Bridging Topology and Deep Representation Learning: A TDA-ViT Fusion Model for Four-Class Brain Tumor Classification
Researchers have developed a novel fusion model that combines Topological Data Analysis (TDA) with Vision Transformers (ViTs) for improved brain tumor classification from MRI scans. This TDA-ViT model extracts both geometric/topological features and semantic representations, fusing them for enhanced discrimination. The approach achieved a remarkable 99.10% accuracy on the BRISC2025 dataset, outperforming existing state-of-the-art methods. AI
IMPACT Enhances medical imaging diagnostics by improving accuracy and robustness in brain tumor classification.