Three-Dimensional Reconstruction and Quantification of Hip Joint Cartilages from Magnetic Resonance Images

Abstract

A method for automated determination of the distribution of hip joint cartilage thickness from in vivo magnetic resonance (MR) images is described. Three-dimensional (3-D) filtering techniques are combined with the shape constraint of the ball and socket constitution of the hip joint to accurately detect and quantitate thin structures of cartilages. The method consists of three steps: 3-D cartilage enhancement filtering, detection, and quantification. First, the cartilage and articular space regions in MR images are effectively enhanced by 3-D radial directional second-derivative filtering based on the sphere approximation of the hip joint. Next, the initial regions of the cartilages and articular space are detected from the filtered images using thresholding and connectivity analysis. The boundaries of these regions provide the initial descriptions for the subsequent refinement processes. Finally, the boundaries of cartilage regions are accurately localized and quantitated three-dimensionally through subpixel edge searching. The effect of partial voluming on the accuracy of the estimated thickness is evaluated by means of software simulation studies, and the usefulness of the method is demonstrated through experiments using in vivo MR images of a normal volunteer and actual patients.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.