Reproduce “TumorXAI: Self-Supervised Deep Learning Framework for Explainable Brain MRI Tumor Classification” on OpenResearch

About this paper

Abrar Hossain Zahin, Amit Kumar Saha, Tanvir Mridha, Saifur Rahman, Jannatul Ferdous Prome, Raima Husna, Israt Jahan, Ahmed Wasif Reza

Researchers developed TumorXAI, a self-supervised deep learning framework for brain MRI tumor classification that addresses data scarcity and interpretability challenges. The framework, specifically its SimCLR implementation, achieved 99.64% accuracy, precision, recall, and F1-score in multi-class brain tumor classification, complemented by visual explanations from integrated XAI techniques.

Key insights

• Self-supervised learning effectively addresses the medical data scarcity problem by enabling models to learn robust feature representations from unlabeled MRI images.
• Fine-tuning SSL-pretrained models significantly enhances classification performance, with SimCLR demonstrating superior overall metrics across various evaluation phases.
• Integrating XAI with SSL models provides critical interpretability, visually confirming that the models focus on clinically relevant tumor areas, thereby fostering trust in AI-driven diagnoses.

Summary by alphaXiv.