A mathematical model for measuring scapula motion, validated with a fluoroscopy image matching process
Date
2015
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
The primary goal of this study was to develop and validate a new clinically applicable approach to measuring scapulothoracic kinematics among typically developed individuals. Accurate assessment of scapulothoracic kinematics is important for understanding functional limitations of the shoulder and upper extremity. Current efforts to measure shoulder function often provide no differentiation between scapulothoracic and glenohumeral joint contributions to shoulder motion. Techniques that attempt to quantify these measures are often inaccurate, invasive, time consuming, induce large amounts of radiation exposure, and/or require the subject to lie horizontally in an enclosed space. Traditional 3D motion capture has been used effectively to measure kinematics of bones that move with the skin, such as the humerus, but has failed to provide accurate estimates of scapula kinematics. We believe, however, that 3D motion capture data can be used in an innovative way to estimate scapula kinematics. We hypothesized that individualized mathematical algorithms that estimate dynamic scapula orientation based on humeral orientations and/or acromion process (AP) displacements collected in multiple static poses can be developed using motion capture data. The mathematical algorithms were validated using a technique recently established by our lab that utilized bone models and biplane fluoroscopy to perform a 2D to 3D matching process. This process was time consuming and involves some radiation exposure, and, therefore is not clinically applicable, but will provide a standard for validation of the mathematical algorithms. The algorithms that result from this work are expected to provide conceptual evidence that a clinically useful tool for measuring scapulothoracic kinematics may be adaptable to patient populations and used to drive musculoskeletal modeling. Individualized multiple linear regression algorithms and artificial neural networks were developed for nine healthy adult shoulders. Algorithms were used to predicted scapulothoracic kinematics, and the predicted kinematics were compared to the fluoroscopy obtained kinematics to determine the accuracy and clinical practicality of the algorithms. Results showed strong correlations between mathematically predicted scapulothoracic kinematics and the validation kinematics. Clinically, both mathematical approaches estimated kinematics within 10 degrees of actual kinematics. As a result, both multiple linear regression and artificial neural networks showed promise for being accurate, non-invasive, techniques for measuring scapulothoracic kinematics in a clinical setting.