AI-Guided Ultrasound Needling of Xinwu Point for Allergic Rhinitis

Overview

This study evaluated a new way to perform acupuncture at a deep point called the Xinwu point (sphenopalatine ganglion) for allergic rhinitis (hay fever-like nasal allergy). The point is hard to reach with the usual blind needling method. Artificial intelligence (AI) with real-time ultrasound images was used to help the acupuncturist see nearby structures and guide the needle. Sixty adults were randomly assigned to AI-guided needling or usual blind needling. Both groups received the same acupuncture schedule: both sides of the face, once a week, for six sessions. The primary aim was to measure how accurately the needle reached the target. Side effects, nasal symptom scores, overall improvement, and how acceptable participants found the procedure were also recorded.

This was a single-center, parallel-group, randomized controlled trial conducted at Daxing Teaching Hospital of Capital Medical University (Beijing Daxing District People's Hospital), China, from September 2025 to May 2026. Participants: 60 adults aged 18-60 years with allergic rhinitis diagnosed according to Chinese clinical guidelines were randomized 1:1 to an experimental group (AI-assisted ultrasound-guided needling) or a control group (conventional blind needling), 30 per group. All needling was performed by the same acupuncturist. Intervention: Both groups received bilateral acupuncture at the Xinwu point (sphenopalatine ganglion) once every 7 days for 6 sessions. In the experimental group, a low-frequency ultrasound probe with an AI image-recognition program labeled anatomical landmarks (including the condylar process, coronoid process, neurovascular bundle, maxilla, and lateral pterygoid plate) in real time, and the needle was advanced in-plane to the pterygopalatine fossa. In the control group, the same needle type and success criteria were used without ultrasound or AI guidance. Primary outcomes: needling success rate, first-pass success rate, and mean number of punctures. Secondary outcomes: adverse-event rate, Total Nasal Symptom Score (TNSS), total effective rate based on TNSS improvement, and scores on a self-designed patient-acceptance scale. This registration is retrospective because enrollment and follow-up were completed before the record was submitted.