Kidney Disease Segmentation and Classification Using Firefly Sigma Seeker and MagWeight Rank Techniques

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Nephrology Diagnostics

Background:

• Deep learning models offer automated analysis of medical images (MRI, CT, ultrasound) for early kidney disease detection.
• Automated diagnosis expedites intervention and treatment, reducing reliance on manual interpretation and improving healthcare efficiency.

Purpose of the Study:

• To enhance parallel convolutional layer architectures for improved kidney disease segmentation.
• To integrate advanced optimization techniques for greater accuracy and computational efficiency.

Main Methods:

• Utilized Firefly Sigma Seeker for dynamic parameter adjustment and early stopping.
• Employed MagWeight Rank to optimize parameter weighting, prune less important weights, and reduce computational time.
• Developed a Multi-Stream Neural Network (MSNN) for efficient kidney disease classification.

Main Results:

• Achieved optimal kidney disease segmentation with 98.2% accuracy.
• Minimized loss to 0.1 and reduced computational time to 15 min 4 s.
• Demonstrated successful avoidance of overfitting through experimental evaluation.

Conclusions:

• The proposed framework significantly enhances kidney disease segmentation accuracy and computational efficiency.
• Advanced optimization techniques integrated into parallel convolutional layers improve diagnostic capabilities.
• The MSNN model provides an efficient and scalable solution for kidney disease classification.