A retear after anterior cruciate ligament (ACL) reconstruction remains a common and devastating complication. Knee bone morphology is associated with the risk of ACL injuries, ACL retears, and osteoarthritis, and a combination of tools that derive bone shape from clinical imaging, such as magnetic resonance imaging (MRI) and statistical shape modeling, could identify patients at risk of developing these joint conditions. To identify bone shape features before primary ACL reconstruction in patients with an eventual retear compared to those with a known intact ACL graft. Case-control study; Level of evidence, 3. Bone was automatically segmented on 2-dimensional proton density-weighted MRI of the knee in patients at the time of the initial ACL injury using deep convolutional neural networks. Patients with a subsequent retear after reconstruction within 3 years (22 femurs, 19 tibias) were compared with those with an intact ACL graft at 3 years (20 femurs, 22 tibias) using statistical shape modeling to identify preoperative bone shape features predictive of a retear after ACL reconstruction. Statistical shape modeling revealed 2 specific bone shape features (modes) in the femur and 1 mode in the tibia that demonstrated significant differences at the time of the initial injury in patients with subsequent retears. In the femur, a narrower intercondylar notch width, a widened medial condylar width, an increased femoral condylar offset ratio, increased surface area along the lateral femoral condyle relative to the medial condyle, and a more prominent trochlear sulcus at the time of the initial injury were associated with retears after ACL reconstruction. In the tibia, a diminished ACL facet prominence, a squared lateral and medial tibial plateaus, and a broader and flattened tibial spine at the time of the initial injury were associated with retears after ACL reconstruction. Using the automatic bone segmentation pipeline on preoperative MRI, the authors identified bone shape features associated with a retear after ACL reconstruction. The use of this pipeline enables large-scale studies of bone shape on MRI and could predict patients at risk of ACL retears to alter treatment decisions.
Read full abstract