This study was approved by the Institutional Review Board and informed consent was obtained for each subject. Five right dominant upper extremities without a history of previous injury from right-hand dominant male volunteers (age 24 to31) were imaged while in neutral forearm rotation and elbow extension by computed tomography (GE Light Speed Pro 16-Slice scanner). Parallel axial CT images separated at 0.625mm with a resolution of 512 × 512 pixels were obtained. Each CT image was processed using a Canny filter programmed in a commercially available mathematics software package (Matlab, Mathworks, Canton, MA). The Canny filter calculates gradients in pixel intensity to detect edges between objects. The calculated edges were used to help trace the outlines of the radius and ulna within the axial plane image using solid-modeling software (Rhinoceros®, Robert McNeel & Associates, Seattle, WA). The contours were then placed in the appropriate plane in a 3D space, and the surface was meshed using the solid-modeling software.
After CT image-based computer 3D models were constructed, each subject was imaged using two orthogonally placed fluoroscopes (BV Pulsera, Philips, Bothell, WA) at elbow flexion angles of 0°, 45°, and 90° as the subject rotated his/her forearm from maximum supination, 50% supination, neutral, 50% pronation, and maximum pronation (). The wrist was placed in zero degrees of extension and zero degrees of radioulnar deviation and the metacarpalphalangeal, proximal interphalangeal, and distal interphalangeal joints were held in extension. These positions were confirmed with goniometer. Each subject practiced the testing positions five times before actual testing. The degree of prononation/supination was estimated from the pre-conditioning trials with the understanding that the true amount of rotation would be defined with the modeling techniques. During imaging, the plane of the C-arm was co-planar to the floor and the subject was seated with the elbow stabilized on a radiolucent table (). The subject held each position which was confirmed with a goniometer and then the image was acquired. The entire testing procedure took about 10 minutes.
Figure 1 Dual fluoroscopic imaging system. Subject with elbow at 45° flexion and forearm in neutral rotation. Images were captured at elbow flexion angles of 0°, 45°, and 90° with the forearm at maximum supination, mid-supination, (more ...)
The orthogonal images were then imported into a solid modeling software (Rhinoceros®, Robert McNeel & Associates, Seattle, WA Rhinoceros, Seattle, Wash) and used to determine the in vivo
forearm positions at each of the targeted flexion angles. The orthogonal images were placed in the software to reproduce the positions of the two intensifiers of the fluoroscope during image acquisition. The forearm model was imported into the virtual space and was viewed simultaneously from two orthogonal directions corresponding to the positions of the x-ray source of the fluoroscope during image acquisition. The 3D forearm model was then manipulated in six degrees of freedom inside the 3D C-arm until its projections viewed from two orthogonal directions matched the outlines of the fluoroscopic images obtained, thus reproducing the in vivo forearm position using the 3D anatomical models. The method has been previously validated on a knee model to an accuracy of 0.1 mm position and 0.1 degrees rotation (6
To describe the six degrees of freedom (6DOF) kinematics of the distal radioulnar joint (DRUJ), anatomically based ulnar and radial Cartesian coordinate systems similar to that reported previously in the literature were constructed for each subject (). The position of the forearm in neutral position during CT scanning was used as a reference. For the ulnar coordinate system, the longitudinal axis (z-axis) was defined as the line passing through the longitudinal axis of the ulna. The y-axis was defined as the line passing from the center of the ulnar head to through the center of the ulnar styloid base. The x-axis was an axis perpendicular to those two axes pointing dorsally. The radial longitudinal axis (z-axis) was the long-axis of the radius. The y-axis was defined as the line passing through the anatomic center of the distal radius to the tip of the radial styloid. The x-axis was an axis perpendicular to those 2 axes pointing dorsally. In this manner, a coordinate system for each wrist could be established in a consistent way. The axes used to describe the radius are similar to those used clinically to describe the forearm.
The coordinate systems used to describe radioulnar kinematics. The ulnar y-axis, a line the center of the ulnar head to the center of the ulna styloid, was used to delineate supination from pronation.
The 6DOF kinematics of the wrist was described by the relative position and orientation of the distal radius with respect to the ulnar head using a script written by our laboratory for Rhinoceros solid modeling software. The three-dimensional positions were determined by the position of the origin of the distal radius coordinate system in the ulnar head coordinate system. The orientation was represented by the relative orientation of the distal radius coordinate system with respect to the ulnar styloid coordinate system using 3 Euler angles (in x-y-z sequence) (). In this study, 6DOF was expressed using the radial displacements along the x,y, and z-axes of the ulna for anterior/posterior, ulnar/radial, and proximal/distal displacements and rotation about the z-axis for pronation/supination. Proximal translation of the radius relative to the ulna was defined as negative, whereas distal translation was positive.
Figure 3 Change in ulna variance was measured from the neutral position for each angle of elbow flexion. Ulna variance was measured as the distance along the ulnar z-axis between the center of the radial coordinate system and the center of the ulnar coordinate (more ...)
In order to characterize the overall effect of flexion and rotation at a specified elbow flexion angle, we examined the difference observed in proximal-distal translation between full supination and full pronation (terminal difference) versus the difference observed between the peak proximal and peak distal translation (peak difference).
To compare the actual path of rotation of the radius to the expected path of a perfect circle, we defined a circular path about the center of the ulnar coordinate system using the center of the radial coordinate systems and the “best fit circle” function within Rhinoceros (). Lines extending from the center of the ulnar coordinate system were extended to the centers of the radial coordinate systems such that they intersected with the path of the best fit circle. The difference in distance between the actual and expected path of rotation were then measured along this line.
Center of rotation was determined by finding the best fit circle of the center points of the radial coordinate system. Deviations from this circular path were random and minor.
Because of the small sample size of the cohort, significance was pre-defined using an alpha of 0.10. Analysis of Variance (ANOVA) was used to compare the motion of the forearm among the different elbow flexion and rotation angles with Statistica 6.0 (StatSoft, Tulsa OK). Post-hoc testing using a Neuman Keul’s test was performed to evaluate differences between groups.