Image-assisted non-invasive and dynamic biomechanical analysis of human joints

Kinematic analysis provides a strong link between musculoskeletal injuries, chronic joint conditions, treatment planning/monitoring and prosthesis design/outcome. However, fast and accurate 3D kinematic analysis still remains a challenge in order to translate this procedure into clinical scenarios. 3D computed tomography (CT) to 2D single-plane fluoroscopy registration is a promising non-invasive technology for biomechanical examination of human joints. Although this technique has proven to be very precise in terms of in-plane translation and rotation measurements, out-of-plane motion estimations have been a difficulty so far. Therefore, to enable this technology into clinical translation, precise and fast estimation of both in-plane and out-of-plane movements is crucial, which is the aim of this paper. Here, a fast and accurate 3D/2D registration technique is proposed to evaluate biomechanical/kinematic analysis. The proposed algorithm utilizes a new multi-modal similarity measure called 'sum of conditional variances', a coarse-to-fine Laplacian of Gaussian filtering approach for robust gradient-descent optimization and a novel technique for the analytic calculation of the required gradients for out-of-plane rotations. Computer simulations and in vitro experiments showed that the new approach was robust in terms of the capture range, required significantly less iterations to converge and achieved good registration and kinematic accuracy when compared to existing techniques and to the 'gold-standard' Roentgen stereo analysis.