Auto-segmentation of the tibia and femur from knee MR images via deep learning and its application to cartilage strain and recovery.

The ability to efficiently and reproducibly generate subject-specific 3D models of bone and soft tissue is important to many areas of musculoskeletal research. However, methodologies requiring such models have largely been limited by lengthy manual segmentation times. Recently, machine learning, and more specifically, convolutional neural networks, have shown potential to alleviate this bottleneck in research throughput. Thus, the purpose of this work was to develop a modified version of the convolutional neural network architecture U-Net to automate segmentation of the tibia and femur from double echo steady state knee magnetic resonance (MR) images. Our model was trained on a dataset of over 4,000 MR images from 34 subjects, segmented by three experienced researchers, and reviewed by a musculoskeletal radiologist. For our validation and testing sets, we achieved dice coefficients of 0.985 and 0.984, respectively. As further testing, we applied our trained model to a prior study of tibial cartilage strain and recovery. In this analysis, across all subjects, there were no statistically significant differences in cartilage strain between the machine learning and ground truth bone models, with a mean difference of 0.2 ± 0.7 % (mean ± 95 % confidence interval). This difference is within the measurement resolution of previous cartilage strain studies from our lab using manual segmentation. In summary, we successfully trained, validated, and tested a machine learning model capable of segmenting MR images of the knee, achieving results that are comparable to trained human segmenters.

Elevated In Vivo ACL Strain Is Associated With a Straight Knee in Both the Sagittal and the Coronal Planes.

BACKGROUND: Noncontact anterior cruciate ligament (ACL) injuries typically occur during deceleration movements such as landing or cutting. However, conflicting data have left the kinematic mechanisms leading to these injuries unclear. Quantifying the influence of sagittal and coronal plane knee kinematics on in vivo ACL strain may help to elucidate noncontact ACL injury mechanisms. PURPOSE/HYPOTHESIS: The purpose of this study was to measure in vivo sagittal and coronal plane knee kinematics and ACL strain during a single-leg jump. We hypothesized that ACL strain would be modulated primarily by motion in the sagittal plane and that limited coronal plane motion would be measured during this activity. STUDY DESIGN: Descriptive laboratory study. METHODS: Seventeen healthy participants (8 male/9 female) underwent magnetic resonance imaging (MRI) followed by high-speed biplanar radiography, obtained as participants performed a single-leg jump. Three-dimensional models of the femur, tibia, and associated ACL attachment site footprints were created from the MRIs and registered to the radiographs to reproduce the position of the knee during the jump. ACL strain, knee flexion/extension angles, and varus/valgus angles were measured throughout the jump. Spearman rank correlations were used to assess relationships between mean ACL strain and kinematic variables. RESULTS: Mean ACL strain increased with decreasing knee flexion angle (ρ = -0.3; P = .002), and local maxima in ACL strain occurred with the knee in a straight position in both the sagittal and the coronal planes. In addition, limited coronal plane motion (varus/valgus angle) was measured during this activity (mean ± SD, -0.5°± 0.3°). Furthermore, we did not detect a statistically significant relationship between ACL strain and varus/valgus angle (ρ = -0.01; P = .9). CONCLUSION: ACL strain was maximized when the knee was in a straight position in both the sagittal and coronal planes. Participants remained in <1° of varus/valgus position on average throughout the jump. As a ligament under elevated strain is more vulnerable to injury, landing on a straight knee may be an important risk factor for ACL rupture. CLINICAL RELEVANCE: These data may improve understanding of risk factors for noncontact ACL injury, which may be useful in designing ACL injury prevention programs.

The Interplay of Biomechanical and Biological Changes Following Meniscus Injury.

PURPOSE OF REVIEW: Meniscus injury often leads to joint degeneration and post-traumatic osteoarthritis (PTOA) development. Therefore, the purpose of this review is to outline the current understanding of biomechanical and biological repercussions following meniscus injury and how these changes impact meniscus repair and PTOA development. Moreover, we identify key gaps in knowledge that must be further investigated to improve meniscus healing and prevent PTOA. RECENT FINDINGS: Following meniscus injury, both biomechanical and biological alterations frequently occur in multiple tissues in the joint. Biomechanically, meniscus tears compromise the ability of the meniscus to transfer load in the joint, making the cartilage more vulnerable to increased strain. Biologically, the post-injury environment is often characterized by an increase in pro-inflammatory cytokines, catabolic enzymes, and immune cells. These multi-faceted changes have a significant interplay and result in an environment that opposes tissue repair and contributes to PTOA development. Additionally, degenerative changes associated with OA may cause a feedback cycle, negatively impacting the healing capacity of the meniscus. Strides have been made towards understanding post-injury biological and biomechanical changes in the joint, their interplay, and how they affect healing and PTOA development. However, in order to improve clinical treatments to promote meniscus healing and prevent PTOA development, there is an urgent need to understand the physiologic changes in the joint following injury. In particular, work is needed on the in vivo characterization of the temporal biomechanical and biological changes that occur in patients following meniscus injury and how these changes contribute to PTOA development.