A non-invasive method for kinematic analysis of knee joints

Abstract

The most popular and accurate kinematic analysis methods for musculoskeletal injuries or diseases, prosthesis planning and design are invasive in nature. The invasive methods are harmful and painful for the patient and not practically feasible to use for pre-operation planning on healthy knees. This paper presents a novel non-invasive, accurate kinematic analysis approach for measuring the relative motion of knee joints. A multi-modal image registration algorithm is used which successfully registers 3D CT to 2D single-plane fluoroscopy data for large initial displacements between the images. The proposed algorithm utilizes a log-polar transforms (LPT) mapping for in-pane translations, rotation and out-of-plane translation, a conjugate directional search (CDS) for out-of-plane rotation, and a fast digitally reconstructed radiograph (DRR) generation method using partial volume interpolation. Experimental results show that the achieved accuracy for kinematic analysis is similar to the invasive Roentgen stereophotogrammetric analysis (RSA) method.