Validity and reliability of automatic range of motion measurement using the elbow joint photograph

Abstract

Background: Photographic-based arc of motion measurement methods required human assessors and its accuracy is depend on observer experience.

Objectives: Current research proposed method of using digital image processing technique (DIPT) for measuring elbow range of motion.

Methods: Participants were enrolled from students and staffs in the university. Fluoroscopic images of both elbows were taken in flexion and extension positions. Two physiotherapists performed goniometer, inclinometer and smartphone gyroscope range of motion (ROM) measurement on bilateral elbows. Photographer took elbow images in fully extension and fully flexion three times for each position with 8-megapixel smartphone camera. The extension and flexion angles were calculated using DIPT protocol. Intra-rater reliability and inter-rater reliability of all methods were assessed using intraclass correlation coefficient (ICC). Paired student t test and Wilcoxon signed rank test were used to detect systematic bias. Bland-Altman plot was utilized to show possible range of difference between methods.

Results: There were total 56 elbows. Intra-rater and inter-rater ICCs of fluoroscope, goniometer, inclinometer, and gyroscope showed moderate to excellent agreement. Extension and flexion score of fluoroscopic images were higher than DIPT results. Mean extension and flexion angle of DIPT group was higher than goniometer, inclinometer and gyroscope group (P < 0.05), but total ROM were equaled, (vs goniometer P = 0.322, vs inclinometer P = 0.534, vs gyroscope P = 0.899). Limit of agreement of extension angles, flexion angles and total ROMs were 9.93-13.32, 9.81-12.66 and 13.84-16.66 degrees respectively.

Conclusions: Elbow ROM measurement from current DIPT protocol had comparable result with flexion agnle of fluoroscopic images and flexion-extension angle of goniometer, inclinometer and gyroscope, but it can be difference from other reference methods up to 16 degree. Further study and protocol adjustment are needed to improve accuracy of the image analytic technique.