gms | German Medical Science

Objectives: The complexity of the shoulder joint has long been a difficulty when trying to investigate clinical problems. This is more evident when attempting to simulate an active, muscle-controlled shoulder motion as it requires knowledge about the activity ratios of the muscles involved. Since the rotator cuff (RC) cannot be completely imaged with surface electromyography (EMG) [1], using musculoskeletal modelling to estimate muscle and joint loading during upper limb motion could be advantageous. Therefore, the aim of this study was to develop and validate a reliable musculoskeletal shoulder model, based on newly acquired EMG data, in order to determine muscle activity ratios during different movements.

Methods: Sixteen healthy subjects were measured during six motions of the upper limb in the scapular plane (abduction/adduction, anteversion/retroversion, external/internal rotation). Joint motion was measured with an inertial motion capture system (Xsens Technologies B.V., An Enschede, The Netherlands) and EMG data of the deltoid (anterior, middle and posterior) and infraspinatus muscles (Zebris Medical GmbH, Isny, Germany) was acquired for validation of the musculoskeletal shoulder model. The model was created with AnyBody Modeling Software (AnyBody Technology A/S, Aalborg, Denmark), including the deltoid and RC muscles and using the subject-specific motion data from the Xsens sensors. Pearson's correlation of EMG activity and AnyBody muscle activity was used for validation. In order to calculate the ratio of muscle activation, a main/primary muscle was defined for each movement. The activity of the remaining/secondary muscles was defined as a percentage in relation to the main muscle.

Results and conclusion: The model showed moderate to strong correlation with the EMG data (r = [0.38; 0.51]) and thus the muscle activities calculated with the model were subsequently used to determine the muscle ratios. The middle deltoid was defined as the main muscle acting during abduction/adduction (Figure 1 [Fig. 1]).

During retroversion, the middle deltoid had a higher activity than the anterior deltoid, and the posterior deltoid exhibited a lower activity than both muscles.

A muscle model of the shoulder was successfully created based on subject data. The musculoskeletal model can thus calculate the activity of muscles that would not be accessible with a surface EMG. With the knowledge of the ratios of individual muscles of the RC during certain movements, a more accurate in-vitro simulation of the shoulder motion for investigating clinical questions can be achieved.