A slice-to-volume registration method based on real-time magnetic resonance imaging for measuring three-dimensional kinematics of the knee.

PURPOSE

This study developed and assessed a slice-to-volume registration method that integrated three-dimensional (3D) static MRI volumes of the bones with a novel single-slice, real-time radial fast low-angle shot MRI for measuring the 3D kinematics of the knee.

METHODS

Multislice 3D images (for establishing bone models) and 2D real-time images of the knee at five static positions, and 2D real-time images of the knee during flexion/extension were acquired from three healthy adults. The 3D bone poses, and thus the 3D kinematics of the knee, were obtained by registering the real-time images to a reformed slice interpolated from the bone models according to the WEMS similarity measure. The ensemble means (biases) and standard deviations (precisions) of the measurement errors of the proposed measurement method, i.e., differences between the 3D images and the registered poses, were calculated across all the static trials of all subjects. Ensemble standard deviations of all the repeated registrations for the dynamic data of all subjects were obtained to indicate the repeatability of the registration method.

RESULTS

The ensemble means (standard deviations) of the measurement errors of the femoral poses were less than 0.6 (0.6) mm for translations and -0.2° (1.3°) degrees for rotations. The corresponding values for the tibia were 0.5 (0.7) mm and -0.4° (1.1°), respectively. The ensemble means (standard deviations) of the measurement errors of knee joint poses were less than 0.9 (1.4) mm for translations and -0.3° (1.8°) degrees for rotations. For registration repeatability of dynamic tests, the ensemble standard deviations were all less than 1.2 mm for translations and 1.5° for rotations.

CONCLUSIONS

With the accuracy and repeatability achieved, and without the use of ionizing radiation and multiple repetitive motions, the proposed method combining the novel real-time MR imaging promises to be a valuable tool for studying 3D knee kinematics noninvasively.