FWNNet: Presentation of a New Classifier of Brain Tumor Diagnosis Based on Fuzzy Logic and the Wavelet-Based Neural...

Area of Science:

• Artificial Intelligence
• Medical Imaging
• Computational Neuroscience

Background:

• Accurate brain tumor diagnosis and segmentation are critical for effective treatment planning.
• Traditional machine learning classifiers exhibit varying performance in analyzing complex medical image data.
• Existing feature extraction and classification methods may not fully capture the intricate details of brain tumors in MRI scans.

Purpose of the Study:

• To introduce a novel Fuzzy Wavelet Neural Network (FWNNet) classifier for enhanced brain tumor diagnosis.
• To evaluate the performance of FWNNet in brain tumor classification and segmentation using MRI images.
• To compare the efficacy of FWNNet against established classifiers like Decision Trees (DT), K-Nearest Neighbors (KNN), and Support Vector Machines (SVM).

Main Methods:

• Development of a novel classifier integrating fuzzy logic and wavelet transformation within a neural network architecture (FWNNet).
• Feature extraction using a fractal model with four Gaussian functions applied to 2000 MRI images for classification.
• Implementation of a supervised segmentation method utilizing the FWNNet layer, optimized with a Particle Swarm Optimization (PSO) algorithm for training on 80 MRI images.

Main Results:

• The proposed FWNNet classifier achieved a 100% accuracy rate in brain tumor diagnosis, outperforming DT (93.5%), KNN (87.6%), Linear Discriminant Analysis (LDA) (61.5%), Naive Bayes (NB) (57.5%), Multilayer Perceptron (MLP) (68.5%), and SVM (43.6%).
• FWNNet demonstrated superior performance as the best classifier for brain tumor diagnosis, followed by DT and KNN.
• The FWNNet-based segmentation method showed a high true-positive rate, as indicated by ROC curve analysis.

Conclusions:

• The novel FWNNet architecture, combined with fractal feature extraction, represents a highly accurate and effective tool for brain tumor diagnosis from MRI data.
• FWNNet offers a significant advancement in brain tumor segmentation, providing reliable results with a high true-positive rate.
• This study highlights the potential of integrating fuzzy logic, wavelet transformation, and neural networks for improving medical image analysis in oncology.