An Acupoint Detection Approach for Robotic Upper Limb Acupuncture Therapy

Abstract

Acupuncture is one of the most effective traditional Chinese therapies, which emphasizes that the intensity of acupuncture must reach a threshold to generate “De-Qi”, which is necessary to achieve the best therapeutic effect. To achieve the De-Qi status, microneedles used for acupuncture have to be penetrated into some specific key locations, called acupoints, for different diseases. During acupuncture therapy, doctors/experts are required to find the acupoints and treat them one by one manually and repeatedly. To lower the burden of these precious experienced experts while maintaining effectiveness and standardizing the therapy process, the development of robot-assisted automatic manipulation systems is necessary. To perform robot-assisted acupuncture therapy, a visual-based acupoint detection process should be realized firstly, however, it is not an easy task for conventional vision-based feature detection methods due to the varying conditions of human bodies. In this paper, as an instance, a deep learning-based approach is developed to perform upper limbs acupoints detection.