Efficient Segmentation of Brain Tumor Using FL-SNM with a Metaheuristic Approach to Optimization

Area of Science:

• Medical Image Analysis
• Artificial Intelligence
• Machine Learning

Background:

• Automatic brain tumor detection from MRI is crucial for diagnosis and treatment.
• Existing methods face challenges with high processing times and lower accuracy due to training complexities.
• Accurate tumor segmentation is vital for effective medical image analysis.

Purpose of the Study:

• To introduce a new, efficient automatic segmentation process for brain tumors using MRI data.
• To propose a fuzzy logic with spiking neuron model (FL-SNM) for enhanced tumor segmentation.
• To optimize the segmentation process using swarm intelligence for improved accuracy and reduced processing time.

Main Methods:

• Preprocessing using modified Kuan filter (MKF) with random search algorithm (RSA) for noise reduction and improved PSNR.
• Image smoothing via anisotropic diffusion filter (ADF) to mitigate over-filtering issues.
• Feature extraction using Fisher's linear-discriminant analysis (FLDA) followed by FL-SNM optimized with chicken swarm intelligence (CSI).

Main Results:

• The proposed FL-SNM scheme achieved high accuracy (94.87%), sensitivity (92.07%), and specificity (99.34%).
• Performance metrics including precision (89.36%), recall (88.39%), F-measure (95.06%), G-mean (95.63%), and DSC (91.2%) demonstrated superior results.
• The FL-SNM method outperformed existing convolutional neural networks (CNNs) and hierarchical self-organizing maps (HSOMs).

Conclusions:

• The developed FL-SNM approach offers a highly accurate and efficient solution for automatic brain tumor segmentation from MRI.
• Optimization using chicken swarm intelligence significantly enhances the performance of the fuzzy logic with spiking neuron model.
• This method presents a promising advancement in medical image analysis for neurological diagnostics.