Abstract
Background
Hip-joint shape is an important factor that affects an individual's risk of developing osteoarthritis (OA). However, validated tools to quantify these morphological characteristics on clinical plain radiographs are few. We developed a Matlab-based Semi-automated HIP Software programme (SHIPS) that measures 10 morphologic characteristics of the hip that are risk factors for OA using a plain digitised antero-posterior pelvic radiograph. In this study we validated the accuracy and repeatability of this freeware-tool.
Methods
Software accuracy was assessed using a test pelvic radiograph, and by repeated measurements of an AP-pelvic radiograph digitally recreated from pelvis computed-tomography images reformatted to create images rotated in 2-dimensions (2.5° increments, range −15° to +15°). Intra- and inter-observer repeatability was assessed using pelvic radiographs from 30 subjects analysed twice using the software by two readers, and expressed as coefficient of variation (CV). Clinical-repeatability was assessed by measuring sequential pelvic radiographs taken on the same day after re-positioning in 23 subjects.
Results
Software accuracy was within 0.1% for linear-ratios and 0.4° for angular-measurements. Changes in pelvic inclination and rotation of ±15° resulted in <14% change in linear-measurement ratios and <7° change in angular-measurements. The intra-observer CV was between 0.3 to 4.1%, and inter-observer CV 0.7 to 9.7% with the exception of horizontal-toit-externa (HTE, 14.6 and 24.2% respectively). Short-term clinical-repeatability varied from 0.4 to 6.1%, with the exception of HTE (37.4%).
Conclusion
The software showed good accuracy and repeatability for measurement of most hip-joint morphologic risk factors for OA apart from HTE. This tool has particular value in studying large or retrospective datasets where cross-sectional imaging is not feasible or available.
Get full access to this article
View all access options for this article.