Enhancing image-based diagnosis of gastrointestinal tract diseases through deep learning with EfficientNet and...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Gastroenterology

Background:

• Early detection of gastrointestinal diseases (ulcerative colitis, polyps, esophagitis) is vital for effective treatment.
• Manual interpretation of medical images introduces variability and potential inaccuracies.
• Existing deep learning models struggle with overfitting and generalization in medical image analysis.

Purpose of the Study:

• To develop a robust deep learning model for accurate gastrointestinal disease classification.
• To overcome limitations of conventional models, including overfitting and poor generalization.
• To enhance diagnostic accuracy in medical imaging for gastrointestinal conditions.

Main Methods:

• Utilized the EfficientNetB5 architecture for image analysis.
• Implemented a sophisticated data augmentation strategy including horizontal flipping and dynamic scaling.
• Integrated transfer learning, maximal pooling, and extensive regularization techniques.

Main Results:

• Achieved a high test accuracy of 98.89% in classifying gastrointestinal diseases.
• Demonstrated superior performance compared to traditional diagnostic methods.
• Reported a low test loss of 0.230 and high precision metrics, indicating strong generalization.

Conclusions:

• The proposed deep learning framework offers superior performance for automated gastrointestinal disease classification.
• Advanced regularization and augmentation techniques significantly improve model generalization and accuracy.
• This approach enhances medical imaging diagnostics, potentially leading to earlier and more precise disease interventions.