Table 1.
Summary of models with scapula mechanics.
Published models are listed with their modeling approach along with their suitability for real-time analyses and their accessibility for modification and re-use by other researchers and clinicians.
Fig 1.
The four modeled degrees of freedom of the scapulothoracic joint.
The joint reference frame on the scapula (axes X,Y,Z) is used to locate the scapula with respect to the thorax. The joint reference frame on the scapula is computed according to the ISB recommendations [18] (shown as XS, YS, ZS), however our joint origin is located at the centroid of the anatomical markers used to define the joint frame instead of the Angulus Acromialis and its axes are rotated -90° about Y (to enable positive upward rotation about Z). The joint frame on the thorax defines the center of the scapulothoracic surface modeled as an ellipsoid (red shaded surface). Abduction (adduction) followed by elevation (depression) locate the joint frame origin of the scapula (blue) on the ellipsoid fixed to the parent thorax body (green). The scapula rotates upward (downward) about the normal to the surface (scapula Z-axis). Internal rotation or “winging” is a positive rotation about the Y-axis of the joint frame in the scapular plane, which remains tangent to the thoracic surface.
Table 2.
Marker errors between model reconstruction and bone-pin experiments (RMS in mm) from three shoulder activities.
The mean error across all markers and the worst marker for a task are also provided. Scapula markers include Angulus Acromialis (AA), Angulus Inferior (AI) and Trigonum Spinae (TS). Thorax markers include Incisura Jugularis (IJ), the spinous process of the seventh cervical (C7) and eighth thoracic (T8) vertebra. Humerus markers include the glenohumeral joint center (GH) and the lateral (EL) and medial (EM) epicondyles. Marker definitions from [18].
Fig 2.
Scapula and Scapulothoracic joint kinematics during shoulder flexion, abduction, and rotation tasks.
Scapular kinematics described by the relative rotation of the scapula with respect to the thorax expressed as a body-fixed Y-X-Z Euler angle sequence according to the ISB standard (left panel: Y-internal rotation, solid; X-downward rotation, dashed; Z-posterior-tilting, dotted) and the scapulothoracic joint coordinates (right panel) with abduction (black solid), elevation (dashed), upward rotation (gray solid), and internal rotation or winging (dotted) reconstructed motion from measured bone-pin marker locations during shoulder tasks of: flexion, abduction, and rotation at 90° of humerus abduction.
Fig 3.
Mean and standard deviation of root-mean-squared errors (RMSE) of scapular kinematics in the presence of noise compared to noise-free kinematics.
Scapular kinematics were computed with and without the scapulothoracic joint model during shoulder (humeral) flexion, abduction and rotation tasks. At every noise level, the model (green) and, in particular, the use of the scapulothoracic joint model coordinates (black), reduces RMSE by over 65% compared to direct scapula Euler angle calculations from markers (red). Standard deviations of Euler angles and joint coordinates are indicated by vertical bars and gray shading, respectively. The horizontal dotted line at 4.7° indicates where errors in scapular angles would result in an inability to distinguish the movement between different subjects (Bourne et al. 2011).
Fig 4.
Scapulothoracic generalized coordinate forces (Nm) during shoulder flexion, abduction and rotation at 90° of shoulder abduction tasks.
Scapula abduction (bold), elevation (dashed), upward rotation (gray), and internal rotation (dotted) generalized torques computed from an inverse dynamics analysis. A large sustained torque is required to keep the scapula elevated against gravity and requires additional upward rotation torque (gray) to rotate the scapula and lift the humerus during arm elevation tasks.
Table 3.
Summary of computational speed and accuracy of scapulothoracic joint mechanics.
Performance is presented as a ratio of the real movement duration to the computation period, where a result > 1 represents a factor faster than real time. Computation times evaluated from single threaded calculations on an i7-2820QM 2.4GHz processor.