Neural Gas Network Image Features and Segmentation for Brain Tumor Detection Using Magnetic Resonance Imaging Data

Neural Gas Network Image Features and Segmentation for Brain Tumor Detection Using Magnetic Resonance Imaging Data

Link to the paper:

• https://arxiv.org/abs/2301.12176
• https://arxiv.org/pdf/2301.12176
• DOI: https://doi.org/10.48550/arXiv.2301.12176

Please cite:

• Mousavi, S. "Neural Gas Network Image Features and Segmentation for Brain Tumor Detection Using Magnetic Resonance Imaging Data." arXiv preprint arXiv:2301.12176 (2023).

This repository contains the Python implementation of the methods presented in the paper titled "Neural Gas Network Image Features and Segmentation for Brain Tumor Detection Using Magnetic Resonance Imaging Data". The paper combines Neural Gas Networks (NGN) for feature extraction and segmentation with the Firefly Algorithm (FA) for optimizing contrast enhancement.

Table of Contents

• Overview
• Key Contributions
• Features

Overview

Brain tumor detection and segmentation are critical tasks in medical imaging. This repository demonstrates:

• The use of Neural Gas Networks (NGN) for clustering-based feature extraction and segmentation.
• Firefly Algorithm (FA) for optimizing contrast enhancement of MRI images.
• Comparison of methods for accuracy, speed, and computational efficiency.

Key Contributions

1. Feature Extraction with NGN:

   • Neural Gas Network is used for extracting features from MRI images, enabling clustering-based analysis.
   • Efficient handling of high-dimensional pixel data.

2. Image Segmentation:

   • NGN-based segmentation produces interpretable and computationally lightweight results.
   • Demonstrated comparison with traditional techniques like K-means and Otsu thresholding.

3. Contrast Enhancement:

   • Firefly Algorithm optimizes gamma correction parameters to maximize contrast while preserving details.

4. Evaluation Metrics:

   • Accuracy, precision, recall, IoU, and F1-measure are used for quantitative evaluation.

Features

• Neural Gas Network (NGN):

  • Dynamic clustering and feature extraction.
  • Scalable for high-dimensional data.

• Firefly Algorithm (FA):

  • Metaheuristic optimization for image contrast enhancement.
  • Parameter tuning for gamma correction.

• Image Processing Workflow:

  • Preprocessing, feature extraction, segmentation, and contrast enhancement.

Link to the paper:

• https://arxiv.org/abs/2301.12176
• https://arxiv.org/pdf/2301.12176
• DOI: https://doi.org/10.48550/arXiv.2301.12176

Please cite:

• Mousavi, S. "Neural Gas Network Image Features and Segmentation for Brain Tumor Detection Using Magnetic Resonance Imaging Data." arXiv preprint arXiv:2301.12176 (2023).