Three-dimensional convolutional neural network-based classification of chronic kidney disease severity using kidney MRI.

A three-dimensional convolutional neural network model was developed to classify the severity of chronic kidney disease (CKD) using magnetic resonance imaging (MRI) Dixon-based T1-weighted in-phase (IP)/opposed-phase (OP)/water-only (WO) imaging. Seventy-three patients with severe renal dysfunction (estimated glomerular filtration rate [eGFR] < 30 mL/min/1.73 m2, CKD stage G4-5); 172 with moderate renal dysfunction (30 ≤ eGFR < 60 mL/min/1.73 m2, CKD stage G3a/b); and 76 with mild renal dysfunction (eGFR ≥ 60 mL/min/1.73 m2, CKD stage G1-2) participated in this study. The model was applied to the right, left, and both kidneys, as well as to each imaging method (T1-weighted IP/OP/WO images). The best performance was obtained when using bilateral kidneys and IP images, with an accuracy of 0.862 ± 0.036. The overall accuracy was better for the bilateral kidney models than for the unilateral kidney models. Our deep learning approach using kidney MRI can be applied to classify patients with CKD based on the severity of kidney disease.

A high precision laser scanning system for measuring shape and volume of transtibial amputee residual limbs: Design and validation.

Changes in limb volume and shape among transtibial amputees affects socket fit and comfort. The ability to accurately measure residual limb volume and shape and relate it to comfort could contribute to advances in socket design and overall care. This work designed and validated a novel 3D laser scanner that measures the volume and shape of residual limbs. The system was designed to provide accurate and repeatable scans, minimize scan duration, and account for limb motion during scans. The scanner was first validated using a cylindrical body with a known shape. Mean volumetric errors of 0.17% were found under static conditions, corresponding to a radial spatial resolution of 0.1 mm. Limb scans were also performed on a transtibial amputee and yielded a standard deviation of 8.1 ml (0.7%) across five scans, and a 46 ml (4%) change in limb volume when the socket was doffed after 15 minutes of standing.

Pipeline for Multi-Scale Three-Dimensional Anatomic Study of the Human Heart.

Detailed study of non-failing human hearts rejected for transplantation provides a unique opportunity to perform structural analyses across microscopic and macroscopic scales. These techniques include tissue clearing (modified immunolabeling-enabled three-dimensional (3D) imaging of solvent-cleared organs) and immunohistochemical staining. Mesoscopic examination procedures include stereoscopic dissection and micro-computed tomographic (CT) scanning. Macroscopic examination procedures include gross dissection, photography (including anaglyphs and photogrammetry), CT, and 3D printing of the physically or virtually dissected or whole heart. Before macroscopic examination, pressure-perfusion fixation may be performed to maintain the 3D architecture and physiologically relevant morphology of the heart. The application of these techniques in combination to study the human heart is unique and crucial in understanding the relationship between distinct anatomic features such as coronary vasculature and myocardial innervation in the context of the 3D architecture of the heart. This protocol describes the methodologies in detail and includes representative results to illustrate progress in the research of human cardiac anatomy.

Efficient musculoskeletal annotation using free-form deformation.

Traditionally, constructing training datasets for automatic muscle segmentation from medical images involved skilled operators, leading to high labor costs and limited scalability. To address this issue, we developed a tool that enables efficient annotation by non-experts and assessed its effectiveness for training an automatic segmentation network. Our system allows users to deform a template three-dimensional (3D) anatomical model to fit a target magnetic-resonance image using free-form deformation with independent control points for axial, sagittal, and coronal directions. This method simplifies the annotation process by allowing non-experts to intuitively adjust the model, enabling simultaneous annotation of all muscles in the template. We evaluated the quality of the tool-assisted segmentation performed by non-experts, which achieved a Dice coefficient greater than 0.75 compared to expert segmentation, without significant errors such as mislabeling adjacent muscles or omitting musculature. An automatic segmentation network trained with datasets created using this tool demonstrated performance comparable to or superior to that of networks trained with expert-generated datasets. This innovative tool significantly reduces the time and labor costs associated with dataset creation for automatic muscle segmentation, potentially revolutionizing medical image annotation and accelerating the development of deep learning-based segmentation networks in various clinical applications.

Axially swept open-top light-sheet microscopy for densely labeled clinical specimens.

Open-top light-sheet (OTLS) microscopy offers rapid 3D imaging of large optically cleared specimens. This enables nondestructive 3D pathology, which provides key advantages over conventional slide-based histology including comprehensive sampling without tissue sectioning/destruction and visualization of diagnostically important 3D structures. With 3D pathology, clinical specimens are often labeled with small-molecule stains that broadly target nucleic acids and proteins, mimicking conventional hematoxylin and eosin (H&E) dyes. Tight optical sectioning helps to minimize out-of-focus fluorescence for high-contrast imaging in these densely labeled tissues but has been challenging to achieve in OTLS systems due to trade-offs between optical sectioning and field of view. Here we present an OTLS microscope with voice-coil-based axial sweeping to circumvent this trade-off, achieving 2 µm axial resolution over a 750 × 375 µm field of view. We implement our design in a non-orthogonal dual-objective (NODO) architecture, which enables a 10-mm working distance with minimal sensitivity to refractive index mismatches, for high-contrast 3D imaging of clinical specimens.

Unveiling nuclear chromatin distribution using IsoConcentraChromJ: A flourescence imaging plugin for IsoRegional and IsoVolumetric based ratios analysis.

Chromatin exhibits non-random distribution within the nucleus being arranged into discrete domains that are spatially organized throughout the nuclear space. Both the spatial distribution and structural rearrangement of chromatin domains in the nucleus depend on epigenetic modifications of DNA and/or histones and structural elements such as the nuclear envelope. These components collectively contribute to the organization and rearrangement of chromatin domains, thereby influencing genome architecture and functional regulation. This study develops an innovative, user-friendly, ImageJ-based plugin, called IsoConcentraChromJ, aimed quantitatively delineating the spatial distribution of chromatin regions in concentric patterns. The IsoConcentraChromJ can be applied to quantitative chromatin analysis in both two- and three-dimensional spaces. After DNA and histone staining with fluorescent probes, high-resolution images of nuclei have been obtained using advanced fluorescence microscopy approaches, including confocal and stimulated emission depletion (STED) microscopy. IsoConcentraChromJ workflow comprises the following sequential steps: nucleus segmentation, thresholding, masking, normalization, and trisection with specified ratios for either 2D or 3D acquisitions. The effectiveness of the IsoConcentraChromJ has been validated and demonstrated using experimental datasets consisting in nuclei images of pre-adipocytes and mature adipocytes, encompassing both 2D and 3D imaging. The outcomes allow to characterize the nuclear architecture by calculating the ratios between specific concentric nuclear areas/volumes of acetylated chromatin with respect to total acetylated chromatin and/or total DNA. The novel IsoConcentrapChromJ plugin could represent a valuable resource for researchers investigating the rearrangement of chromatin architecture driven by epigenetic mechanisms using nuclear images obtained by different fluorescence microscopy methods.

HVSMR-2.0: A 3D cardiovascular MR dataset for whole-heart segmentation in congenital heart disease.

Patients with congenital heart disease often have cardiac anatomy that deviates significantly from normal, frequently requiring multiple heart surgeries. Image segmentation from a preoperative cardiovascular magnetic resonance (CMR) scan would enable creation of patient-specific 3D surface models of the heart, which have potential to improve surgical planning, enable surgical simulation, and allow automatic computation of quantitative metrics of heart function. However, there is no publicly available CMR dataset for whole-heart segmentation in patients with congenital heart disease. Here, we release the HVSMR-2.0 dataset, comprising 60 CMR scans alongside manual segmentation masks of the 4 cardiac chambers and 4 great vessels. The images showcase a wide range of heart defects and prior surgical interventions. The dataset also includes masks of required and optional extents of the great vessels, enabling fairer comparisons across algorithms. Detailed diagnoses for each subject are also provided. By releasing HVSMR-2.0, we aim to encourage development of robust segmentation algorithms and clinically relevant tools for congenital heart disease.

Optimizing immunofluorescence with high-dynamic-range imaging to enhance PD-L1 expression evaluation for 3D pathology assessment from NSCLC tumor tissue.

Assessing programmed death ligand 1 (PD-L1) expression through immunohistochemistry (IHC) is the golden standard in predicting immunotherapy response of non-small cell lung cancer (NSCLC). However, observation of heterogeneous PD-L1 distribution in tumor space is a challenge using IHC only. Meanwhile, immunofluorescence (IF) could support both planar and three-dimensional (3D) histological analyses by combining tissue optical clearing with confocal microscopy. We optimized clinical tissue preparation for the IF assay focusing on staining, imaging, and post-processing to achieve quality identical to traditional IHC assay. To overcome limited dynamic range of the fluorescence microscope's detection system, we incorporated a high dynamic range (HDR) algorithm to restore the post imaging IF expression pattern and further 3D IF images. Following HDR processing, a noticeable improvement in the accuracy of diagnosis (85.7%) was achieved using IF images by pathologists. Moreover, 3D IF images revealed a 25% change in tumor proportion score for PD-L1 expression at various depths within tumors. We have established an optimal and reproducible process for PD-L1 IF images in NSCLC, yielding high quality data comparable to traditional IHC assays. The ability to discern accurate spatial PD-L1 distribution through 3D pathology analysis could provide more precise evaluation and prediction for immunotherapy targeting advanced NSCLC.

Common double-lumen tube selection methods overestimate adequate tube sizes in individual patients - a 3D reconstruction study.

BACKGROUND: Appropriate selection of double-lumen tube sizes for one-lung ventilation is crucial to prevent airway damage. Current selection methods rely on demographic factors or 2D radiography. Prediction of left bronchial diameter is indispensable for choosing the adequate tube size. This prospective observational study investigates if current selection methods sufficiently predict individuals' left bronchial diameters for DLT selection compared to the 3D reconstruction. METHODS: 100 patients necessitating thoracic surgery with one-lung ventilation and left-sided double-lumen tubes, ≥ 18 years of age, and a set of chest X-rays and 2D thorax CT scans for 3D reconstruction of the left main bronchus were included between 07/2021 and 06/2023. The cross-validated prediction error and the width of the 95%-prediction intervals of the 3D left main bronchial diameter utilizing linear prediction models were based on current selection methods. RESULTS: The mean bronchial diameter in 3D reconstruction was 13.6 ± 2.1 mm. The ranges of the 95%-prediction intervals for the bronchial diameter were 6.4 mm for demographic variables, 8.3 mm for the tracheal diameter from the X-ray, and 5.9 mm for bronchial diameter from the 2D-CT scans. Current methods violated the suggested '≥1 mm' safety criterion in up to 7% (men) and 42% (women). Particularly, 2D radiography overestimated women's left bronchial diameter. Current methods even allowed the selection of double-lumen tubes with bronchial tube sections greater than the bronchial diameter in women. CONCLUSIONS: Neither demographic nor 2D-radiographic methods sufficiently account for the variability of the bronchial diameter. Wide 95%-prediction intervals for the bronchial diameter hamper accurate individual double-lumen tube selection. This increases women's risk of bronchial damage, particularly if they have other predisposing factors. These patients may benefit from 3D reconstruction of the left main bronchus. TRIAL REGISTRATION: Not applicable.

Effectiveness of different intrusion modes of maxillary anterior teeth with mini-implants in clear aligner treatment: a three-dimensional finite element analysis.

BACKGROUND: The intrusion of maxillary anterior teeth is often required and there are various intrusion modes with mini-implants in clear aligner treatment. The objective of this study was to evaluate the effectiveness of maxillary anterior teeth intrusion with different intrusion modes, aiming to provide references for precise and safe intrusion movements in clinical practice. METHODS: Cone-beam computed tomography and intraoral optical scanning data of a patient were collected. Finite element models of the maxilla, maxillary dentition, periodontal ligaments (PDLs), clear aligner (CA), attachments, and mini-implants were established. Different intrusion modes of the maxillary anterior teeth were simulated by changing the mini-implant site (between central incisors, between central and lateral incisor, between lateral incisor and canine), loading site (between central incisors, on central incisor, between central and lateral incisor, between lateral incisor and canine), and loading mode (labial loading and labiolingual loading). Ten conditions were generated and intrusive forces of 100 g were applied totally. Then displacement tendency of the maxillary anterior teeth and CA, and stress of the PDLs were analyzed. RESULTS: For the central incisor under condition L14 and for the canine under conditions L11, L13, L23, and L33, the intrusion amount was negative. Under other conditions, the intrusion amount was positive. The labiolingual angulation of maxillary anterior teeth exhibited positive changes under all conditions, with greater changes under linguoincisal loading. The mesiodistal angulation of canine exhibited positive changes under labial loading, while negative changes under linguoincisal loading except for condition L14. CONCLUSIONS: The intrusion amount, labiolingual and mesiodistal angulations of the maxillary anterior teeth were affected by the mini-implant site, loading site, and loading mode. Labial and linguoincisal loading may have opposite effects on the intrusion amount of maxillary anterior teeth and the mesiodistal angulation of canine. The labiolingual angulation of the maxillary incisors would increase under all intrusion modes, with greater increases under linguoincisal loading.

Double vision: 2D and 3D mosquito trajectories can be as valuable for behaviour analysis via machine learning.

BACKGROUND: Mosquitoes are carriers of tropical diseases, thus demanding a comprehensive understanding of their behaviour to devise effective disease control strategies. In this article we show that machine learning can provide a performance assessment of 2D and 3D machine vision techniques and thereby guide entomologists towards appropriate experimental approaches for behaviour assessment. Behaviours are best characterised via tracking-giving a full time series of information. However, tracking systems vary in complexity. Single-camera imaging yields two-component position data which generally are a function of all three orthogonal components due to perspective; however, a telecentric imaging setup gives constant magnification with respect to depth and thereby measures two orthogonal position components. Multi-camera or holographic techniques quantify all three components. METHODS: In this study a 3D mosquito mating swarm dataset was used to generate equivalent 2D data via telecentric imaging and a single camera at various imaging distances. The performance of the tracking systems was assessed through an established machine learning classifier that differentiates male and non-male mosquito tracks. SHAPs analysis has been used to explore the trajectory feature values for each model. RESULTS: The results reveal that both telecentric and single-camera models, when placed at large distances from the flying mosquitoes, can produce equivalent accuracy from a classifier as well as preserve characteristic features without resorting to more complex 3D tracking techniques. CONCLUSIONS: Caution should be exercised when employing a single camera at short distances as classifier balanced accuracy is reduced compared to that from 3D or telecentric imaging; the trajectory features also deviate compared to those from the other datasets. It is postulated that measurement of two orthogonal motion components is necessary to optimise the accuracy of machine learning classifiers based on trajectory data. The study increases the evidence base for using machine learning to determine behaviours from insect trajectory data.

Automation of Wilms' tumor segmentation by artificial intelligence.

BACKGROUND: 3D reconstruction of Wilms' tumor provides several advantages but are not systematically performed because manual segmentation is extremely time-consuming. The objective of our study was to develop an artificial intelligence tool to automate the segmentation of tumors and kidneys in children. METHODS: A manual segmentation was carried out by two experts on 14 CT scans. Then, the segmentation of Wilms' tumor and neoplastic kidney was automatically performed using the CNN U-Net and the same CNN U-Net trained according to the OV2ASSION method. The time saving for the expert was estimated depending on the number of sections automatically segmented. RESULTS: When segmentations were performed manually by two experts, the inter-individual variability resulted in a Dice index of 0.95 for tumor and 0.87 for kidney. Fully automatic segmentation with the CNN U-Net yielded a poor Dice index of 0.69 for Wilms' tumor and 0.27 for kidney. With the OV2ASSION method, the Dice index varied depending on the number of manually segmented sections. For the segmentation of the Wilms' tumor and neoplastic kidney, it varied respectively from 0.97 to 0.94 for a gap of 1 (2 out of 3 sections performed manually) to 0.94 and 0.86 for a gap of 10 (1 section out of 6 performed manually). CONCLUSION: Fully automated segmentation remains a challenge in the field of medical image processing. Although it is possible to use already developed neural networks, such as U-Net, we found that the results obtained were not satisfactory for segmentation of neoplastic kidneys or Wilms' tumors in children. We developed an innovative CNN U-Net training method that makes it possible to segment the kidney and its tumor with the same precision as an expert while reducing their intervention time by 80%.

Dynamic computed tomography angiography for noninvasive diagnosis of bow Hunter's syndrome: a case report.

The focus of this case report is to technically describe a noninvasive diagnostic evaluation of bow Hunter's syndrome using a dynamic computed tomography angiography protocol and discuss its advantages. In addition, we aimed to exemplify the quality of the study by presenting images of a 3D-printed model generated to help plan the surgical treatment for the patient. The dynamic computed tomography angiography protocol consisted of a first image acquisition with the patient in the anatomic position of the head and neck. This was followed by a second acquisition with the head and neck rotated to the side that triggered the symptoms, with technical parameters similar to the first acquisition. The acquired images were used to print a 3D model to better depict the findings for the surgical team. The dynamic computed tomography angiography protocol developed in this study helped visualize the vertebrobasilar arterial anatomy, detect vertebral artery stenosis produced by head and neck rotation, depict the structure responsible for artery stenosis (e.g., bony structure or membranes), and study possible complications of the disease (e.g., posterior cerebral circulation infarction). Additionally, the 3D-printed model better illustrated the findings of stenosis, aiding in surgical planning. In conclusion, dynamic computed tomography angiography for the evaluation of bow Hunter's syndrome is a feasible noninvasive technique that can be used as an alternative to traditional diagnostic methods.

Changes in femoral anteversion after intramedullary nailing for pediatric femoral shaft fracture: a multicenter study.

BACKGROUND: The rotational change after using a flexible intramedullary (IM) nail for femoral shaft fractures has been a concern for many surgeons. Recently, a statistical shape model (SSM) was developed for the three-dimensional reconstruction of the femur from two-dimensional plain radiographs. In this study, we measured postoperative femoral anteversion (FAV) in patients diagnosed with femoral shaft fractures who were treated with flexible IM nails and investigated age-related changes in FAV using the SSM. METHODS: This study used radiographic data collected from six regional tertiary centers specializing in pediatric trauma in South Korea. Patients diagnosed with femoral shaft fractures between September 2002 and June 2020 and patients aged < 18 years with at least two anteroposterior (AP) and lateral (LAT) femur plain radiographs obtained at least three months apart were included. A linear mixed model (LMM) was used for statistical analysis. RESULTS: Overall, 72 patients were included in the study. The average patient age was 7.6 years and the average follow-up duration was 6.8 years. The average FAV of immediate postoperative images was 27.5 ± 11.5°. Out of 72 patients, 52 patients (72.2%) showed immediate postoperative FAV greater than 20°. The average FAV in patients with initial FAV > 20° was 32.74°, and the LMM showed that FAV decreased by 2.5° (p = 0.0001) with each 1-year increase from the time of initial trauma. CONCLUSIONS: This study explored changes in FAV after femoral shaft fracture using a newly developed technology that allows 3D reconstruction from uncalibrated 2D images. There was a pattern of change on the rotation of the femur after initial fixation, with a 2.5° decrease of FAV per year.

An Automatic Method for Elbow Joint Recognition, Segmentation and Reconstruction.

Elbow computerized tomography (CT) scans have been widely applied for describing elbow morphology. To enhance the objectivity and efficiency of clinical diagnosis, an automatic method to recognize, segment, and reconstruct elbow joint bones is proposed in this study. The method involves three steps: initially, the humerus, ulna, and radius are automatically recognized based on the anatomical features of the elbow joint, and the prompt boxes are generated. Subsequently, elbow MedSAM is obtained through transfer learning, which accurately segments the CT images by integrating the prompt boxes. After that, hole-filling and object reclassification steps are executed to refine the mask. Finally, three-dimensional (3D) reconstruction is conducted seamlessly using the marching cube algorithm. To validate the reliability and accuracy of the method, the images were compared to the masks labeled by senior surgeons. Quantitative evaluation of segmentation results revealed median intersection over union (IoU) values of 0.963, 0.959, and 0.950 for the humerus, ulna, and radius, respectively. Additionally, the reconstructed surface errors were measured at 1.127, 1.523, and 2.062 mm, respectively. Consequently, the automatic elbow reconstruction method demonstrates promising capabilities in clinical diagnosis, preoperative planning, and intraoperative navigation for elbow joint diseases.

Learning Temporal-Spatial Contextual Adaptation for Three-Dimensional Human Pose Estimation.

Three-dimensional human pose estimation focuses on generating 3D pose sequences from 2D videos. It has enormous potential in the fields of human-robot interaction, remote sensing, virtual reality, and computer vision. Existing excellent methods primarily focus on exploring spatial or temporal encoding to achieve 3D pose inference. However, various architectures exploit the independent effects of spatial and temporal cues on 3D pose estimation, while neglecting the spatial-temporal synergistic influence. To address this issue, this paper proposes a novel 3D pose estimation method with a dual-adaptive spatial-temporal former (DASTFormer) and additional supervised training. The DASTFormer contains attention-adaptive (AtA) and pure-adaptive (PuA) modes, which will enhance pose inference from 2D to 3D by adaptively learning spatial-temporal effects, considering both their cooperative and independent influences. In addition, an additional supervised training with batch variance loss is proposed in this work. Different from common training strategy, a two-round parameter update is conducted on the same batch data. Not only can it better explore the potential relationship between spatial-temporal encoding and 3D poses, but it can also alleviate the batch size limitations imposed by graphics cards on transformer-based frameworks. Extensive experimental results show that the proposed method significantly outperforms most state-of-the-art approaches on Human3.6 and HumanEVA datasets.

Automatic 3D reconstruction of vertebrae from orthogonal bi-planar radiographs.

When conducting spine-related diagnosis and surgery, the three-dimensional (3D) upright posture of the spine under natural weight bearing is of significant clinical value for physicians to analyze the force on the spine. However, existing medical imaging technologies cannot meet current requirements of medical service. On the one hand, the mainstream 3D volumetric imaging modalities (e.g. CT and MRI) require patients to lie down during the imaging process. On the other hand, the imaging modalities conducted in an upright posture (e.g. radiograph) can only realize 2D projections, which lose the valid information of spinal anatomy and curvature. Developments of deep learning-based 3D reconstruction methods bring potential to overcome the limitations of the existing medical imaging technologies. To deal with the limitations of current medical imaging technologies as is described above, in this paper, we propose a novel deep learning framework, ReVerteR, which can realize automatic 3D Reconstruction of Vertebrae from orthogonal bi-planar Radiographs. With the utilization of self-attention mechanism and specially designed loss function combining Dice, Hausdorff, Focal, and MSE, ReVerteR can alleviate the sample-imbalance problem during the reconstruction process and realize the fusion of the centroid annotation and the focused vertebra. Furthermore, aiming at automatic and customized 3D spinal reconstruction in real-world scenarios, we extend ReVerteR to a clinical deployment-oriented framework, and develop an interactive interface with all functions in the framework integrated so as to enhance human-computer interaction during clinical decision-making. Extensive experiments and visualization conducted on our constructed datasets based on two benchmark datasets of spinal CT, VerSe 2019 and VerSe 2020, demonstrate the effectiveness of our proposed ReVerteR. In this paper, we propose an automatic 3D reconstruction method of vertebrae based on orthogonal bi-planar radiographs. With the 3D upright posture of the spine under natural weight bearing effectively constructed, our proposed method is expected to better support doctors make clinical decision during spine-related diagnosis and surgery.

Classification of variant portal vein anatomy based on three-dimensional CT: surgical implications.

PURPOSES: The purpose of this study was to develop a new and more comprehensive classification system for portal vein (PV) variations using three-dimensional visualization and evaluation (3DVE) and to discuss the prevalence rates and clinical implications of the variants. METHODS: The anatomies of PVs were tracked and analyzed by using three-dimensional visualization of CT images acquired between 2013 and 2022. Scans from 200 adults were evaluated and a total of 178 patients (N = 178) were included in the study. The new classification system, named BLB classification, was developed based on the level of the absent PV branch in each variant anatomy. RESULTS: Using the BLB classification system, PVs were divided into thirteen subtypes. Only 82.6-84.8% of the portal veins of the 178 patients were depicted in Atri's, Cheng's or Covey's classification, compared with 100% identified by the BLB classification. The BLB classification was validated against external data sets from previous studies, with 97.0-98.9% of patients classified by the BLB system. CONCLUSION: Variant PV anatomies are more commonly seen based on 3DVE than in previous reports. The BLB classification covers almost all portal vein variants and may be used for planning liver surgery.

Development and clinical validation of a deep learning-based knee CT image segmentation method for robotic-assisted total knee arthroplasty.

BACKGROUND: This study aimed to develop a novel deep convolutional neural network called Dual-path Double Attention Transformer (DDA-Transformer) designed to achieve precise and fast knee joint CT image segmentation and to validate it in robotic-assisted total knee arthroplasty (TKA). METHODS: The femoral, tibial, patellar, and fibular segmentation performance and speed were evaluated and the accuracy of component sizing, bone resection and alignment of the robotic-assisted TKA system constructed using this deep learning network was clinically validated. RESULTS: Overall, DDA-Transformer outperformed six other networks in terms of the Dice coefficient, intersection over union, average surface distance, and Hausdorff distance. DDA-Transformer exhibited significantly faster segmentation speeds than nnUnet, TransUnet and 3D-Unet (p < 0.01). Furthermore, the robotic-assisted TKA system outperforms the manual group in surgical accuracy. CONCLUSIONS: DDA-Transformer exhibited significantly improved accuracy and robustness in knee joint segmentation, and this convenient and stable knee joint CT image segmentation network significantly improved the accuracy of the TKA procedure.

A human activity recognition method based on Vision Transformer.

Human activity recognition has a wide range of applications in various fields, such as video surveillance, virtual reality and human-computer intelligent interaction. It has emerged as a significant research area in computer vision. GCN (Graph Convolutional networks) have recently been widely used in these fields and have made great performance. However, there are still some challenges including over-smoothing problem caused by stack graph convolutions and deficient semantics correlation to capture the large movements between time sequences. Vision Transformer (ViT) is utilized in many 2D and 3D image fields and has surprised results. In our work, we propose a novel human activity recognition method based on ViT (HAR-ViT). We integrate enhanced AGCL (eAGCL) in 2s-AGCN to ViT to make it process spatio-temporal data (3D skeleton) and make full use of spatial features. The position encoder module orders the non-sequenced information while the transformer encoder efficiently compresses sequence data features to enhance calculation speed. Human activity recognition is accomplished through multi-layer perceptron (MLP) classifier. Experimental results demonstrate that the proposed method achieves SOTA performance on three extensively used datasets, NTU RGB+D 60, NTU RGB+D 120 and Kinetics-Skeleton 400.