Merged CT and MRI imaging of ACL footprints A novel in vitro technique for individualized footprint analysis

Purpose: In contrast to knee arthroplasty, use of robotics in sports medicine surgery is a great void. Currently used robotic interfaces in arthroplasty are based on preoperative computed-tomography (CT) scans or on image-free systems. To optimize anterior cruciate ligament (ACL) reconstruction to a more patient specific approach, aiming to improve clinical outcomes, implementation of computer assisted robotic surgery might be the next step to individualized ACL reconstruction. As sports surgery is most often soft tissue surgery, it might be beneficial to incorporate magnetic resonance imaging (MRI) imaging into preoperative planning, intra-operative guidance, and perioperative kinematic analysis. In a first step, goal of this study was to evaluate the in-vitro feasibility and precision of merging MRI and full leg CT images with patient specific kinematic data to analyse patient specific inter-footprint behaviour of the ACL. Methods: CT and (low- and high-quality) MRI scans were acquired from 20 cadaveric lower limbs and scans were subsequently rendered in 3D. Kinematic data were obtained during testing on a passive knee rig. On 4 knees, marker sets were applied prior to scanning. Surface matching using the iterative closest point (ICP) and marker-based matching of CT, MRI and kinematic data was acquired and matching accuracy of the combined models were evaluated with statistical inter-point deviation. Results: Matching low-quality and high-quality MRI scans on the full-leg CT images (with applied kinematics), the absolute mean difference between matching distances was found to be significantly differing (1.15 vs 0.89 mm / p=0.0000052) in favour of high-quality MRI scans. Furthermore, comparison of the percentage within 1 mm (48.33% vs 66.45%) and 2 mm (87.95% and 97.30%) showed both significant differences (p=0.0000152 and p=0.0014237). Adding markers significantly increased the precision of CT-MRI matching (p=0.0000001, 0.89 mm vs 0.62 mm). The percentage of CT-high quality MRI point distances within 1 mm and 2 mm was also significantly different with respective p-values of 0.0000152 and 0.0014237. Trimming the femur to the metaphysis also increased matching accuracy (p=0.00000). Conclusion: Merged CT and MRI imaging with incorporation of patient specific data might be a valuable first step towards a more individualized perioperative ACL footprint analysis and planning to achieve a more tailored and personalized (robotic assisted) ACL reconstruction in the future. Level of evidence: V