Related Experiment Video
Updated: Jun 30, 2026
08:18
Visualizing Motion Patterns in Acupuncture Manipulation
Published on: July 16, 2016
9.1K
MetaAcuPoint: MetaHuman-Generated Synthetic Data for Hand Acupoint Localization.
Kasunika Guruge1, Prathiksha Padmanabha1, H M K K M B Herath1
1Industry 4.0 Convergence Bionics Engineering, Pukyong National University, Busan 48513, Republic of Korea.
Healthcare (Basel, Switzerland)
|December 11, 2025
Summary
This study introduces MetaAcuPoint, a synthetic dataset for precise acupoint localization in Traditional Korean Medicine. The synthetic data enables accurate, standardized acupoint identification, overcoming limitations of traditional methods.
Area of Science:
- Biomedical Engineering
- Medical Imaging
- Traditional Korean Medicine
Background:
- Precise acupuncture point (acupoint) localization is vital for Traditional Korean Medicine (TKM) clinical success.
- Traditional acupoint identification methods are subjective, hindering standardization.
- Existing data-driven techniques face challenges due to limited, inconsistently labeled datasets.
Purpose of the Study:
- To introduce MetaAcuPoint, a novel synthetic dataset for high-fidelity acupoint localization.
- To provide anatomically consistent hand images for overcoming dataset limitations in TKM research.
- To facilitate standardized, data-driven acupoint identification.
Main Methods:
- Generated 900 RGB hand images using MetaHuman avatars in Unreal Engine.
- Implemented precise, bone-attached sockets for five hand acupoints with millimeter accuracy.
- Assessed dataset validity by training a High-Resolution Network (HRNet-W48) and testing on real-world images.
Main Results:
- Synthetic-trained model achieved a mean distance error (MDE) of 5.67 pixels, comparable to real-data baseline (4.81 pixels).
- Synthetic data improved performance when added to real data (MDE: 4.95 pixels).
- Synthetic data-trained models generalized better to external datasets than real-data-trained models (MDE: 5.84-6.45 mm vs. 10.63-15.80 mm).
Conclusions:
- MetaAcuPoint represents the first synthetic-to-real generalization for hand acupoint localization.
- The dataset combines photorealistic rendering with anatomically grounded annotation.
- MetaAcuPoint offers a reliable resource for advancing standardized, data-driven acupuncture research and practice.
Keywords:
MetaHumanacupoint localizationannotation consistencydeep learningsynthetic datavirtual realityMore Related Videos
Related Concept Videos
01:09Bones of the Upper Limb: Radius
The radius is longer of the two bones that make up the human antebrachium or forearm. At the proximal end, the radius articulates with the capitulum of the humerus and the radial notch of the ulna to form the elbow joint. At the distal end, the radius articulates with the ulna via the ulnar notch, forming the distal radioulnar joint. Distally, the radius also attaches to the carpal wrist bones (scaphoid and lunate) to form the radiocarpal joint.
The radius has a nail-shaped head, and a short...
The radius has a nail-shaped head, and a short...
01:16Muscles of the Forearm that Move the Hand and Fingers
The muscles of the forearm that move the wrist, hand, and digits are numerous and diverse. They can be classified into two groups based on their location and function — the anterior and posterior compartment muscles.
Anterior Compartment
The anterior compartment muscles originate from the humerus. They primarily function as flexors and are also known as flexor muscles. They typically insert on the carpals, metacarpals, and phalanges. The superficial layer includes the flexor carpi radialis,...
Anterior Compartment
The anterior compartment muscles originate from the humerus. They primarily function as flexors and are also known as flexor muscles. They typically insert on the carpals, metacarpals, and phalanges. The superficial layer includes the flexor carpi radialis,...