FastColitisDetector-XAI: An efficient AI model utilizing sparse Autoencoder with explainable AI for ulcerative...

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Gastroenterology

Background:

• Accurate and timely diagnosis of Ulcerative Colitis (UC) is crucial for effective patient management.
• Current diagnostic methods can be invasive and may lack detailed visual insights into disease severity.
• There is a need for AI-driven tools that enhance diagnostic accuracy and provide interpretable results.

Purpose of the Study:

• To develop and evaluate an AI framework for UC diagnosis using Sparse Autoencoders (SA) and Explainable AI (XAI) with Grad-CAM.
• To improve the interpretability and transparency of AI models in medical image analysis for UC detection.
• To achieve high performance metrics in accuracy, precision, recall, and F1 score for UC diagnosis.

Main Methods:

• Utilized Sparse Autoencoders (SA) for dimensionality reduction and feature extraction from medical images.
• Integrated Grad-CAM, a technique within Explainable AI (XAI), to visualize and highlight critical disease regions.
• Employed a machine learning classifier for the final classification of UC presence based on extracted features.

Main Results:

• The proposed SA-XAI model demonstrated superior performance compared to existing methods.
• Achieved remarkable diagnostic accuracy of 98%, precision of 97.5%, recall of 96.4%, and F1 score of 95%.
• Grad-CAM visualizations effectively identified key pathological areas such as inflammation and ulcers.

Conclusions:

• The combined SA-XAI approach offers a highly accurate and interpretable AI solution for UC diagnosis.
• This framework enhances clinician trust by providing transparent decision-making processes.
• The model's high performance indicates its potential as a valuable tool in gastroenterological diagnostics.