Intelligent Diabetes Detection System based on Tongue Datasets

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Diabetology

Background:

• Scanning Electron Microscope (SEM) camera imaging aids in analyzing hidden organs and detecting abnormalities.
• Tongue reflexology indicates organ dysfunction, with changes reflecting in tongue appearance.
• Existing imaging modalities for diagnosis are often expensive and invasive.

Purpose of the Study:

• To develop a non-invasive, cost-effective method for early diabetes detection using tongue images.
• To utilize SEM camera imaging for capturing detailed tongue features.
• To classify individuals as diabetic or normal based on tongue image analysis.

Main Methods:

• Tongue images acquired using an SEM camera.
• Image preprocessing for noise reduction and resizing.
• Segmentation of tongue images using the Greedy Snake Algorithm (GSA).
• Analysis of tongue texture features for classification.

Main Results:

• Organ failure, particularly pancreatic dysfunction, alters tongue color, coating thickness, and crack formation.
• Tongue coating texture can range from white to yellow, with increased thickness observed.
• Specific changes in tongue appearance correlate with diabetic conditions.

Conclusions:

• Early diagnosis of Type 2 Diabetes (DT2) is proposed using digital tongue images.
• The Greedy Snake Algorithm (GSA) effectively extracts edge and texture features from tongue images for diagnosis.