A lightweight network for brain MRI segmentation.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neuroscience

Background:

• Accurate brain Magnetic Resonance Imaging (MRI) segmentation is vital for diagnosing and tracking neurological conditions.
• Existing deep learning models often struggle to balance high accuracy with computational efficiency for clinical applications.

Purpose of the Study:

• To develop a novel, lightweight deep learning framework for precise brain MRI segmentation.
• To improve the efficiency and accuracy of automated segmentation for clinical neuroimaging.

Main Methods:

• Utilized EfficientNet B0 as an encoder for multi-scale feature extraction with reduced complexity.
• Incorporated Visual State-Space blocks and multi-scale attention mechanisms for enhanced global context and feature refinement.
• Employed a U-Net-inspired decoder with skip connections and a hybrid loss function (Active Contour Loss and Focal Loss) for robust training.

Main Results:

• Achieved high segmentation accuracy with a computationally efficient, lightweight architecture.
• Demonstrated superior segmentation performance compared to existing state-of-the-art methods.
• Successfully segmented complex anatomical structures and lesions with precise boundary delineation.

Conclusions:

• The proposed deep learning framework offers a promising solution for accurate and efficient brain MRI segmentation.
• This approach facilitates improved disease identification and monitoring in clinical neuroimaging.
• The lightweight design makes the model suitable for real-world deployment in medical settings.