A NOVEL MAGNETIC RESONANCE IMAGING METHOD ACCURATELY ASSESSES THREE-DIMENSIONAL PATELLAR TRACKING

Abstract

We evaluated the accuracy of a Magnetic Resonance Imaging (MRI)-based method to measure three-dimensional patellar tracking during loaded knee flexion. This method determines the relative positions of the knee bones by shape matching high-resolution three-dimensional geometric models of these bones to fast low-resolution scans taken during loaded flexion.

The accuracy of the method’s assessment of patellar position and orientation was determined by comparing test measurements in four cadaver specimens to measurements made in the same specimens using Roentgen Stereophotogrammetric Analysis (RSA). This MRI-based method is more accurate than current two-dimensional imaging methods.

The purpose of this study was to determine the accuracy of a MRI-based technique for measuring patellar tracking in loaded flexion.

This novel, noninvasive, MRI-based method measures three-dimensional patellar tracking during loaded knee flexion with sufficient accuracy to detect clinically significant changes.

Although abnormal patellar tracking is widely believed to be associated with pain and cartilage degeneration at the patella, these relationships have not been clearly established because most current methods assess only the two-dimensional alignment of the patella at one position. Measurements possible with this method should be sufficiently accurate to yield new insights into these relationships.

Four cadaver knee specimens were flexed through seventy-five degrees of flexion in an MRI-compatible knee loading rig. A high-resolution image was acquired with each knee in extension and then a series of low-resolution scans (in two slice directions: axial and sagittal) were acquired through a flexion cycle. Segmenting bone outlines from high-resolution scans generated models of the femur, tibia and patella. These models were shape matched to the segmented bone outlines in the low resolution scans. Patellar attitude and position were determined and compared to measurements made using RSA.

The mean measurement error in every kinematic parameter was lower for “fast” sagittal plane slices than for “fast” axial plane slices. In general, the mean measurement error was increased by decreasing the number of low-resolution slices.

This method is more accurate than many two-dimensional methods, exposes participants to no ionizing radiation, and can be used through a large range of loaded knee flexion.

Funding: Supported by an operating grant from the Canadian Institutes for Health Research and a Strategic Grant from the Natural Sciences and Engineering Research Council. NJM is supported by the Arthritis Society/CIHR Partnership Fund.

Please contact author for figures and/or tables.