A review of acupoint localization based on deep learning

Area of Science:

• Integrative Medicine
• Artificial Intelligence in Healthcare
• Medical Imaging Analysis

Background:

• Traditional Chinese Medicine (TCM) relies on precise acupoint localization.
• Conventional methods face limitations in accuracy and consistency.
• Deep learning (DL) offers a powerful solution for automated acupoint identification.

Purpose of the Study:

• To comprehensively review and analyze recent deep learning research for automatic acupoint localization.
• To examine principles, classifications, datasets, evaluation metrics, and applications of DL-based acupoint localization algorithms.
• To provide practical guidance for future research and clinical practice in TCM.

Main Methods:

• Systematic review and analysis of recent research on deep learning for acupoint localization.
• Categorization of algorithms by body part, architecture, localization strategy, and image modality.
• Identification and summary of key datasets and evaluation metrics.

Main Results:

• Deep learning algorithms significantly improve upon traditional and machine learning methods for acupoint localization.
• Categorization highlights the strengths, weaknesses, and optimal application scenarios for various DL approaches.
• Representative datasets and critical evaluation metrics are identified for standardized assessment.

Conclusions:

• Deep learning is revolutionizing acupoint localization, driving the modernization of TCM.
• The review offers a valuable resource for researchers and practitioners, outlining current capabilities and future directions.
• Standardized evaluation and further development of DL models promise enhanced TCM efficacy and accessibility.