resonance Imaging

2026-07-30

AI Summary

This study addresses the lack of a standardized benchmark for fair comparison among deep learning models in multi-task 3D brain tumor segmentation. For the first time, it systematically evaluates the performance of five prominent architectures—3D U-Net, SegResNet, Swin UNETR, SegMamba, and SegMambaV2—under identical conditions, including the same dataset (BraTS 2023/2024), preprocessing pipeline, training protocol, and evaluation metrics. Through comprehensive assessment across multiple dimensions—segmentation accuracy, inference time, and model size—the work elucidates the inherent trade-offs between accuracy and efficiency. These findings provide clinically relevant guidance for model selection and deployment, effectively establishing a much-needed standardized reference framework in the field.

Abstract

Automatic brain tumor segmentation from magnetic resonance imaging (MRI) has become a fundamental task in computer-assisted diagnosis, treatment planning, and disease monitoring. Although numerous deep learning architectures have recently been proposed, objective comparisons remain challenging because published studies often employ different datasets, preprocessing strategies, training protocols, and evaluation procedures. This work presents a unified experimental benchmark for comparing representative convolutional neural networks (CNNs), Transformer-based models, and recent State Space Model (SSM) architectures under homogeneous experimental conditions. Five state-of-the-art three-dimensional segmentation models, including 3D U-Net, SegResNet, Swin UNETR, SegMamba, and SegMambaV2, are evaluated on two brain tumor segmentation datasets representing distinct clinical scenarios: intracranial meningioma segmentation (BraTS 2023) and post-treatment glioma segmentation (BraTS 2024). All architectures are trained using identical preprocessing, data augmentation, optimization strategies, and evaluation protocols to ensure a fair comparison. Performance is assessed using segmentation accuracy metrics together with computational cost indicators, including inference time and the size of each model. The results provide practical insights into the trade-offs between segmentation accuracy and computational efficiency, highlighting the suitability of different architectural paradigms for challenging three-dimensional brain tumor segmentation tasks.

Problem

• brain tumor segmentation
• MRI
• deep learning
• benchmark
• 3D segmentation

Innovation

• unified benchmark
• 3D brain tumor segmentation
• deep learning architectures
• State Space Models
• fair model comparison