A new brain tumor diagnostic model: Selection of textural feature extraction algorithms and convolution neural...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Brain tumors pose significant health risks, often requiring invasive biopsies for diagnosis.
• Medical imaging techniques like MRI are crucial for non-invasive tumor detection and staging.
• Current diagnostic methods can be limited, necessitating advanced computational approaches.

Purpose of the Study:

• To develop a highly accurate, computer-based hybrid diagnostic model for detecting brain tumors from MRI images.
• To explore the efficacy of combining traditional feature extraction, deep learning, and optimization algorithms for brain tumor diagnosis.
• To provide a reliable and non-invasive tool for clinical application in brain tumor identification.

Main Methods:

• A three-stage hybrid model integrating traditional feature extraction and convolutional neural networks.
• Application of metaheuristic optimization algorithms (genetic algorithms, particle swarm optimization, artificial bee colony) for feature selection.
• Classification of selected features using support vector machine kernels.

Main Results:

• The proposed hybrid model achieved a diagnostic accuracy of 98.22%.
• The integration of multiple feature sets and optimized selection significantly improved diagnostic performance.
• The model demonstrated superior performance compared to methods relying solely on traditional or deep learning approaches.

Conclusions:

• The developed hybrid diagnostic model offers a highly accurate and non-invasive method for brain tumor detection using MRI.
• This approach shows significant potential for integration into clinical practice for improved patient care.
• The study highlights the power of combining AI techniques for enhanced medical image analysis and diagnosis.