Knee biomechanics after total knee arthroplasty (TKA) has received more attention in recent years. One critical biomechanical aspect involved in the workflow of present TKA strategies is the intraoperative optimisation of ligament balancing. Ligament balancing is usually performed with passive flexion-extension in unloaded situations. Medial and lateral ligaments strains after TKA differ in loaded flexion compared to unloaded passive flexion making the passive unloaded ligament balancing for TKA questionable. To address this problem, the development of detailed and specific knowledge on the biomechanical behaviour of loaded knee structures is essential. Stress MRI techniques were introduced in previous studies to evaluate loaded joint kinematics. Previous studies captured the knee movement either in atypical loading supine positions, or in upright positions with help of inclined supporting backrests being insufficient for movement capture under full body weight-bearing conditions.

In this work, we proposed a combined MR imaging approach for measurement and assessment of knee kinematics under full body weight-bearing in single legged stance as a first step towards the understanding of complex biomechanical aspects of bony structures and soft tissue envelope. The proposed method is based on registration of high resolution static MRI data (supine acquisition) with low resolution data, quasi-static upright-MRI data (loaded flexion positions) and was applied for the measurement of tibio-femoral kinematics in 10 healthy volunteers. The high resolution MRI data were acquired using a 1.5T Philips-Intera system, while the quasi-static MRI data (full bodyweight-bearing) was obtained with a 0.6T Fonar-Upright™ system. Contours of femur, tibia, and patella from both MRI techniques were extracted using expert manual segmentation. Anatomical surface models were then obtained for the high resolution static data.

The upright-MRI acquisition consisted of Multi-2D, quasi-static sagittal scans each including 4 slices for each flexion angle. Starting with full knee extension, the subjects were asked to increase the flexion in 4–5 steps to reach the maximum flexion angle possible under space and force limitations. Knees were softly padded for stabilisation in lateral-medial direction only in order to reduce motion artifacts. During the upright acquisition the subjects were asked to transfer their bodyweight onto the leg being imaged and maintain the predefined flexion position in single legged stance. The acquisition at every flexion angle was obtained near the scanner's isocenter and takes ∼39 seconds.

The anatomical surface models of the static data were each registered to their corresponding contours from the weight-bearing scans using an iterative closest point (ICP) based approach. A reference registration step was carried out to register the surface models to the full extension loaded position. The registered surfaces from this step were then considered as initial conditions for next ICP registration step. This procedure was similarly repeated to ensure successful registrations between subsequent flexion acquisitions.

The tibio-femoral kinematics was calculated using the joint coordinate system (JCS). The combined MR imaging approach allows the non-invasive measurement of kinematics in single legged stance and under physiological full weight-bearing conditions. We believe that this method can provide valuable insights for TKA for the validation of patient-specific biomechanical models.

Knee biomechanics after total knee arthroplasty (TKA) has received more attention in recent years. One critical biomechanical aspect involved in the workflow of present TKA strategies is the intraoperative optimisation of ligament balancing. Ligament balancing is usually performed with passive flexion-extension in unloaded situations. Medial and lateral ligaments strains after TKA differ in loaded flexion compared to unloaded passive flexion making the passive unloaded ligament balancing for TKA questionable. To address this problem, the development of detailed and specific knowledge on the biomechanical behavior of loaded knee structures is essential. Stress MRI techniques were introduced in previous studies to evaluate loaded joint kinematics. Previous studies captured the knee movement either in atypical loading supine positions, or in upright positions with help of inclined supporting backrests being insufficient for movement capture under full body weight-bearing conditions.

In this work, we proposed a combined MR imaging approach for measurement and assessment of knee kinematics under full body weight-bearing in single legged stance as a first step towards the understanding of complex biomechanical aspects of bony structures and soft tissue envelope. The proposed method is based on registration of high resolution static MRI data (supine acquisition) with low resolution data, quasi-static upright-MRI data (loaded flexion positions) and was applied for the measurement of tibio-femoral kinematics in 10 healthy volunteers. The high resolution MRI data were acquired using a 1.5T Philips-Intera system, while the quasi-static MRI data (full bodyweight-bearing) was obtained with a 0.6T Fonar-Upright™ system. Contours of femur, tibia, and patella from both MRI techniques were extracted using expert manual segmentation. Anatomical surface models were then obtained for the high resolution static data.

The upright-MRI acquisition consisted of Multi-2D, quasi-static sagittal scans each including 4 slices for each flexion angle. Starting with full knee extension, the subjects were asked to increase the flexion in 4–5 steps to reach the maximum flexion angle possible under space and force limitations. Knees were softly padded for stabilisation in lateral-medial direction only in order to reduce motion artifacts. During the upright acquisition the subjects were asked to transfer their bodyweight onto the leg being imaged and maintain the predefined flexion position in single legged stance. The acquisition at every flexion angle was obtained near the scanner's isocenter and takes ∼39 seconds.

The anatomical surface models of the static data were each registered to their corresponding contours from the weight-bearing scans using an iterative closest point (ICP) based approach. A reference registration step was carried out to register the surface models to the full extension loaded position. The registered surfaces from this step were then considered as initial conditions for next ICP registration step. This procedure was similarly repeated to ensure successful registrations between subsequent flexion acquisitions.

The tibio-femoral kinematics was calculated using the joint coordinate system (JCS). The combined MR imaging approach allows the non-invasive measurement of kinematics in single legged stance and under physiological full weight-bearing conditions. We believe that this method can provide valuable insights for TKA for the validation of patient-specific biomechanical models.