Proximal humeral coordinate systems can predict humerothoracic and glenohumeral kinematics of a full bone system

Clinical imaging often excludes the distal humerus, confounding definition of common whole-bone coordinate systems. While proximal anatomy coordinate systems exist, no simple method transforms them to whole-bone systems. Their influence on humeral kinematics is unknown.

How do humeral kinematics vary based on proximal and whole-bone coordinate systems, and can average rotation matrices accurately convert kinematics between them?

Three proximal coordinate systems were defined by the lesser and greater tuberosities (LT, GT), Crest of the greater tuberosity, and humeral shaft. Average rotation matrices derived from anatomic landmarks on cadaver humeri were generated between the proximal and whole-bone coordinate systems. Absolute angle of rotation was used to determine if anatomical variability within the cadaver population influenced the matrices. The matrices were applied to humerothoracic and glenohumeral motion (collected previously) and analyzed using the proximal coordinate systems, then expressed in the whole-bone system. RMSE was used to compare kinematics from the proximal and whole-bone systems.

A single average rotation matrix between a given proximal and whole-bone coordinate system achieved consistent error, regardless of landmarks. Elevation and plane of elevation had <2° mean error when proximal coordinate systems were transformed to whole-bone kinematics. Axial rotation had a mean 7° error, primarily due to variable humeral head retroversion. Absolute angles of rotation did not statistically differ between subgroups. The average rotation matrices were independent of sex, side, and motion.

Proximal humerus coordinate systems can accurately predict whole-bone kinematics, with most error concentrated in axial rotation due to anatomic twist along the bone. These results enhance interpretability and reproducibility in expressing humerothoracic and glenohumeral motion data between laboratories by providing a simple means to convert data between common coordinate systems. This is necessitated by the lack of distal humerus anatomy present in most clinical imaging.