Three-dimensional measurement of the hip joint space from clinical CT for the assessment and prediction of osteoarthritis

Advances in our understanding of osteoarthritis have been hampered by an ongoing inability to detect relevant structural changes in vivo. Radiography, which remains the gold standard for disease assessment, is limited by its two-dimensional representation of the whole joint and the need for observer-generated grading or measurement of disease. The aim of this study was to develop and validate a novel algorithm for three-dimensional (3D) measurement of the hip joint space width (JSW) with clinical CT data.

Right hips were dissected from 20 female cadavers, all with ante-mortem informed consent for participation in post-mortem imaging research. Each hip was imaged with clinical CT (test set) and high resolution peripheral quantitative CT (validation set). An automated model-based deconvolution approach was then applied to measure JSW in 3D from clinical CT data. This study was approved by the Cambridge Human Biology Research Ethics Committee.

Each hip joint was processed with a pipeline of automatic sphere fitting to the femoral head, manual 3D segmentation of joint space limits on the surface of this sphere, and subsequent automatic JSW measurement with a technique called joint space mapping. With a model-based approach, JSW was measured as the distance between opposing subchondral bone edges after deconvolution of the clinical CT data. Measurement validation was performed after cross-registration with the corresponding high resolution data, using a thresholding technique to determine JSW in this data set. Joint space mapping of clinical CT data was both technically accurate and precise over a 95% range of high resolution JSW values, from 1·48 to 4·61 mm (mean bias +0·02 mm [SD 0·43]).

Joint space mapping is a new image analysis technique that can automatically measure hip joint space in 3D from clinical CT data with high accuracy and precision, subsequently building a 3D model of the joint itself. This technique could deliver a step-change both in research trials and in the clinical setting, where the ability to stratify individuals for therapy and then evaluate disease accordingly is essential for prognostication and monitoring treatment efficacy over time.

Wellcome Trust.