Estimating Jaw, Neck, and Shoulder Range of Motion Using an AI Model

Due to the involvement of various structures, patients with oral cancer may experience functional impairments after treatment, such as trismus, spinal accessory nerve dysfunction, neck dystonia, radiation fibrosis, and fatigue. This observational study aims to develop an AI-based system for tracking mandibular and shoulder movements using deep learning techniques. AI-generated pose estimations will be compared with gold standard measurements: maximal mouth opening will be compared with caliper measurements, and Therabilte scale, while shoulder abduction range of motion will be compared with universal goniometer measurements. We will recruit 20 healthy adults and 20 oral cancer patients. Data on maximal mouth opening and shoulder abduction will be collected through video recordings, calipers, Therabilte scale, and universal goniometers. The videos will be analyzed using deep learning to estimate mouth opening and shoulder abduction angles. These estimates will then be compared with the gold standard measurements. The Intraclass Correlation Coefficient (ICC), Mean Absolute Error (MAE), and Coefficient of Variation (CV) will be used as performance indicators to assess and compare the reliability, accuracy, and consistency of the models.

Inclusion criteria

• Healthy adults without a history of head, neck or shoulder injury or surgery, and without HNC-related radiotherapy or chemoradiotherapy
• Oral cancer patients with trismus, clinical signs of neck or shoulder joint impairment after oral cancer surgery or radiotherapy
• Age between 20 and 65 years

Exclusion criteria

• Could not communicate
• Had any disorder that could influence movement performance