Neural network segmentation of disc volume from magnetic resonance images and the effect of degeneration and spinal level.

Background: Magnetic resonance imaging (MRI) noninvasively quantifies disc structure but requires segmentation that is both time intensive and susceptible to human error. Recent advances in neural networks can improve on manual segmentation. The aim of this study was to establish a method for automatic slice-wise segmentation of 3D disc volumes from subjects with a wide range of age and degrees of disc degeneration. A U-Net convolutional neural network was trained to segment 3D T1-weighted spine MRI. Methods: Lumbar spine MRIs were acquired from 43 subjects (23-83 years old) and manually segmented. A U-Net architecture was trained using the TensorFlow framework. Two rounds of model tuning were performed. The performance of the model was measured using a validation set that did not cross over from the training set. The model version with the best Dice similarity coefficient (DSC) was selected in each tuning round. After model development was complete and a final U-Net model was selected, performance of this model was compared between disc levels and degeneration grades. Results: Performance of the final model was equivalent to manual segmentation, with a mean DSC = 0.935 ± 0.014 for degeneration grades I-IV. Neither the manual segmentation nor the U-Net model performed as well for grade V disc segmentation. Compared with the baseline model at the beginning of round 1, the best model had fewer filters/parameters (75%), was trained using only slices with at least one disc-labeled pixel, applied contrast stretching to its input images, and used a greater dropout rate. Conclusion: This study successfully trained a U-Net model for automatic slice-wise segmentation of 3D disc volumes from populations with a wide range of ages and disc degeneration. The final trained model is available to support scientific use.

Acute Repair of Meniscus Root Tear Partially Restores Joint Displacements as Measured With Magnetic Resonance Images and Loading in a Cadaveric Porcine Knee.

The meniscus serves important load-bearing functions and protects the underlying articular cartilage. Unfortunately, meniscus tears are common and impair the ability of the meniscus to distribute loads, increasing the risk of developing osteoarthritis. Therefore, surgical repair of the meniscus is a frequently performed procedure; however, repair does not always prevent osteoarthritis. This is hypothesized to be due to altered joint loading post-injury and repair, where the functional deficit of the meniscus prevents it from performing its role of distributing forces. The objective of this study was to quantify joint kinematics in an intact joint, after a meniscus root tear, and after suture repair in cadaveric porcine knees, a frequently used in vivo model. We utilized an magnetic resonance images-compatible loading device and novel use of a T1 vibe sequence to measure meniscus and femur displacements under physiological axial loads. We found that anterior root tear led to large meniscus displacements under physiological axial loading and that suture anchor repair reduced these displacements but did not fully restore intact joint kinematics. After tear and repair, the anterior region of the meniscus moved posteriorly and medially as it was forced out of the joint space under loading, while the posterior region had small displacements as the posterior attachment acted as a hinge about which the meniscus pivoted in the axial plane. Methods from this study can be applied to assess altered joint kinematics following human knee injuries and evaluate repair strategies aimed to restore joint kinematics.

Acute repair of meniscus root tear partially restores joint displacements as measured with MRI and loading in a porcine knee.

The meniscus serves important load-bearing functions and protects the underlying articular cartilage. Unfortunately, meniscus tears are common and impair the ability of the meniscus to distribute loads, greatly increasing the risk for developing osteoarthritis. Therefore, surgical repair of the meniscus is a frequently performed procedure; however, this repair does not always prevent osteoarthritis. This is hypothesized to be due to altered joint loading post injury and repair, where the functional deficit of the meniscus prevents it from performing its role of distributing forces. However, many studies of meniscus function required opening the joint, which alters kinematics. The objective of this study was to use novel MRI methods to image the intact joint under axial load and measure the acute meniscus and femur displacements in an intact joint, after a meniscus root tear, and after suture repair in the porcine knee, a frequently used in vivo model. We found that anterior root tear led to large meniscus and femur displacements under physiological axial loading, and that suture anchor repair reduced these displacements, but did not fully restore intact joint kinematics. After tear and repair, the anterior region of the meniscus moved posteriorly and medially as it was forced out of the joint space under loading, while the posterior region had small displacements as the posterior attachment acted as a hinge about which the meniscus rotated in the axial plane. This technique can be applied to evaluate the effect of knee injuries and to develop improved repair strategies to restore joint kinematics.