A Feature Extraction Using Probabilistic Neural Network and BTFSC-Net Model with Deep Learning for Brain Tumor...

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Computational Pathology

Background:

• Brain tumors require accurate segmentation and classification for effective treatment.
• Existing methods for brain tumor analysis often face challenges with noise, feature extraction, and classification accuracy.

Purpose of the Study:

• To develop a hybrid system, Brain Tumor Fusion-based Segments and Classification-Non-enhancing tumor (BTFSC-Net), for enhanced brain tumor classification.
• To integrate advanced techniques for image preprocessing, fusion, segmentation, feature extraction, and classification.

Main Methods:

• Applied Hybrid Probabilistic Wiener Filter (HPWF) for noise reduction.
• Utilized deep learning convolutional neural networks (DLCNN) for image fusion incorporating Robust Edge Analysis (REA).
• Employed Adaptive Fuzzy C-Means integrated K-Means (HFCMIK) for segmentation and extracted features using Redundant Discrete Wavelet Transform (RDWT), empirical color, and Gray-Level Co-occurrence Matrix (GLCM).
• Deployed a Deep Learning Probabilistic Neural Network (DLPNN) for final tumor classification.

Main Results:

• The proposed BTFSC-Net model demonstrated superior performance compared to traditional techniques.
• Achieved 99.21% accuracy in image segmentation.
• Reached 99.46% accuracy in brain tumor classification.

Conclusions:

• The BTFSC-Net system significantly outperforms previous approaches in brain tumor image fusion, segmentation, feature extraction, and classification.
• The developed method offers enhanced quantitative and visual performance for brain tumor analysis.