A novel multi-atlas segmentation approach under the semi-supervised learning framework: Application to knee cartilage segmentation.

BACKGROUND AND OBJECTIVE: Multi-atlas based segmentation techniques, which rely on an atlas library comprised of training images labeled by an expert, have proven their effectiveness in multiple automatic segmentation applications. However, the usage of exhaustive patch libraries combined with the voxel-wise labeling incur a large computational cost in terms of memory requirements and execution times. METHODS: To confront this shortcoming, we propose a novel two-stage multi-atlas approach designed under the Semi-Supervised Learning (SSL) framework. The main properties of our method are as follows: First, instead of the voxel-wise labeling approach, the labeling of target voxels is accomplished here by exploiting the spectral content of globally sampled datasets from the target image, along with their spatially correspondent data collected from the atlases. Following SSL, voxels classification is boosted by incorporating unlabeled data from the target image, in addition to the labeled ones from atlas library. Our scheme integrates constructively fruitful concepts, including sparse reconstructions of voxels from linear neighborhoods, HOG feature descriptors of patches/regions, and label propagation via sparse graph constructions. Segmentation of the target image is carried out in two stages: stage-1 focuses on the sampling and labeling of global data, while stage-2 undertakes the above tasks for the out-of-sample data. Finally, we propose different graph-based methods for the labeling of global data, while these methods are extended to deal with the out-of-sample voxels. RESULTS: A thorough experimental investigation is conducted on 76 subjects provided by the publicly accessible Osteoarthritis Initiative (OAI) repository. Comparative results and statistical analysis demonstrate that the suggested methodology exhibits superior segmentation performance compared to the existing patch-based methods, across all evaluation metrics (DSC:88.89%, Precision: 89.86%, Recall: 88.12%), while at the same time it requires a considerably reduced computational load (>70% reduction on average execution time with respect to other patch-based). In addition, our approach is favorably compared against other non patch-based and deep learning methods in terms of performance accuracy (on the 3-class problem). A final experimentation on a 5-class setting of the problems demonstrates that our approach is capable of achieving performance comparable to existing state-of-the-art knee cartilage segmentation methods (DSC:88.22% and DSC:85.84% for femoral and tibial cartilage respectively).

A Review on Finite Element Modeling and Simulation of the Anterior Cruciate Ligament Reconstruction.

The anterior cruciate ligament (ACL) constitutes one of the most important stabilizing tissues of the knee joint whose rapture is very prevalent. ACL reconstruction (ACLR) from a graft is a surgery which yields the best outcome. Taking into account the complicated nature of this operation and the high cost of experiments, finite element (FE) simulations can become a valuable tool for evaluating the surgery in a pre-clinical setting. The present study summarizes, for the first time, the current advancement in ACLR in both clinical and computational level. It also emphasizes on the material modeling and properties of the most popular grafts as well as modeling of different surgery techniques. It can be concluded that more effort is needed to be put toward more realistic simulation of the surgery, including also the use of two bundles for graft representation, graft pretension and artificial grafts. Furthermore, muscles and synovial fluid need to be included, while patellofemoral joint is an important bone that is rarely used. More realistic models are also required for soft tissues, as most articles used isotropic linear elastic models and springs. In summary, accurate and realistic FE analysis in conjunction with multidisciplinary collaboration could contribute to ACLR improvement provided that several important aspects are carefully considered.

Biomechanics of sit-to-stand transition after muscle damage.

The purpose of the study was to examine the effects of exercise-induced muscle damage on the biomechanics of the sit-to-stand transition (STST). Seventeen volunteers participated in an intense, eccentric based, muscle damage protocol of knee flexors and extensors via an isokinetic dynamometer. Kinematic and kinetic data were collected using a 10-camera optoelectronic system and a force plate 24h before and 48h after exercise. Statistical analysis showed significant differences in kinematic and kinetic parameters after exercise. Forty-eight hours after exercise, the strategy did change and the knee joint relative effort level increased significantly. Pelvic and hip kinematics, in conjunction with the knee extension joint moment, provided an efficient mechanism to support the participants' locomotor system during the STST. These results may be of great significance in designing supportive devices, as well as composing rehabilitation programs for young or elderly individuals, with various musculoskeletal pathologies.

Mechanical correction of dynamometer moment for the effects of segment motion during isometric knee-extension tests.

The purpose of this study was to determine the effect of dynamometer and joint axis misalignment on measured isometric knee-extension moments using inverse dynamics based on the actual joint kinematic information derived from the real-time X-ray video and to compare the errors when the moments were calculated using measurements from external anatomical surface markers or obtained from the isokinetic dynamometer. Six healthy males participated in this study. They performed isometric contractions at 90° and 20° of knee flexion, gradually increasing to maximum effort. For the calculation of the actual knee-joint moment and the joint moment relative to the knee-joint center, determined using the external marker, two free body diagrams were used of the Cybex arm and the lower leg segment system. In the first free body diagram, the mean center of the circular profiles of the femoral epicondyles was used as the knee-joint center, whereas in the second diagram, the joint center was assumed to coincide with the external marker. Then, the calculated knee-joint moments were compared with those measured by the dynamometer. The results indicate that 1) the actual knee-joint moment was different from the dynamometer recorded moment (difference ranged between 1.9% and 4.3%) and the moment calculated using the skin marker (difference ranged between 2.5% and 3%), and 2) during isometric knee extension, the internal knee angle changed significantly from rest to the maximum contraction state by about 19°. Therefore, these differences cannot be neglected if the moment-knee-joint angle relationship or the muscle mechanical properties, such as length-tension relationship, need to be determined.

The effects of muscle damage on walking biomechanics are speed-dependent.

The purpose of the present study was to examine the effects of muscle damage on walking biomechanics at different speeds. Seventeen young women completed a muscle damage protocol of 5 × 15 maximal eccentric actions of the knee extensors and flexors of both legs at 60°/s. Lower body kinematics and swing-phase kinetics were assessed on a horizontal treadmill pre- and 48 h post-muscle damaging exercise at four walking speeds. Evaluated muscle damage indices included isometric torque, delayed onset muscle soreness, and serum creatine kinase. All muscle damage indices changed significantly after exercise, indicating muscle injury. Kinematic results indicated that post-exercise knee joint was significantly more flexed (31-260%) during stance-phase and knee range of motion was reduced at certain phases of the gait cycle at all speeds. Walking post-exercise at the two lower speeds revealed a more extended knee joint (3.1-3.6%) during the swing-phase, but no differences were found between pre- and post-exercise conditions at the two higher speeds. As speed increased, maximum dorsiflexion angle during stance-phase significantly decreased pre-exercise (5.7-11.8%), but remained unaltered post-exercise across all speeds (p > 0.05). Moreover, post-exercise maximum hip extension decreased (3.6-18.8%), pelvic tilt increased (5.5-10.6%), and tempo-spatial differences were found across all speeds (p < 0.05). Limited effects of muscle damage were observed regarding swing-phase kinetics. In conclusion, walking biomechanics following muscle damage are affected differently at relatively higher walking speeds, especially with respect to knee and ankle joint motion. The importance of speed in evaluating walking biomechanics following muscle damage is highlighted.

A comparison of different two-dimensional approaches for the determination of the patellar tendon moment arm length.

The purpose of this study was to estimate and compare the moment arm length of the patellar tendon (d) during passive knee extension using three different reference landmarks; instant centre of rotation (ICR), tibiofemoral contact point (TFCP) and geometrical centre of the posterior femoral condyles (GCFC). Measurements were taken on the right leg on seven healthy males during passive knee rotation performed by the motor of a Cybex Norm isokinetic dynamometer. Moment arms lengths were obtained by analysing lateral X-ray images recorded using a GE FlexiView 8800 C-arm videofluoroscopy system. The d-knee joint angle relations with respect to GCFC and ICR were similar, with decreasing values from full knee extension (~5.8 cm for d (GCFC) and ~5.9 cm for d (ICR)) to 90 degrees of knee flexion (~4.8 cm for both d (GCFC) and d (ICR)). However, the d (TFCP)-knee joint angle relation had an ascending-descending shape, with the highest d (TFCP) value (~5 cm) at 60 degrees of knee flexion. There was no significant difference between the GCFC and ICR methods at any knee joint angle. In contrast, there were significant differences (P < 0.01) between d (ICR) and d (TFCP) at 0 degrees , 15 degrees , 30 degrees and 45 degrees of knee flexion and between d (GCFC) and d (TFCP) at 0 degrees , 15 degrees and 30 degrees of knee flexion (P < 0.01). This study shows that when using different knee joint rotation centre definitions, there are significant differences in the estimates of the patellar tendon moment arm length, especially in more extended knee joint positions. These differences can have serious implications for joint modelling and loading applications.

In vivo changes in the human patellar tendon moment arm length with different modes and intensities of muscle contraction.

The purpose of this study was to examine the effect of different muscle contraction modes and intensities on patellar tendon moment arm length (d(PT)). Five men performed isokinetic concentric, eccentric and passive knee extensions at an angular velocity of 60 deg/s and six men performed gradually increasing to maximum effort isometric muscle contractions at 90( composite function) and 20( composite function) of knee flexion. During the tests, lateral X-ray fluoroscopy imaging was used to scan the knee joint. The d(PT) differences between the passive state and the isokinetic concentric and extension were quantified at 15( composite function) intervals of knee joint flexion angle. Furthermore, the changes of the d(PT) as a function of the isometric muscle contraction intensities were determined during the isometric knee extension at 90( composite function) and 20( composite function) of knee joint flexion. Muscle contraction-induced changes in knee joint flexion angle during the isometric muscle contraction were also taken into account for the d(PT) measurements. During the two isometric knee extensions, d(PT) increased from rest to maximum voluntary muscle contraction (MVC) by 14-15%. However, when changes in knee joint flexion angle induced by the muscle contraction were taken into account, d(PT) during MVC increased by 6-26% compared with rest. Moreover, d(PT) increased during concentric and eccentric knee extension by 3-15%, depending on knee flexion angle, compared with passive knee extension. These findings have important implications for estimating musculoskeletal loads using modelling under static and dynamic conditions.

Can the patellar tendon moment arm be predicted from anthropometric measurements?

The purpose of this study was to examine the relations between patellar tendon moment arm length and several relevant anthropometric characteristics of 22 healthy men. The patellar tendon moment arm length was measured using magnetic resonance imaging with two different methods: (1) measurement of patellar tendon moment arm length (d(PT)) with respect to the tibiofemoral contact point (d(PTCP)) and (2) measurement of d(PT) with respect to the intersection point of the anterior and posterior cruciate ligament (d(PTIP)). Pearson correlation coefficients and a stepwise linear regression analysis were used to examine the relationships between the d(PT) and anthropometric measurements taken. Furthermore, a Student's t-test was used to determine differences between the d(PTCP) and d(PTIP) values. Only knee circumference was a significant d(PTCP) predictor (P < 0.05) but with a very low R2 (0.139). None of the anthropometric parameters examined was found to be a significant d(PTIP) predictor. The correlation coefficients ranged from -0.04 to 0.42. The d(PTIP) values were significantly higher (by 0.84-1.89 cm) than the d(PTCP) values (P < 0.05). These results are in disagreement with previous in vitro findings that d(PT) variance may be explained by knee joint size differences. Hence, existing imaging-based methodologies remain necessary for accurate quantification of the patellar tendon moment arm.

Human patellar tendon moment arm length: measurement considerations and clinical implications for joint loading assessment.

Detailed understanding of the knee joint loading requires the calculation of muscle and joint forces in different conditions. In these applications the patellar tendon moment arm length is essential for the accurate estimation of the tibiofemoral joint loading. In this article, different methods that have been used to determine the patellar tendon moment arm length under in vivo and in vitro conditions are reviewed. The limitations and advantages associated with each of the methods are evaluated together with their applications in the different loading conditions that the musculoskeletal system is subjected to. The three main measurement methods that this review considers are the geometric method, the tendon excursion method and the direct load method. A comparison of relevant quantitative results is presented to asses the impact of the errors of each method on the quantification of the patellar tendon moment arm and the implications for joint loading assessment in clinical applications.