Generative adversarial network for Multimodal Contrastive Domain Sharing based on efficient invariant feature-centric...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Accurate brain tumor classification is crucial but challenging with current methods.
• Existing techniques often use generic features, limiting classification performance.
• There is a need for advanced methods to improve the accuracy of brain tumor categorization.

Purpose of the Study:

• To propose a novel Multimodal Contrastive Domain Sharing Generative Adversarial Network for Improved Brain Tumor Classification Based on Efficient Invariant Feature Centric Growth Analysis (MCDS-GNN-IBTC-CGA).
• To enhance the accuracy and efficiency of classifying brain tumor images into Glioma, Meningioma, and Pituitary types.
• To address the limitations of generic feature extraction in existing brain tumor classification methods.

Main Methods:

• Image preprocessing using Range-Doppler Matched Filter (RDMF) to enhance image quality.
• Feature extraction via Ternary Pattern and Discrete Wavelet Transforms (TPDWT), focusing on specific image characteristics.
• Classification using a Multimodal Contrastive Domain Sharing Generative Adversarial Network (MCDS-GNN), optimized by the Coati Optimization Algorithm (COA).

Main Results:

• The proposed MCDS-GNN-IBTC-CGA method demonstrated superior performance compared to state-of-the-art techniques.
• Achieved significant improvements in accuracy, specificity, sensitivity, precision, and F1-score.
• Outperformed methods like PDCNN-BTC, AGCNN-BTC, DCRN-BTC, FCNN-BTC, and CNN-MLP-BTC by notable margins.

Conclusions:

• The MCDS-GNN-IBTC-CGA offers a significant advancement in automated brain tumor classification.
• The integration of RDMF, TPDWT, MCDS-GNN, and COA provides a robust framework for medical image analysis.
• This approach holds promise for improving diagnostic accuracy and patient outcomes in neuro-oncology.