Validation of Functional Methods for Calculating Shoulder Joint Centers Using 3D Motion Analysis

Abstract

Research in upper body kinematics and kinetics requires accurate estimation of anatomical joints. Currently the use of regressive techniques using anatomical landmarks is the most common way of calculating upper limb joint centers. Research has shown that functional joint center methods can produce more accurate results than traditional regressive methods in the estimation of hip joint center. This paper investigates the use of functional methods for the estimation of the shoulder joint center using 3D motion analysis data. Three methods for calculating the functional joint center were tested: 1) a standard sphere fit regression, 2) a regression developed and tested for use finding the hip joint center (Piazza method) [1], and 3) a gradient method developed for this paper similar to the one used by Schonauer [2]. First the functional joint center methods were tested in MATLAB using data with random points rotating around a known joint center with varying amounts of noise. Using the MATLAB calculations the accuracy and repeatability of each method was analyzed. Functional joint centers were then calculated from two sets of motion analysis data. The first data set contained shoulder range of motion data, and the second set was gathered during activities of daily living (ADL). Both motion analysis sets used data collected from a healthy adult male subject using a Vicon motion analysis system. The repeatability of each method using the motion analysis data was then analyzed. The MATLAB tests show that the gradient method has the highest tolerance to noise in the data. Results from the motion analysis test show that, of the methods tested, no functional method was found to have consistent results for individual tasks. Each of the functional methods requires a range of motion not prevalent in most ADLs in order to generate a reliable joint center. Joint centers calculations improved in accuracy and reliability with a greater number of trials and larger range of motion. The functional methods are suitable for use in future studies that include a large range of motion.