MMRINet: Efficient Mamba-Based Segmentation with Dual-Path Refinement for Low-Resource MRI Analysis
Researchers have developed MMRINet, a lightweight AI model designed for efficient brain tumor segmentation in MRI scans, particularly for low-resource clinical settings. The model utilizes Mamba state-space models to replace computationally intensive self-attention mechanisms, enabling effective long-range context modeling with fewer parameters. MMRINet incorporates dual-path feature refinement and progressive feature aggregation to enhance segmentation accuracy and boundary sharpness, even with limited data. Tested on a dataset from Nigerian clinical sites, MMRINet achieved competitive performance, demonstrating its potential for AI-assisted neuro-oncology in underserved regions. AI
IMPACT Offers a practical solution for AI-assisted neuro-oncology in resource-constrained environments by reducing computational demands.