TumorXAI: Self-Supervised Deep Learning Framework for Explainable Brain MRI Tumor Classification

2026-05-03

AI Summary

This study addresses the challenges of brain MRI tumor classification posed by high tumor heterogeneity and scarce annotated data. The authors systematically evaluate four self-supervised learning methods—SimCLR, BYOL, DINO, and MoCo v3—by pretraining ResNet-50 on 4,448 unlabeled brain MRIs, followed by fine-tuning and linear evaluation with tailored data augmentation. Their approach achieves high-accuracy classification across 17 brain tumor types, with SimCLR attaining 99.64% accuracy, precision, recall, and F1 score, substantially outperforming supervised baselines. Furthermore, integrating Grad-CAM-based interpretability techniques enhances decision transparency without compromising performance, demonstrating the efficacy and reliability of self-supervised learning in medical imaging scenarios where labeled data are limited.

Abstract

Classifying brain tumors using magnetic resonance imaging (MRI) is crucial for early diagnosis and treatment; however, tumor heterogeneity and a dearth of annotated datasets restrict the use of supervised deep learning approaches. In this work, we use self-supervised learning (SSL) to study multi-class brain tumor classification. Using a ResNet-50 backbone, we evaluate four SSL frameworks including SimCLR, BYOL, DINO, and Moco v3 on a publicly available dataset of 4,448 MRIs with 17 distinct tumor types. On the dataset, SimCLR achieved 99.64% accuracy, 99.64% precision, 99.64% recall, and 99.64% F1-score. The workflow includes preprocessing, fine-tuning, linear evaluation, and SSL pretraining with data augmentations. Results show that, when labels are limited, SSL-pretrained models outperform supervised baselines in terms of F1-score, recall, accuracy, and precision. Additionally, by providing visual insights into model decisions, Explainable AI techniques (Grad-CAM, Grad-CAM++, EigenCAM) enhance interpretability. These results demonstrate SSL's scalability and dependability in diagnosing brain tumors from unlabeled medical data.

Problem

• brain tumor classification
• self-supervised learning
• MRI
• limited annotations
• tumor heterogeneity

Innovation

• Self-Supervised Learning
• Explainable AI
• Brain Tumor Classification
• MRI
• ResNet-50