Validation of video-assisted computer vision goniometry to measure shoulder abduction motor function

Introduction Goniometry is used to measure shoulder abduction range of motion aiding in diagnosis, rehabilitation planning and monitoring progress in rehabilitation evaluating a patient's shoulder function. Computer vision technology holds promising potential for the assessment of movement by unifying and objectifying goniometric studies of different somatometric parameters. The objective was to validate a video-assisted computer vision goniometry of the motor function of shoulder abduction using the potential of neural networks. Material and methods The study involved 33 volunteers, males and females aged 18 to 56 years, with the weight of 53 to 108 kg and the height of 155 to 195 cm. Measurements of related samples were compared to validate the author's method of goniometric examination of shoulder abduction. Classical goniometry was used for patients of group 1. Changes in the shoulder position were radiologically explored in group 2 and video-assisted goniometry computer vision was employed for examinations in group 3. The study was performed using hardware and software "Arthro-Pro" system. Statistical processing was produced using the Statgraphics software package. Results The average difference in the abduction was insignificant in groups 1 and 2 measuring (0.62 ± 0.63)° from a minimum of 5.2° to a maximum of 1° with confidence interval of p = 0.95. The difference in the abduction angle ranged from -11.8° to 22.7° in groups 1 and 3 with the average difference of 6° and confidence interval of p = 0.95. Discussion The minor difference in the abduction angles obtained with computer vision technologies and classical goniometry indicated the comparability of the two methods facilitating the possibility of introducing artificial intelligence for assessing musculoskeletal function in clinical practice. Conclusion The video-assisted computer vision goniometry is practical for measurements of shoulder abduction in clinical practice.