Deep Convolutional Neural Networks for Classifying Body Constitution Based on Face Image.

Area of Science:

• Integrative Medicine
• Artificial Intelligence in Healthcare
• Traditional Chinese Medicine (TCM)

Background:

• Body constitution classification is fundamental to Traditional Chinese Medicine (TCM) research.
• Current identification methods, such as questionnaires, are inefficient and lack accuracy.
• A need exists for objective and efficient methods to classify body constitutions.

Purpose of the Study:

• To develop and evaluate a novel algorithm for body constitution recognition based on facial images.
• To improve the accuracy and efficiency of body constitution classification using deep learning techniques.
• To provide an objective tool for TCM constitution research.

Main Methods:

• A deep convolutional neural network (CNN) was employed for feature extraction from facial images.
• Color features were combined with CNN-extracted features.
• A Softmax classifier was utilized for final constitution type classification.
• The algorithm was validated through comparative experiments.

Main Results:

• The proposed deep convolutional neural network algorithm achieved an accuracy of 65.29% for body constitution classification.
• The performance of the algorithm was deemed acceptable by traditional Chinese medicine practitioners.
• The method demonstrated improved efficiency compared to traditional identification techniques.

Conclusions:

• Deep convolutional neural networks offer a promising approach for objective body constitution classification using facial images.
• The developed algorithm provides a more accurate and efficient alternative to traditional methods in TCM constitution research.
• Facial image analysis combined with deep learning holds potential for advancing TCM diagnostics.