Distant Joints Research Articles

Local and global self-attention enhanced graph convolutional network for skeleton-based action recognition

The current successful paradigm for skeleton-based action recognition is the combination of Graph Convolutional Networks (GCNs) modeling spatial correlations, and Temporal Convolution Networks (TCNs), extracting motion features. Such GCN-TCN-based approaches usually rely on local graph convolution operations, which limits their ability to capture complicated correlations among distant joints, as well as represent long-range dependencies. Although the self-attention originated from Transformers shows great potential in correlation modeling of global joints, the Transformer-based methods are usually computationally expensive and ignore the physical connectivity structure of the human skeleton. To address these issues, we propose a novel Local-Global Self-Attention Enhanced Graph Convolutional Network (LG-SGNet) to simultaneously learn both local and global representations in the spatial–temporal dimension. Our approach consists of three components: The Local-Global Graph Convolutional Network (LG-GCN) module extracts local and global spatial feature representations by parallel channel-specific global and local spatial modeling. The Local-Global Temporal Convolutional Network (LG-TCN) module performs a joint-wise global temporal modeling using multi-head self-attention in parallel with local temporal modeling. This constitutes a new multi-branch temporal convolution structure that effectively captures both long-range dependencies and subtle temporal structures. Finally, the Dynamic Frame Weighting Module (DFWM) adjusts the weights of skeleton action sequence frames, allowing the model to adaptively focus on the features of representative frames for more efficient action recognition. Extensive experiments demonstrate that our LG-SGNet performs very competitively compared to the state-of-the-art methods. Our project website is available at https://github.com/DingYyue/LG-SGNet.

Combine multi-order representation learning and frame optimization learning for skeleton-based action recognition

Skeleton-based action recognition has broad application prospects in many fields such as virtual reality. Currently, the most popular way is to employ Graph Convolutional Networks (GCNs) or Hypergraph Convolutional Networks (HGCNs) for this task. However, GCN-based methods may heavily rely on the physical connectivity relationship between joints while lack the capture of higher-order information about interactions among distant joints, and HGCN-based methods usually introduce unnecessary noise when capturing low-order information of skeleton structures with simple topology. Besides, the current methods do not deal well with redundant frames and confusing frames. These limitations hinder the improvement of recognition accuracy. In this paper, we propose a novel network, called Hyper-Net, which combines multi-order representation learning and frame optimization learning for skeleton-based action recognition. Specifically, the proposed Hyper-Net contains Temporal-Channel Aggregation Graph Convolution (TCA-GC), Spatial-Temporal Aggregation Hypergraph Convolution (STA-HC) and Frame Optimization Learning (F-OL) modules. The TCA-GC aggregates low-order and local information from simple joint and bone topologies across different temporal and channel dimensions. The STA-HC captures high-order and global information from complex motion streams as well as solving the problem of spatial-temporal weight imbalance. The F-OL can adaptively extract key frames and distinguish confusing frames, thus improving the ability of the network to recognize confusing actions. A large number of experiments are conducted on the NTU RGB+D, NTU RGB+D 120 and NW-UCLA datasets for action recognition task. Experimental results demonstrate the superiority and effectiveness of the proposed network.

Regular Splitting Graph Network for 3D Human Pose Estimation.

In human pose estimation methods based on graph convolutional architectures, the human skeleton is usually modeled as an undirected graph whose nodes are body joints and edges are connections between neighboring joints. However, most of these methods tend to focus on learning relationships between body joints of the skeleton using first-order neighbors, ignoring higher-order neighbors and hence limiting their ability to exploit relationships between distant joints. In this paper, we introduce a higher-order regular splitting graph network (RS-Net) for 2D-to-3D human pose estimation using matrix splitting in conjunction with weight and adjacency modulation. The core idea is to capture long-range dependencies between body joints using multi-hop neighborhoods and also to learn different modulation vectors for different body joints as well as a modulation matrix added to the adjacency matrix associated to the skeleton. This learnable modulation matrix helps adjust the graph structure by adding extra graph edges in an effort to learn additional connections between body joints. Instead of using a shared weight matrix for all neighboring body joints, the proposed RS-Net model applies weight unsharing before aggregating the feature vectors associated to the joints in order to capture the different relations between them. Experiments and ablations studies performed on two benchmark datasets demonstrate the effectiveness of our model, achieving superior performance over recent state-of-the-art methods for 3D human pose estimation.

Dynamic Hand Gesture Recognition Using Improved Spatio-Temporal Graph Convolutional Network

Hand gesture recognition is essential to human-computer interaction as the most natural way of communicating. Furthermore, with the development of 3D hand pose estimation technology and the performance improvement of low-cost depth cameras, skeleton-based dynamic hand gesture recognition has received much attention. This paper proposes a novel multi-stream improved spatio-temporal graph convolutional network (MS-ISTGCN) for skeleton-based dynamic hand gesture recognition. We adopt an adaptive spatial graph convolution that can learn the relationship between distant hand joints and propose an extended temporal graph convolution with multiple dilation rates that can extract informative temporal features from short to long periods. Furthermore, we add a new attention layer consisting of effective spatio-temporal attention and channel attention between the spatial and temporal graph convolution layers to find and focus on key features. Finally, we propose a multi-stream structure that feeds multiple data modalities (i.e., joints, bones, and motions) as inputs to improve performance using the ensemble technique. Each of the three-stream networks is independently trained and fused to predict the final hand gesture. The performance of the proposed method is verified through extensive experiments with two widely used public dynamic hand gesture datasets: SHREC’17 Track and DHG-14/28. Our proposed method achieves the highest recognition accuracy in various gesture categories for both datasets compared with state-of-the-art methods.

Simplified-attention Enhanced Graph Convolutional Network for 3D human pose estimation

Optical motion capture systems have been used intensively to obtain human body poses. However, there still exist several problems. First is the dislocation problem caused by joints being too close together. The second is the joint lost problem. Restricted by severe self-occlusions, cameras may not capture the target joints. Given this observation, we investigate the high-level constraints over human poses to solve these two problems. In this work, we present a Simplified-attention Enhanced Graph Convolutional Network (SaEGC-Net) to extract both spatial and temporal features from monocular videos flexibly. The SaEGC-Net for 3D human pose estimation is U-shaped and involves the Cascaded Spatial-Temporal Graph Convolutional (CST-GC) blocks and the Simplified Spatial-Temporal Attention (SST-Att) blocks, allowing for drawing long-range dependencies between unconnected joints by graph topologies and attention mechanism, respectively. Specifically, the CST-GC block embeds two predefined graph structures into a convolutional network, incorporating discriminative features from distant joints. The proposed SST-Att block disregards redundant information by sharing part of the attention map, which is highly lightweight. It also considers dimension-expanded joint relationships to maintain the diversity of dependence. To evaluate the effectiveness of our method, we conduct extensive experiments on two datasets: Human3.6M and our own dataset FDU-Motion. Results demonstrate that our model achieves excellent performance and can competently handle the above two problems. Also, ablation studies show that our network’s submodules can better exploit the motion information of the human body.

Effects of experimental pain on the cervical spine reposition errors

BackgroundHealthy subjects showed normal variance of cervical spine reposition errors of approximately 2 degrees. Effects of experimental pain on cervical spine reposition errors were unknown; thus, the purpose of this study was to investigate the effects of experimental pain on cervical spine reposition errors.MethodsA repeated measured study design was applied. Thirty healthy subjects (12 males) were recruited. Reposition errors were extracted from upright cervical positions before and after cervical flexion movement in healthy subjects before and during experimental neck pain. Cervical spine reposition errors were calculated based on anatomical landmarks of each cervical joint. Reposition errors were extracted in degrees as constant errors and absolute errors for further statistical analysis. Repeated measures analysis of variance (RM-ANOVA) was applied to analyse experimental pain effects on either constant errors or absolute errors of different cervical joints.ResultsThe cervical spine showed non-significant difference in reposition errors regarding the constant errors (P>0.05) while larger reposition errors regarding the absolute errors during experimental pain compared to before experimental pain (P<0.001). In addition, the pain level joint (C4/C5) and its adjacent joints (C3/C4 and C5/C6) indicated larger reposition errors regarding absolute errors (P=0.035, P=0.329 and P=0.103, respectively).ConclusionsThis study firstly investigated the cervical spine reposition errors in experimental neck pain and further found the joints adjacent to the pain level showed larger errors compared to the distant joints regarding absolute errors. It may imply that the larger reposition errors in specific cervical joint indicate probable injury or pain existed adjacent to the joints.

Multi-Scale Spatial Temporal Graph Convolutional Network for Skeleton-Based Action Recognition

Graph convolutional networks have been widely used for skeleton-based action recognition due to their excellent modeling ability of non-Euclidean data. As the graph convolution is a local operation, it can only utilize the short-range joint dependencies and short-term trajectory but fails to directly model the distant joints relations and long-range temporal information that are vital to distinguishing various actions. To solve this problem, we present a multi-scale spatial graph convolution (MS-GC) module and a multi-scale temporal graph convolution (MT-GC) module to enrich the receptive field of the model in spatial and temporal dimensions. Concretely, the MS-GC and MT-GC modules decompose the corresponding local graph convolution into a set of sub-graph convolution, forming a hierarchical residual architecture. Without introducing additional parameters, the features will be processed with a series of sub-graph convolutions, and each node could complete multiple spatial and temporal aggregations with its neighborhoods. The final equivalent receptive field is accordingly enlarged, which is capable of capturing both short- and long-range dependencies in spatial and temporal domains. By coupling these two modules as a basic block, we further propose a multi-scale spatial temporal graph convolutional network (MST-GCN), which stacks multiple blocks to learn effective motion representations for action recognition. The proposed MST-GCN achieves remarkable performance on three challenging benchmark datasets, NTU RGB+D, NTU-120 RGB+D and Kinetics-Skeleton, for skeleton-based action recognition.

Double anchor embedding for accurate multi-person 2D pose estimation

Multi-person pose estimation is an important field in computer vision. Due to the lower time complexity, the bottom-up approaches have recently received more attention in multi-person 2D pose estimation, however, they are more sensitive to challenges in real-world scenarios. In this paper, we propose a multi-person pose estimation algorithm based on the Double Anchor Embedding (DAE), which shows that bottom-up algorithms are still competitive in precision. Firstly, for reducing the modeling difficulty of the detection task we divide the human joints into upper and lower half groups which are internally continuous and highly correlated. Accordingly, a novel joint affinity cue, called Double Anchor Embedding is designed, which can help the network effectively extract the information of both local contexts and global contexts, so that can better cope with occluded scenes and complex postures. Secondly, the parallel greedy joint inference algorithm is proposed to alleviate the mismatching problem of distant joints in the post-processing stage, which can also accelerate the matching process to some extent. Extensive experiments on two challenging datasets demonstrate the effectiveness and potential of our proposed framework, which is comparable to the current state-of-the-art methods.

PGCN-TCA: Pseudo Graph Convolutional Network With Temporal and Channel-Wise Attention for Skeleton-Based Action Recognition

Skeleton-based human action recognition has become an active research area in recent years. The key to this task is to fully explore both spatial and temporal features. Recently, GCN-based methods modeling the human body skeletons as spatial-temporal graphs, have achieved remarkable performances. However, most GCN-based methods use a fixed adjacency matrix defined by the dataset, which can only capture the structural information provided by joints directly connected through bones and ignore the dependencies between distant joints that are not connected. In addition, such a fixed adjacency matrix used in all layers leads to the network failing to extract multi-level semantic features. In this paper we propose a pseudo graph convolutional network with temporal and channel-wise attention (PGCN-TCA) to solve this problem. The fixed normalized adjacent matrix is substituted with a learnable matrix. In this way, the matrix can learn the dependencies between connected joints and joints that are not physically connected. At the same time, learnable matrices in different layers can help the network capture multi-level features in spatial domain. Moreover, Since frames and input channels that contain outstanding characteristics play significant roles in distinguishing the action from others, we propose a mixed temporal and channel-wise attention. Our method achieves comparable performances to state-of-the-art methods on NTU-RGB+D and HDM05 datasets.

Does Chondrocalcinosis Associate With a Distinct Radiographic Phenotype of Osteoarthritis in Knees and Hips? A Case–Control Study

To examine if knee chondrocalcinosis (CC), hip CC, or CC at distant joints associates with a distinct radiographic phenotype of osteoarthritis (OA) in knees or hips. We conducted a case-control study using data from the Genetics of Osteoarthritis and Lifestyle (GOAL) study (n = 3,170). All participants of the GOAL study had radiographs of knees, hands, and pelvis, which have been scored for CC and for individual radiographic features of OA. For this study, cases had radiographic OA and CC, and controls had radiographic OA without CC at the index joint. Data for knees and hips were analyzed separately. Binary logistic regression was used to examine the association between each radiographic phenotype and CC in joints with OA. Generalized estimating equation analysis was used to account for correlated data. Knee CC, and CC at any distant joint (without knee CC), associated with attrition in knee OA (adjusted odds ratio 2.32 [95% confidence interval 1.42-3.79] and 2.42 [1.41-4.13], respectively). There was no association between knee CC and osteophytosis or joint space narrowing (JSN) in knees with OA. Hip CC associated negatively with the summated osteophyte score and minimum JSN in hip OA. However, in hips with OA, CC did not associate with cysts or sclerosis. Additionally, distant joint CC did not associate with any structural change in hip OA. This study demonstrates that knee CC and CC at distant joints associate with attrition in knee OA, and hip CC associates with a milder hip OA phenotype. There was no evidence that CC associates with a hypertrophic OA phenotype.

Evidence of a Systemic Predisposition to Chondrocalcinosis and Association Between Chondrocalcinosis and Osteoarthritis at Distant Joints: A Cross‐Sectional Study

To determine whether there is a systemic predisposition to chondrocalcinosis (CC) and to examine the association between CC and osteoarthritis (OA) at distant joints. We performed a cross-sectional study embedded in the Genetics of Osteoarthritis and Lifestyle (GOAL) database (n = 3,170). All GOAL participants have had radiographs of the knees, hands, and pelvis performed. These were scored for OA at the knee, hip, wrist, and metacarpophalangeal (MCP) joints and for CC at the knee, hip, wrist, and symphysis pubis joints and for MCP joint calcification. Systemic predisposition to CC was established using cluster analysis. Odds ratios (95% confidence intervals) were used to examine the association between CC at the index and distant joints, CC and OA at the same joint, and index joint OA and distant joint CC. We adjusted for age, sex, and body mass index, and for distant joint OA, if required. Joints with CC clustered together. This was also observed when participants with OA were excluded from the analysis. CC at each joint was associated with CC at distant joints. Knee and wrist OA but not hip OA was associated with CC at the same joint. MCP joint OA was associated with MCP joint calcification. Knee OA was associated with CC at other joints, and this was independent of OA at the distant joint. There was no association between hip OA and distant joint CC. There is a systemic predisposition to the apparently sporadic CC. OA is associated with CC at the same joint and at distant joints, except hip OA, which is not associated with hip CC or with CC at distant joints.

AB1047 Evidence that a systemic predisposition to chondrocalcinosis exists, and is mediated by osteoarthritis at most but not all joints

BackgroundIt is not known whether radiographic chondrocalcinosis (CC) results from a systemic predisposition or whether the association between CC and osteoarthritis (OA) is joint specific or generalized.ObjectivesTo ascertain if there...

AB1048 Does chondrocalcinosis associate with a different radiographic phenotype of osteoarthritis?

BackgroundSome studies suggest that the structural changes in joints with osteoarthritis (OA) and co-existing radiographic chondrocalcinosis (CC) are distinct from joints with OA alone. However, other studies suggest that there...

The contralateral effect conferred by intra-articular adenovirus-mediated gene transfer of viral IL-10 is specific to the immunizing antigen.

We have demonstrated previously that local, adenoviral-mediated gene transfer of vIL-10 to a single joint of rabbits and mice with experimental arthritis can suppress disease in both the treated and untreated contralateral joints. These therapeutic effects observed in distant untreated joints following local intra-articular gene delivery have been termed the 'contralateral effect'. To begin to understand the underlying immunologic mechanism that confers this effect, a dual-antigen model of antigen-induced arthritis (AIA) in rabbit knee joints was utilized. Rabbits were immunized against two antigens, ovalbumin and keyhole limpet hemocyanin, and AIA generated by intra-articular injection of each antigen into contralateral knees. Intra-articular adenovirus-mediated gene transfer of vIL-10 significantly reduced intra-articular leukocytosis and cartilage matrix degradation, while preserving near normal levels of cartilage matrix synthesis within treated joints. However, no antiarthritic effect was conferred in the contralateral control joints that received only a marker gene, in contrast to the results seen in a single-antigen AIA model. These results suggest that the distant antiarthritic effects associated with local gene delivery to joints are antigen-specific, and not due to vIL-10-induced generalized immunosuppression of the animal.

Ex Vivo Gene Delivery of IL-1Ra and Soluble TNF Receptor Confers a Distal Synergistic Therapeutic Effect in Antigen-Induced Arthritis

Intra-articular expression of antagonists of interleukin-1beta (IL-1beta) and tumor necrosis factor-alpha (TNF-alpha) in arthritic rabbit knee and mouse ankle joints by direct adenoviral-mediated intraarticular delivery results in amelioration of disease pathology in both the treated and contralateral untreated joints. Previous experiments suggest that direct adenoviral infection of resident antigen-presenting cells (APCs) and subsequent traveling of these cells to other sites of inflammation and lymph nodes might be responsible for this "contralateral effect." To determine whether genetic modification of APCs is required for the contralateral effect, we have used an ex vivo approach utilizing genetically modified fibroblasts to express IL-1 receptor antagonist protein (IL-1Ra) and soluble TNF-alpha receptor (sTNFR) locally in arthritic joints. Retroviral vectors carrying IL-1Ra, sTNFR-Ig, or both genes together were used to stably infect autologous rabbit fibroblasts that were then injected intra-articularly into arthritic rabbit knee joints. The intra-articular delivery of either IL-1Ra- or sTNFR-Ig-expressing fibroblasts was antiinflammatory and chondro-protective in both the injected and noninjected contralateral joints. In addition, we demonstrate that the co-delivery of both antagonists in combination results in a synergistic effect in disease amelioration in both the treated and nontreated joints. These ex vivo results suggest that trafficking of vector-modified inflammatory cells is not the main mechanism responsible for the observed distal spread of the therapeutic effect. Moreover, the results demonstrate that local, ex vivo gene therapy for arthritis could be effective in blocking pathologies within untreated, distant arthritic joints.

Monosynaptic Ia pathways at the cat shoulder.

1. The study aimed to describe in cat forelimb and shoulder motoneurones the convergence and projection patterns from large muscle spindle afferents (Ia). In 11 chloralose-anaesthetized cats maximum Ia EPSPs evoked by electrical stimulation of ipsilateral forelimb nerves were obtained in 309 intracellularly recorded alpha-motoneurones. 2. Groups of motor nuclei displayed similar Ia patterns. As in the distal forelimb they were often interconnected by bidirectional pathways, which were used to combine Ia synergistic groups. Three such groups are described at the shoulder. 3. The first group was composed of the main flexors of the scapulo-humeral joint. Regular disto-proximal Ia excitation from elbow extensors (and median afferents) indicates a coupling of flexion in the scapulo-humeral joint to the angular position of the elbow. 4. The second group comprised the outward rotators of the humerus with differentiated Ia convergence onto the different group members. The patterns of Ia excitation received and sent by the group members demonstrate that the outward rotators are incorporated in versatile synergisms and may occupy a central position in steering forelimb movements. 5. The third group was formed by the spinatus muscle and the subscapularis. This arrangement is suggested by the common convergence onto them from the elbow extensors and flexors. The pattern may serve to guide and keep the humeral head in the joint capsule. 6. The Ia synergistic groups receive Ia convergence from muscles acting at distant joints and also project to distant muscles. This is discussed as part of an extended pattern of Ia connections along the forelimb. In this way the shoulder muscles would be incorporated in flexor and extensor oriented synergisms which are needed to co-ordinate the muscular activation along the multijoint forelimb during locomotion. When the shoulder Ia pathways are compared with those in the distal forelimb, organization of the Ia system apparently follows a few basic principles which have adapted to the mechanical situation at the particular joints and their mechanical interaction.

Polyarticular septic arthritis.

Twenty-five cases of polyarticular septic arthritis (PASA) were observed in our department over a 13-year period. They accounted for 16.6% of all septic arthritis (15% on average in the literature). A male predominance was noted in our patients, as well as in the literature. The knee was the most frequent location followed by the elbow, shoulder, and hip, in varying order depending on the series. An average of 4 joints was involved. The causative microorganism was Staphylococcus aureus in 20/25 of our patients and in about 50% of published cases. Other frequently causative organisms were streptococci and gram-negative bacteria. Blood cultures and joint aspirations were positive in 19/22 and 23/25 of our cases, respectively. Other septic lesions were noted in 10/25 of our cases. Fever and severe leukocytosis were absent at admission in 5/25 (literature, 37%) and 10/25 of our 25 patients, respectively. The underlying disease was rheumatoid arthritis in 13/25, while 9 of the other patients had immunodepression caused by drugs or by concurrent illness. Typically, rheumatoid arthritis was long-standing and erosive, patients having ulcerated calluses on the feet. This skin source was also noted in 23/36 published cases of PASA in rheumatoid arthritis. Systemic lupus erythematosus was an uncommon disease in PASA, but its presence promoted gram-negative infection. Despite effective therapy with 2 antibiotics, 8/25 patients died, a prognosis that is equally severe in cases reported in the literature (30%) and one that has remained surprisingly stable over the last 40 years. For comparison, the death rate was only 4% in our patients with MASA. Factors contributing to a poor prognosis were age greater than 50 years, rheumatoid arthritis as an underlying disease, and disease of staphylococcal origin. Septic polyarthritis should be considered even when the clinical picture is not florid--when patients have low fever and normal white blood cell counts. Nor should the simultaneous involvement of distant joints rule out infection. Indeed, the frequency of underlying rheumatic disease and its treatment may further confuse the clinical presentation. Joints suspected of harboring infection should be aspirated, including those previously affected by the concurrent rheumatism.

A case of pseudogout without radiographic evidence of chondrocalcinosis.

Articular chondrocalcinosis with both arthritis and abnormal calcium deposition in hyaline and fibrocartilage was first described as a clinical entity in 1958 by Zitnan and Sitaj (9). Subsquently McCarty et al. (3) identified calcium pyrophosphate dihydrate (CPPD) crystals from both synovial fluid and cartilaginous tissues using compensated polarized light microscopy, x-ray diffraction and infrared spectrophotometry in patients with articular chondrocalcinosis and termed CPPD deposition disease or pseudogout. Although a considerable variation of clinical features has been described (1, 4), pseudogout without radiographic evidence of chondrocalcinosis seems to be very rare and also difficult to determine a correct diagnosis. In this paper, a case of pseudogout without radiographic evidence in either involved or distant joints is presented.

The measurement of skin blood flow in peripheral vascular disease by epicutaneous application of Xenon133

Radioactive xenon133 applied epicutaneously was used to study the skin blood flow below the knee in sixteen normal subjects, in eight patients with peripheral vascular disease not requiring amputation, and in a blind study of twenty-nine patients requiring amputation. Following these twenty-nine amputations, the flow rates were made known and correlated with the rates of healing. When the flow rates were above 1.5 milliliters per minute per 100 grams of tissue (skin), the wounds generally healed. In thirteen other patients, Syme or below-the-knee amputations were performed on the basis of the flow rates, and all of the wounds healed. The test, therefore, is now used routinely prior to amputation for peripheral vascular disease as an adjunct to clinical judgment in the determination of the level of amputation.