Tongue color analysis for medical application

Area of Science:

• Medical Informatics
• Computer Vision
• Diagnostic Systems

Background:

• Tongue color analysis is a traditional diagnostic method.
• Objective and systematic tongue color assessment is lacking.
• Digital image processing offers potential for standardized analysis.

Purpose of the Study:

• To develop and validate a systematic tongue color analysis system for medical applications.
• To establish a quantitative relationship between tongue color features and health status.
• To explore the potential of tongue color analysis for disease classification.

Main Methods:

• Development of a tongue color analysis system using a defined color gamut.
• Extraction of tongue foreground pixels and assignment to 12 representative colors.
• Calculation of color ratios to form a tongue color feature vector for classification.

Main Results:

• Achieved an average classification accuracy of 91.99% for distinguishing healthy from diseased tongue samples.
• Successfully clustered disease samples into three groups, enabling differentiation of multiple illnesses within clusters.
• Attained classification rates greater than 70% for 11 specific illnesses, highlighting diagnostic potential.

Conclusions:

• The proposed tongue color analysis system provides a systematic and accurate method for medical diagnosis.
• Tongue color features are demonstrably linked to the physiological state of the human body.
• This approach shows promise for non-invasive disease detection and classification.