Introduction: Classical studies have defined axes from prominent scapular landmarks that have been used to synthesise many applications. The morphology of the scapula is however known to be highly variable between individuals1,2,3. This introduces significant variability on the use of these classical axes for various clinical applications. Also, some of the literatureapplied landmarks were highly dependant on the presence of pathology, thus introducing more variability in the products they parented. This limits accuracy in inter-subject comparisons from such applications. Therefore there is a need to identify and define pathology-insensitive anatomical landmarks that are less variable between individuals than the variability of the overall scapular shape. The aim of this study was to define more scapular axes from clearly identifiable landmarks, analysing these and other classical definitions for the best axis that minimizes variability and is closely related to the scapular clinical frame of reference.

Materials and Method: Fourteen different axes of new and classical definitions from clearly identifiable landmarks were quantified by applying medical images of 21 scapulae. The orientations of the quantified axes were calculated. The plane of the blade of the scapula was defined, bounded by the angulus inferior4, the spine/medial border intersection5 and the most inferolateral point of the infra-glenoid tubercle. This was applied to grade the alienation of the quantified axes from the scapular blade. The angular relationships between individual axes of a spcapula were quantified, averaged over the 21 specimens and their standard deviations (SD) applied to grade the sensitivity of each axis to interscapular variations in the others. The volume of data required to define an axis (VDA) was noted for its dependency on pathology. These three criteria were weighted according to relative importance such that

axes bearing 10° or more from the blade deviated significantly and were eliminated;

Results: A least square line through the centre of the spine root was the most optimal medio-lateral axis. The normal to the plane formed by the spine root line and a least square line through the centre of the lateral border ridge was the most optimal antero-posterior axis.

Conclusion: These body-fixed axes are closely aligned to the cardinal planes6 in the anatomical position and thus are clinically applicable, specimen invariant axes that can be used in generalised and patient-specific kinematics modelling.

Many different clinical examinations are used to assess instability of the glenohumeral joint. Validation of these includes clinical data, follow-up, imaging, and arthroscopy. In spite of these many works, there currently exists no clear unique method for identifying and validation novel clinical examinations. The aim of this study was to use a computational tool to quantify the specificity of clinical examinations in assessing glenohumeral ligament (GHL) pathology. Five GHLs were modelled according to the literature [1]. Physiological kinematics data [2] were applied to simulate 23 clinical examination manoeuvres for the glenohumeral joint. Individual ligament forces were computed as a percentage of the total ligamentous restraint. The 0° abduction anterior-laxity test was most specific for the superior GHL (82.3%). The anterior apprehension in the coronal plane and 90° anterior-laxity tests were specific for the anterior band of the inferior GHL (abIGHL – 100%). Pure scapular plane abduction and the 90° abduction inferior-laxity tests were specific for the axillary pouch of the IGHL (apIGHL – 100%, 89.6%). Specific tests for posterior band of the IGHL were posterior apprehension (95.1%), 0° and 20° abduction posterior-laxity, and 30° to 45° flexion Norwood and Terry test (100% each). The middle GHL did not exhibit any exclusive loading pattern for any of the tests. A secondary insertional morphology was simulated with the abIGHL positioned at the 4 o’clock position as opposed to the 3 o’clock position [1,2,3]. Significant loading differences were computed for the same ligaments during the same tests. This study demonstrates the sensitivity of specific tests for individual GHLs, but provides the significant caveat that ligament loading is significantly influenced by normal anatomical variations.