Comparison of computer-aided and manual measurements in the evaluation of carpal alignment.

The purpose of this study was to compare computer-aided analysis and different methods of manual measurements in the evaluation of carpal alignment. The radioscaphoid, radiolunate, radiocapitate and radiometacarpal angles were measured on cone-beam computed tomography (CT) scans of 30 healthy wrists by automated software (Disior Ltd.) and by hand surgeons using lateral radiographs reconstructed from the CT data. Hand surgeons were either given (n = 6) or not given (n = 7) prior instructions on how to perform the measurements. Inter- and intra-observer reliability of manual measurements ranged from good to excellent (intra-class correlation coefficients [ICC] 0.77-0.99), being highest in specialists with standardized methods and in reconstructed radiographs with bone overlap digitally removed. Computer-aided software provided excellent intra-observer reliability (ICC 0.94-1.00) consistently and values that were highly comparable (mean difference range 1°-7°) with the manual measurements made in optimal settings. Computer-aided software provides an accurate and repeatable method to measure carpal alignment in CT scans, minimizing observational errors.

Association of ulnar variance with three-dimensional carpal alignment and demographics in asymptomatic volunteers.

BACKGROUND: Several carpal pathologies are considered to be related to ulnar variance. Recently, computer-aided computed tomography (CT) analysis software was introduced to quantify three-dimensional (3D) carpal alignment with high accuracy and reliability. PURPOSE: To determine the association of ulnar variance with 3D carpal alignment and demographics. MATERIAL AND METHODS: A wrist of 121 asymptomatic volunteers (69 men, 52 women; mean age = 38 ± 10.4 years) was imaged in the neutral wrist position with cone-beam CT. Computer-aided CT analysis software (Bonelogic), based on segmentation and numerical modelling, was used to define ulnar variance and standardized 3D axes for all carpal bones. The association of ulnar variance with 3D carpal alignment, age of the volunteer, and side and dominance of the imaged wrist was assessed. RESULTS: The mean ulnar variance was -1.6 ± 1.5 mm (range = -5.3 to 2.4 mm). The mean ulnar variance was -1.9 mm and -1.1 mm in men and women (P = 0.007), respectively. Of the imaged 121 wrists, 18 (15%) had positive and 103 (85%) negative ulnar variance. There was no association between ulnar variance and any of the radio- or intercarpal angle values in either the sagittal or coronal plane (ρ = -0.16…0.17, r = -0.13….0.12). The ulnar variance showed no association with side (P = 0.51) or dominance (P = 0.27) of the imaged wrist. CONCLUSION: 3D carpal alignment is not affected by ulnar variance. The association of ulnar variance with sex may in part explain the difference in reported prevalence of some carpal pathologies, such as ulnar impaction syndrome and Kienböck's disease.

Three-dimensional carpal alignment: computer-aided CT analysis of carpal axes and normal ranges.

Assessment of carpal alignment traditionally uses carpal bone axes measured on lateral radiographs. In this study, three-dimensional axes were defined for carpal bones using segmentation and numerical modelling of CT data of 121 neutrally positioned, asymptomatic wrists. The geometric axis was used for radius, scaphoid and capitate, whereas the axis based on a line perpendicular to the articular surface was used for the other carpal bones. Normal values of radiocarpal angles in the radial coordinate and the reliability of the computer-aided analysis are reported. The mean sagittal radiocarpal angles (positive in palmar direction) were as follows: scaphoid 58° (SD 10°), lunate 0° (SD 11°), triquetrum 12° (SD 8°), trapezium 17° (SD 8°), trapezoid -10° (SD 7°), capitate -17° (SD 9°) and hamate 2° (SD 7°). The mean coronal radiocarpal angles (positive in ulnar direction) were -42° (SD 9°), -20° (SD 4°), -49° (SD 4°), -32° (SD 6°), -16° (SD 5°), 2° (SD 7°) and 8° (SD 6°), respectively. The intra-observer reliability of the measurements was excellent (mean intraclass correlations coefficient 0.98). This study provides guidelines on how to measure and quantify carpal alignment three-dimensionally, and a database for the normal values. Together, these may be useful when analysing various wrist pathologies and kinematics of the wrist.

Computer-based three-dimensional measurement of carpal alignment: measurement techniques and normal ranges.

The three-dimensional carpal alignment was measured in the neutral wrist position in 121 asymptomatic volunteers using computer-based cone-beam CT analysis. Normal values and the reliability of the automated analysis are reported. An analysis software based on segmentation of CT images and mathematical modelling was used to determine several axis variants based on different landmarks and to calculate the intercarpal angles automatically. Twenty wrists were imaged twice to determine intra-observer reliability. Mean values using the preferred axis variants were: scapholunate angle 57° (SD 9°), radiolunate angle 2° (SD 10°), lunocapitate angle -14° (SD 9°). Using alternate axis variants notably changed the angles produced. The intra-observer reliability of the analysis was excellent (mean intraclass correlation coefficient 0.97, SD 0.03). Computer-based CT analysis enables highly reproducible and automated assessment of carpal alignment. This study provides a reference database for measurement techniques and normal carpal angle values in three-dimensional imaging.

Radiographic measurements of the normal distal radius: reliability of computer-aided CT versus physicians' radiograph interpretation.

We examined the reliability of a computer-aided cone-beam CT analysis of radiographic parameters of 50 normal distal radii and compared it with interobserver agreement of measurements made by three groups of physicians on two-dimensional plain radiographs. The intra-rater reliability of the computer-aided analysis was evaluated on 33 wrists imaged twice by cone-beam CT. The longitudinal axis, anterior tilt, radial inclination and ulnar variance were measured. The reliability of computer-aided analysis was excellent (intraclass correlation coefficient (ICC) 0.94-0.96) while the interobserver agreement of two-dimensional radiograph interpretation was good (ulnar variance, ICC 0.80-0.84) to poor (anterior tilt and radial inclination, ICC 0.20-0.42). We conclude that computer-aided cone-beam CT analysis was a reliable tool for radiographic parameter determination, whereas physicians demonstrated substantial variability especially in interpreting the angular parameters.