Brain tumor classification for MRI images using dual-discriminator conditional generative adversarial network.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• Accurate brain tumor detection and classification are crucial for effective treatment planning.
• Existing methods often face challenges with image noise and computational efficiency.
• Machine learning, particularly deep learning, shows promise for improving diagnostic accuracy.

Purpose of the Study:

• To develop and evaluate an advanced method for brain tumor classification in MRI images.
• To enhance the accuracy and efficiency of brain tumor detection using a novel DDCGAN architecture.
• To optimize the classification model using a metaheuristic approach for improved performance.

Main Methods:

• Brain MRI images from the Brats dataset were preprocessed using Structural interval gradient filtering.
• Feature extraction was performed using Empirical Wavelet Transform (EWT).
• A Dual-Discriminator Conditional Generative Adversarial Network (DDCGAN), optimized with Border Collie Optimization (BCO), was employed for classification.

Main Results:

• The proposed BCO-DDCGAN-MRI-BTC method achieved a high sensitivity of 99.58%.
• The system accurately classifies brain tumors into glioma, meningioma, pituitary gland, and normal categories.
• Experimental results demonstrated superior performance compared to KSVM-HHO-BTC, JT-TCDNN-BTC, and YOLOv2-CNN-BTC methods.

Conclusions:

• The BCO-DDCGAN-MRI-BTC method significantly improves brain tumor classification accuracy in MRI images.
• The proposed approach reduces computational time and errors, offering a more efficient diagnostic tool.
• This research highlights the potential of DDCGANs combined with metaheuristic optimization for medical image analysis.