EnSLDe: an enhanced short-range and long-range dependent system for brain tumor classification

Area of Science:

• Neurology
• Medical Imaging
• Artificial Intelligence

Background:

• Brain tumors significantly impair nervous system function.
• Existing classification models often neglect crucial long-range information, limiting accuracy.
• Accurate brain tumor classification is vital for effective treatment planning.

Purpose of the Study:

• To propose an enhanced system for brain tumor classification that addresses limitations in capturing long-range dependencies.
• To introduce the EnSLDe model, integrating feature extraction, enhancement, and classification for improved performance.
• To enhance the accuracy of brain tumor classification by effectively utilizing both local and global contextual information.

Main Methods:

• The proposed EnSLDe model comprises three modules: Feature Extraction Module (FExM), Feature Enhancement Module (FEnM), and Classification Module.
• FExM utilizes a multi-scale parallel subnetwork to fuse shallow and deep features.
• FEnM captures long-range dependencies and retains critical local-scale information, followed by classification.

Main Results:

• The EnSLDe model was validated on public datasets including glioma, meningioma, and pituitary tumors.
• The system achieved good experimental results, demonstrating its effectiveness.
• The model shows potential for accurate brain tumor classification.

Conclusions:

• The EnSLDe model successfully integrates short-range and long-range information for enhanced brain tumor classification.
• The proposed system demonstrates significant potential in improving the accuracy of brain tumor diagnosis.
• This approach offers a promising direction for developing more robust brain tumor classification systems.