Multi-feature fusion and dandelion optimizer based model for automatically diagnosing the gastrointestinal diseases

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computer-Aided Diagnosis

Background:

• Gastrointestinal diseases like gastritis, reflux, and dyspepsia are highly prevalent.
• Symptom overlap often leads to diagnostic confusion.
• Accurate and rapid diagnosis is crucial for effective treatment and cancer prevention.

Purpose of the Study:

• To develop a novel artificial intelligence-based hybrid method for accurate classification of gastrointestinal images.
• To enhance the diagnostic accuracy of anatomical landmarks, pathological findings, and polyps.
• To improve computer-aided diagnosis systems for gastrointestinal diseases.

Main Methods:

• Utilized pre-trained InceptionV3 and MobileNetV2 architectures for feature extraction.
• Merged features from both architectures to capture diverse image characteristics.
• Employed Dandelion Optimizer (DO) for feature selection and Support Vector Machine (SVM) for classification.

Main Results:

• The proposed hybrid method achieved a high classification accuracy of 93.88%.
• The Dandelion Optimizer effectively reduced irrelevant and redundant features.
• Outperformed other convolutional neural network (CNN) models in experimental comparisons.

Conclusions:

• The developed AI hybrid method offers a promising approach for accurate gastrointestinal image classification.
• This technique can significantly aid in faster and more precise diagnosis of gastrointestinal conditions.
• The study highlights the potential of integrating deep learning with metaheuristic optimization for medical image analysis.