3D Datasets Research Articles

Urban Power Line Extraction from Mobile LiDAR Point Clouds based on Local Line-plane seperation Model and Global Refinement

Abstract Power line extraction from three-dimensional (3D) point clouds is a key step for power inspection. However, the power line is intertwined and covered with trees and other objects in the urban area, which brings a tremendous challenge to its extraction. This article proposed a local-to-global method to extract power lines from mobile LiDAR point clouds based on multiple characteristics. Firstly, initial point clouds clusters are selected through data preprocessing. Secondly, given the linearity and independence of the power line, we innovatively design a local line-plane separation modeling to divide the point clouds into high-confidence power line points, non-power line points, and unconfirmed points. Then, a novel global energy function is designed for the graph cut model according to the long linear and short radius characteristics of the power line. The proposed method has been tested on Toronto 3D and WuHan Avenue datasets. The average Precision, Recall, and F1score can reach up to 0.92, 0.73, and 0.79 respectively, which is superior to state-of-the-art approaches. The result demonstrates the efficacy and robustness of our proposed approach in extracting power lines within urban environments using a mobile laser scanning system.

EMS3D-KITTI: Synthetic 3D dataset in KITTI format with a fair distribution of Emergency Medical Services vehicles for autodrive AI model training.

Contemporary research in 3D object detection for autonomous driving primarily focuses on identifying standard entities like vehicles and pedestrians. However, the need for large, precisely labelled datasets limits the detection of specialized and less common objects, such as Emergency Medical Service (EMS) and law enforcement vehicles. To address this, we leveraged the Car Learning to Act (CARLA) simulator to generate and fairly distribute rare EMS vehicles, automatically labelling these objects in 3D point cloud data. This enriched dataset, organized in the KITTI 3D object detection benchmark format by the Karlsruhe Institute of Technology and the Toyota Technological Institute, improves its utility for training and evaluating autonomous vehicle systems. To bridge the gap between simulated and real-world scenarios, our methodology integrates a wide range of scenarios simulation in CARLA, including variations in weather conditions, human presence, and different environmental settings. This approach enhances the realism and robustness of the dataset, making it more applicable to practical autonomous driving scenarios. The data provided in this article offers a valuable resource for researchers, industry professionals, and stakeholders interested in advancing autonomous vehicle technologies and improving emergency vehicle detection. Furthermore, this dataset contributes to broader efforts in road safety and the development of AI systems capable of handling specialized vehicle identification in real-world applications.

Tricuspid valve annulus remodeling and regurgitation analysis with 3-D transesophageal echocardiography after percutaneous transcatheter edge-to-edge repair

Abstract Introduction Tricuspid transcatheter edge-to-edge repair (Tr-TEER) is becoming an effective treatment for severe tricuspid regurgitation (TrR) in symptomatic patients who are high-risk candidates for surgery. Purpose Aims of the study were to assess immediate changes in tricuspid valve (TV) anatomy following Tr-TEER and to investigate the relationship between baseline geometry and residual intraoperative TrR. Methods Thirty-one consecutive patients (average age 77.3 years) with symptomatic severe TrR (≥ grade 4) underwent Tr-TEER. Intraoperative transesophageal echocardiographic 2D and 3D images were obtained before and immediately after the procedure. 3D datasets were analyzed offline using specialized software for 3D TV morphology analysis. Results All patients successfully underwent Tr-TEER: 24 patients (77%) achieved an optimal result with residual regurgitation ≤ 2, while 7 patients (23%) had a residual grade 3 or 4. The 3D datasets were suitable for analysis in 29 patients. Anatomic parameters of the TV, obtained via intraoperative transesophageal echocardiography before and after Tr-TEER, are presented in Table 1. The 2D and 3D annulus area, perimeter, diameter in the 2- and 4-chamber view, and minor and major diameter of the tricuspid annulus significantly decreased post-procedure. A significant reduction in TrR was noted regardless of annulus changes. Conclusions Tr-TEER is linked to a significant decrease in TrR and notable changes in TV geometry, particularly a substantial reduction in annulus area and perimeter. This may contribute to the effectiveness of Tr-TEER in promptly reducing TrR.

Tricuspid leaflet identification on 2D multiplane transesophageal echocardiography: derivation and validation of a simple two-step approach

Abstract Background Tricuspid valve (TV) leaflet identification by two-dimensional (2D) transesophageal echocardiography (TOE) is complex due to the considerable variability in the visualized leaflets. While three-dimensional (3D) imaging can easily overcome this problem by acquiring the whole valve anatomy, the overall image quality is often suboptimal and the technique somewhat unsuitable for real-time monitoring during interventional procedures like TriClip implantation. Using information derived from both 2D and multiplanar reconstruction of 3D datasets, we elaborated a step-wise approach to identify TV leaflets during multiplane 2D TOE and tested it on a validation cohort. Methods TOE data from 83 patients was retrospectively collected. The 3D data of each patient was appropriately sliced to simulate and accurately match what would be seen by "sweeping" the cursor across the TV during real-time multiplane imaging (Fig. 1). Following data analysis, a "two-step" reference table was created and subsequently evaluated for accuracy on a validation cohort of 41 patients from a distinct hospital and country. To evaluate the usability of the approach, 11 echocardiographers were presented with a random selection of 5 multiplane images. They were then instructed to identify the visualized leaflets initially without using the reference table, and subsequently with its assistance. Results After removal of redundant information, 231 unique multiplane sets were obtained. Combining anatomical features (number of leaflets visualized and presence of the interventricular septum) from the 2D views allowed easy and consistent identification of TV leaflets (Fig. 2), as confirmed by correlation with the 3D dataset. The same "two-step" approach was subsequently tested on an independent validation cohort of 41 patients (96 unique multiplane sets). Accurate leaflet identification was possible in 35 patients (85%), while in 6 cases (15%) the image sector was too narrow to confidently assess the presence of the interventricular septum. The accuracy in exact leaflet recognition without the aid of the reference table was rather modest (mean score of 1 out of 5) among echocardiographers with different experience in TV TOE, but significantly increased (mean score of 3.2 out of 5) with the adoption of the proposed approach, accounting for a 44% improvement in accuracy (p< 0.001). Conclusion Simple anatomical features allow precise TV leaflet identification during multiplane 2D TOE. A streamlined "two-step" approach, validated on an independent cohort of patients, significantly improves accuracy in leaflet recognition among both beginner and advanced echocardiographers.

Right ventricular volumes and function indices by automated 3D speckle tracking echocardiography with artificial intelligence software

Abstract Introduction Normal values of three dimensional right ventricular (RV) volumes, ejection fraction (EF) and systolic function indices by three - dimensional speckle tracking echcardiography (3D - STE) are not well established. All measurements relied on older 3D RV software analysis and most of these are operator - depended with a wide variability in measured values. The gold standard method for evaluation of RV volumes and EF remains the Magnetic Resonance Imaging (MRI)despite the rapid improvement in echocardiography softwares for RV evaluation. Purpose Aim of the study was to assess the feasibility of modern advanced technology echocardiographic software based on 3D imaging powered by Artificial Intelligence (AI) in measurement of RV volumes, RV EF and systolic function indices by 3D - STE in normal subjects. Methods Healthy subjects without any cardiovascular, pulmonary and renal disease, underwent complete echocardiographic evaluation including the 3D datasets from each of the parasternal, apical and subcostal views. 3D RV datasets were obtained and analyzed using RV software powered by AI, to measure Global RV Volumes, End - Diastolic volume (EDV), End - Systolic volume (ESV), RV EF and Tricuspid Annular Plane Systolic Excursion (TAPSE). Additionally 3D Free Wall Longitudinal Strain (FWLS) and 3D Global Strain (3D GS) were obtained. Three or more datasets obtained from the apical view were available for comprehensive analysis in each subject. Results Of the 54 healthy subjects, 40 (50% male and 50% female, mean age 51 +/- 15 years old) had adequate image quality for measurements. Mean indexed 3D EDV, ESV and RV EF were measured at 48.5 +/- 10.3 mL/m2, 16.7 +/- 5.8 mL/m2, 56,1 +/- 5.9 % respectively. in addition 3D FWLS and 3D GS were measured at - 27.9 +/- 4.5 % and - 24.7 +/- 3.6 % respectively. TAPSE was measured at 22.4 +/- 5.1 mm. Conclusions Modern softwares based on AI are a new era in Cardiology and Echocardiography. A rapid increase of AI contribution in cardiovascular imaging techiniques is expected in the near future. 3D RV acquisition is feasible in the majority of subjects, although the quality of imaging and appropriate echocardiographic views remains operator - depended. Reference values of 3D RV volumes and 3D - STE systolic function indices were obtained by advanced technology AI software. Larger studies are required to establish normal values for the above indices.

Acute tricuspid valve annulus remodeling after transcatheter edge-to-edge repair with Triclip

Abstract Background Right ventricular (RV) adverse remodeling together with RV dysfunction, and not only tricuspid regurgitation (TR) severity, are independently associated with survival in patients with medically treated significant TR. Nevertheless, changes in RV dimensions and function after transcatheter tricuspid valve intervention and their association with outcomes remain largely unexplored, as well as tricuspid valve annulus (TVA) reverse remodeling. Aim of our study is to assess acute remodeling of TVA after tricuspid transcatheter edge-to-edge repair (T-TEER) with the TriClip device (Abbott Vascular, Santa Clara, CA, USA) using intraprocedural trans-esophageal echocardiography (TEE). Materials and methods This prospective, single-center study, conducted at Fondazione Policlinico Universitario Campus Bio-Medico of Rome, enrolled 60 consecutive symptomatic patients referred with the diagnosis of at least severe TR who underwent T-TEER from March 2021 to June 2024 with the TriClip System. Intraprocedural TEE was performed using Siemens Acuson SC2000 (Siemens Medical Systems, Pleasanton, CA, USA) at the beginning, and then using Philips EpiQ CVx (Philips Medical System, Andover, MA). A full-volume 3D echocardiographic data set was acquired during the procedure for each patient before and after the grasping. The following parameters were assessed using a multiplanar reconstruction analysis performed off-line using a 3D dataset: septal-lateral (SL) and antero-posterior (AP) annulus diameters; annulus area; annulus perimeter and eccentricity index. Results The mean age of the study cohort was 77,77 ± 6,18 years, and 77% were female. Twenty-eight patients (46,7%) had severe (3+), twenty-four patients (40%) massive (4+) and eight patients (13,3%) torrential (5+) tricuspid regurgitation. The majority of them had atrial functional tricuspid regurgitation, mostly due to annular dilatation. The acute procedural success was achieved in 82.22%. We observed an acute reduction in SL (from a median of 45 to 38 mm, p<0.0001), AP (from a median of 47.5 to 45.8 mm, p<0.0001), area (from a median of 18.4 to 16.2 cm2, p<0.0001), perimeter (from a median of 149 to 141 mm, p<0.0001) and eccentricity index (from 0.92 to 0.87, p <0.0001). The TV annulus was larger in patients with residual TR >2+. Analysis of the subgroups according to procedural success showed an acute inverse remodeling of the TV annulus independent of residual TR grade. We observed a significative reduction of TVA three-dimensional measurements immediately after the procedure. Conclusions In our study we observed that the reduction of TVA begins immediately after TriClip implantation, proving that T-TEER is effective in reducing TR and in inducing reverse remodeling of the TV apparatus. Concurrently, the baseline TV geometry influences the procedural results. Our data are gathered during T-TEER procedure, follow up data are needed for monitoring right ventricular performance. tva measures using mpr tr grade at baseline and after t-teer

Exploring three-dimensional right ventricular shape and function in tetralogy of fallot patients

Abstract Background Tetralogy of Fallot (ToF) accounts for approximately 7 to 10 % of all CHD and is commonly associated with failure of the right ventricle (RV). Parameters providing more insight into the mechanics of RV failure are needed to guide clinical decision making. As advancements in Three-Dimensional (3D) echocardiography (3DE) have enabled more detailed analysis of RV volume and function, the way is paved for new 3D derived parameters. We developed a custom software to assess 3D RV shape and function. Purpose The aim of this study was to evaluate the differences of 3DE derived RV geometry (curvature) and deformation parameters in adults with ToF and controls. Methods In 50 patients with ToF and 50 controls 3DE RV images were acquired. RV geometries from 3D datasets were segmented using commercially available RV analysis software (TomTec 4D-RV function 2.0) and processed using custom software (RV-dynamics) to calculate Curvature, Area Strain (AS) and Longitudinal Strain (LS). Parameters were measured globally and over different segments of the RV (i.e. septal body (SB), inferior free wall (IFW), anterior free wall (AFW), lateral free wall (LFW), apex (Ax), RV inflow tract (RVIT), RV outflow tract (RVOT), anterior boundary (AB) and posterior boundary (PB)). Results Globally and for all independent regions of the RV, AS was significantly reduced in patients with ToF as compared with healthy controls. Except for AFW, global and all regional values of LS were significantly reduced in patients with ToF as compared with healthy controls. Global AS and LS were strongly correlated with RVEF with values for r of -0.9 and -0.6 respectively. RV curvature differed over multiple segments, predominantly in the end-diastolic Ax and end-systolic PB and Ax regions. Conclusion The RV-dynamics software gives the opportunity to analyze 3D RV morphology and RV deformation parameters in the diseased RV in more detail. AS and LS may provide more detailed assessment of the functional condition of the diseased RV in ToF whereas curvature provides an interesting method for the quantification of RV remodeling. Further insight in the different mechanisms of dysfunction of the RV may improve clinical decision making.

Spherical harmonics texture extraction for versatile analysis of biological objects.

The characterization of phenotypes in cells or organisms from microscopy data largely depends on differences in the spatial distribution of image intensity. Multiple methods exist for quantifying the intensity distribution - or image texture - across objects in natural images. However, many of these texture extraction methods do not directly adapt to 3D microscopy data. Here, we present Spherical Texture extraction, which measures the variance in intensity per angular wavelength by calculating the Spherical Harmonics or Fourier power spectrum of a spherical or circular projection of the angular mean intensity of the object. This method provides a 20-value characterization that quantifies the scale of features in the spherical projection of the intensity distribution, giving a different signal if the intensity is, for example, clustered in parts of the volume or spread across the entire volume. We apply this method to different systems and demonstrate its ability to describe various biological problems through feature extraction. The Spherical Texture extraction characterizes biologically defined gene expression patterns in Drosophila melanogaster embryos, giving a quantitative read-out for pattern formation. Our method can also quantify morphological differences in Caenorhabditis elegans germline nuclei, which lack a predefined pattern. We show that the classification of germline nuclei using their Spherical Texture outperforms a convolutional neural net when training data is limited. Additionally, we use a similar pipeline on 2D cell migration data to extract polarization direction, quantifying the alignment of fluorescent markers to the migration direction. We implemented the Spherical Texture method as a plugin in ilastik to provide a parameter-free and data-agnostic application to any segmented 3D or 2D dataset. Additionally, this technique can also be applied through a Python package to provide extra feature extraction for any object classification pipeline or downstream analysis.

3D is superior to 2D echocardiography in identifying high risk patients with heart failure with reduced ejection fraction and indication for device therapy

Abstract Background In patients with heart failure with reduced ejection fraction (HFrEF), a left ventricular ejection fraction (LVEF) less than 35% has been identified as a significant risk factor for increased mortality. As a result, all studies assessing the indications and outcomes of implantable cardioverter-defibrillators and cardiac resynchronization therapy in HFrEF used this cut-off, measured by 2-dimensional echocardiography (2DE). However, many patients with HFrEF and a LVEF >35% by 2DE, who do not qualify for device therapy, still experience significant mortality. Purpose To investigate whether, in patients hospitalized with both ischemic and non-ischemic HFrEF, the LVEF measured by 3-dimensional echocardiography (3DE) could identify better patients at high risk of death, thereby indicating the need for device therapy. Methods 322 consecutive patients (60±11 years, 72% men), hospitalized for HFrEF or HFmrEF, were assessed using both 2DE and 3DE, including full-volume acquisitions of the LV. From 2D datasets, LV end-systolic (LVESV) and LV end-diastolic (LVEDV) volumes, and LVEF were estimated. From the 3D volumes, the same parameters were measured. Patients were discharged on optimal HF treatment, according to current ESC guidelines, and were followed for 24±8 months after the initial event. The primary outcome was cardiovascular death (CD). Results At admission, 43% patients were in NYHA class II, 25% in NYHA class III, and 33% in NYHA class IV. There was a similar burden of ischemic and non-ischemic etiology (52% vs 48%). Mean 2D LVEF was 31±11%, while mean 3D LVEF was 30±10% (p=NS). There were 62 CD (19%) during follow-up. Binary logistic regression for the 2D and 3DE parameters showed that only 3D LVEF was a significant predictor of death (β=-0.04, OR 0.96, 95%CI 0.93-0.98, p=0.003). By 2DE, 206 patients (64%) had LVEF <35%, of whom 47 patients died (23%) and 159 patients survived (77%), whereas by 3DE, 215 patients (67%) had LVEF <35%, of whom 60 patients died (28%) and 155 patients survived (72%). Consequently, 2D LVEF <35% identified 47 out of 62 deaths (76%), whereas 3D LVEF <35% identified 60 out of 62 deaths (97%). By ROC Analysis, 3DE was superior to 2DE for prediction of survival: AUC 0.62, p=0.03 for 3D LVEF vs. AUC 0.56, p=0.09 for 2D LVEF. Kaplan-Meier analysis (Figure) showed that 2D LVEF <35% discriminated between survivors and non-survivors with a Chi-square of 6.8, p=0.01, whereas 3D LVEF <35% discriminated between survivors and non-survivors with a Chi-square of 30.3, p<0.001. Conclusion 3DE assessment of LVEF and its cut-off of 35% enhances the identification of patients with HFrEF at high risk of death. This suggests that 3DE should be utilized for risk stratification in HFrEF, potentially leading to more accurate indications for device therapy. Survival according to 2D vs 3D LVEF

Robotic acquisition of echocardiographic images, a proof of concept and safety study

Abstract Background/Introduction Though echocardiography remains the most utilized imaging modality in daily cardiology practice, a major challenge remains the physical strain and injury that can be induced in the echocardiographer. Purpose We developed a prototype robotic arm assisted echocardiography system which incorporates robotics with traditional echocardiography. We assessed the safety and feasibility of the system in 31 volunteers. Methods An ultrasound transducer was mounted upon a robotic arm (UR10 robotic arm, Universal Robots, Odense, Denmark). The transducer was connected to a Philips EPIQ 7C system. The position of the transducer was modified by the echocardiographer who could manipulate a remote stick controller in their hand whilst orientating its position in relation to the real-time images acquired upon the monitor of the echocardiography scanner. This enabled the sonographer to optimize the images and to initiate recordings of parasternal and apical 2D and 3D images. The maximal force on the robot arm was set to 5 Newtons (N) which was deemed to be both safe and to allow for adequate image quality. The force induced by the probe upon the chest wall was measured continuously. Acquired images were assessed for quality ranging from excellent to poor. Results 17 males and 14 females were scanned. The number of images acquired varied from a minimum of 16 and a maximum of 61. The mean time taken to scan an individual was 17.4 minutes with a standard deviation (SD) of 9.9 minutes. This equated to an average of 1.9 images per minute (SD 0.7). No statistical difference was noted between the number of images per minute obtained between male and female participants (mean (+/- SD) for males was 2.1(0.8) and for females 1.8 (0.7) P=0.27). The average force that was applied by the probe was 3.95+/-0.83 Newtons. None of the volunteers reported significant discomfort in the acquisition of the images. Of the 3D images obtained 5 of the participants had poor apical 3D datasets and 3 had poor parasternal datasets but the remaining images acquired were adequate for diagnostic quality. Conclusion(s) Our study demonstrates the feasibility and safety of acquiring 2D and 3D images using a prototype robotic arm assisted echocardiography system. Despite using a low force to the chest wall, the majority of images obtained were of diagnostic quality. Future studies will assess the techniques validity within a routine patient population compared to standard imaging performed by an experienced sonographer. Set up of robot-arm echocardioghy

Leveraging two-dimensional pre-trained vision transformers for three-dimensional model generation via masked autoencoders

Although the Transformer architecture has established itself as the industry standard for jobs involving natural language processing, it still has few uses in computer vision. In vision, attention is used in conjunction with convolutional networks or to replace individual convolutional network elements while preserving the overall network design. Differences between the two domains, such as significant variations in the scale of visual things and the higher granularity of pixels in images compared to words in the text, make it difficult to transfer Transformer from language to vision. Masking autoencoding is a promising self-supervised learning approach that greatly advances computer vision and natural language processing. For robust 2D representations, pre-training with large image data has become standard practice. On the other hand, the low availability of 3D datasets significantly impedes learning high-quality 3D features because of the high data processing cost. We present a strong multi-scale MAE prior training architecture that uses a trained ViT and a 3D representation model from 2D images to let 3D point clouds learn on their own. We employ the adept 2D information to direct a 3D masking-based autoencoder, which uses an encoder-decoder architecture to rebuild the masked point tokens through self-supervised pre-training. To acquire the input point cloud’s multi-view visual characteristics, we first use pre-trained 2D models. Next, we present a two-dimensional masking method that preserves the visibility of semantically significant point tokens. Numerous tests demonstrate how effectively our method works with pre-trained models and how well it generalizes to a range of downstream tasks. In particular, our pre-trained model achieved 93.63% accuracy for linear SVM on ScanObjectNN and 91.31% accuracy on ModelNet40. Our approach demonstrates how a straightforward architecture solely based on conventional transformers may outperform specialized transformer models from supervised learning.

Improved BCI calibration in multimodal emotion recognition using heterogeneous adversarial transfer learning.

The use of brain-computer interface (BCI) technology to identify emotional states has gained significant interest, especially with the rise of virtual reality (VR) applications. However, the extensive calibration required for precise emotion recognition models presents a significant challenge, particularly for sensitive groups such as children, elderly, and patients. This study presents a novel approach that utilizes heterogeneous adversarial transfer learning (HATL) to synthesize electroencephalography (EEG) data from various other signal modalities, reducing the need for lengthy calibration phases. We benchmark the efficacy of three generative adversarial network (GAN) architectures, such as conditional GAN (CGAN), conditional Wasserstein GAN (CWGAN), and CWGAN with gradient penalty (CWGAN-GP) within this framework. The proposed framework is rigorously tested on two conventional open sourced datasets, SEED-V and DEAP. Additionally, the framework was applied to an immersive three-dimensional (3D) dataset named GraffitiVR, which we collected to capture the emotional and behavioral reactions of individuals experiencing urban graffiti in a VR environment. This expanded application provides insights into emotion recognition frameworks in VR settings, providing a wider range of contexts for assessing our methodology. When the accuracy of emotion recognition classifiers trained with CWGAN-GP-generated EEG data combined with non-EEG sensory data was compared against those trained using a combination of real EEG and non-EEG sensory data, the accuracy ratios were 93% on the SEED-V dataset, 99% on the DEAP dataset, and 97% on the GraffitiVR dataset. Moreover, in the GraffitiVR dataset, using CWGAN-GP-generated EEG data with non-EEG sensory data for emotion recognition models resulted in up to a 30% reduction in calibration time compared to classifiers trained on real EEG data with non-EEG sensory data. These results underscore the robustness and versatility of the proposed approach, significantly enhancing emotion recognition processes across a variety of environmental settings.

Segment Any Leaf 3D: A Zero-Shot 3D Leaf Instance Segmentation Method Based on Multi-View Images.

Exploring the relationships between plant phenotypes and genetic information requires advanced phenotypic analysis techniques for precise characterization. However, the diversity and variability of plant morphology challenge existing methods, which often fail to generalize across species and require extensive annotated data, especially for 3D datasets. This paper proposes a zero-shot 3D leaf instance segmentation method using RGB sensors. It extends the 2D segmentation model SAM (Segment Anything Model) to 3D through a multi-view strategy. RGB image sequences captured from multiple viewpoints are used to reconstruct 3D plant point clouds via multi-view stereo. HQ-SAM (High-Quality Segment Anything Model) segments leaves in 2D, and the segmentation is mapped to the 3D point cloud. An incremental fusion method based on confidence scores aggregates results from different views into a final output. Evaluated on a custom peanut seedling dataset, the method achieved point-level precision, recall, and F1 scores over 0.9 and object-level mIoU and precision above 0.75 under two IoU thresholds. The results show that the method achieves state-of-the-art segmentation quality while offering zero-shot capability and generalizability, demonstrating significant potential in plant phenotyping.

3D genomic features across >50 diverse cell types reveal insights into the genomic architecture of childhood obesity.

The prevalence of childhood obesity is increasing worldwide, along with the associated common comorbidities of type 2 diabetes and cardiovascular disease in later life. Motivated by evidence for a strong genetic component, our prior genome-wide association study (GWAS) efforts for childhood obesity revealed 19 independent signals for the trait; however, the mechanism of action of these loci remains to be elucidated. To molecularly characterize these childhood obesity loci, we sought to determine the underlying causal variants and the corresponding effector genes within diverse cellular contexts. Integrating childhood obesity GWAS summary statistics with our existing 3D genomic datasets for 57 human cell types, consisting of high-resolution promoter-focused Capture-C/Hi-C, ATAC-seq, and RNA-seq, we applied stratified LD score regression and calculated the proportion of genome-wide SNP heritability attributable to cell type-specific features, revealing pancreatic alpha cell enrichment as the most statistically significant. Subsequent chromatin contact-based fine-mapping was carried out for genome-wide significant childhood obesity loci and their linkage disequilibrium proxies to implicate effector genes, yielded the most abundant number of candidate variants and target genes at the BDNF, ADCY3, TMEM18, and FTO loci in skeletal muscle myotubes and the pancreatic beta-cell line, EndoC-BH1. One novel implicated effector gene, ALKAL2 - an inflammation-responsive gene in nerve nociceptors - was observed at the key TMEM18 locus across multiple immune cell types. Interestingly, this observation was also supported through colocalization analysis using expression quantitative trait loci (eQTL) derived from the Genotype-Tissue Expression (GTEx) dataset, supporting an inflammatory and neurologic component to the pathogenesis of childhood obesity. Our comprehensive appraisal of 3D genomic datasets generated in a myriad of different cell types provides genomic insights into pediatric obesity pathogenesis.

3D genomic features across >50 diverse cell types reveal insights into the genomic architecture of childhood obesity

The prevalence of childhood obesity is increasing worldwide, along with the associated common comorbidities of type 2 diabetes and cardiovascular disease in later life. Motivated by evidence for a strong genetic component, our prior genome-wide association study (GWAS) efforts for childhood obesity revealed 19 independent signals for the trait; however, the mechanism of action of these loci remains to be elucidated. To molecularly characterize these childhood obesity loci, we sought to determine the underlying causal variants and the corresponding effector genes within diverse cellular contexts. Integrating childhood obesity GWAS summary statistics with our existing 3D genomic datasets for 57 human cell types, consisting of high-resolution promoter-focused Capture-C/Hi-C, ATAC-seq, and RNA-seq, we applied stratified LD score regression and calculated the proportion of genome-wide SNP heritability attributable to cell type-specific features, revealing pancreatic alpha cell enrichment as the most statistically significant. Subsequent chromatin contact-based fine-mapping was carried out for genome-wide significant childhood obesity loci and their linkage disequilibrium proxies to implicate effector genes, yielded the most abundant number of candidate variants and target genes at the BDNF, ADCY3, TMEM18, and FTO loci in skeletal muscle myotubes and the pancreatic beta-cell line, EndoC-BH1. One novel implicated effector gene, ALKAL2 – an inflammation-responsive gene in nerve nociceptors – was observed at the key TMEM18 locus across multiple immune cell types. Interestingly, this observation was also supported through colocalization analysis using expression quantitative trait loci (eQTL) derived from the Genotype-Tissue Expression (GTEx) dataset, supporting an inflammatory and neurologic component to the pathogenesis of childhood obesity. Our comprehensive appraisal of 3D genomic datasets generated in a myriad of different cell types provides genomic insights into pediatric obesity pathogenesis.

High fidelity simulations of contaminant dispersion in an urban environment with comparison to magnetic resonance imaging measurements

The dispersion of a contaminant in an urban environment has the potential to impact a large population of people. In this work, a complex urban canopy flow based on the Oklahoma City downtown business district circa 2003 is studied using Magnetic Resonance Imaging (MRI) and high-fidelity Large Eddy Simulations (LES). MRI is a novel experimental technique that can provide high-resolution measurements in four dimensions (three spatial and temporal) for lab scale models. The experiments and simulations use the same geometry and boundary conditions providing a one-to-one comparison of the two methods. Results are presented on the time-averaged velocity and concentration fields, the temporal dynamics of the concentration plumes for a transient release, and a novel Cloud Identification Algorithm that can separate plumes produced by periodic contaminant releases used for ensemble averaging over many releases. The MRI and LES datasets both include millions of measurement voxels and the comparisons highlight the complex 3D nature of the flow including strong vertical velocities in spanwise street canyons and flow acceleration in streamwise street canyons. The concentration fields are qualitatively similar albeit the LES shows larger dispersion. A quantitative analysis with performance measures compares the datasets pointwise and demonstrates that the two 3D datasets are similar with respect to many measures including a fractional bias of 0.02 (ideal=0.0), correlation coefficient of 0.87 (ideal = 1.0), and the fraction points within a factor of 2 is 0.98 (ideal = 1.0). Plume analysis compares the arrival and residence time of contaminant and is found to vary significantly with location within the urban environment with arrival times between 0 and 1.25 and differences within the contaminant cloud less than 10% at most locations.

Improving 3D deep learning segmentation with biophysically motivated cell synthesis

Biomedical research increasingly relies on three-dimensional (3D) cell culture models and artificial-intelligence-based analysis can potentially facilitate a detailed and accurate feature extraction on a single-cell level. However, this requires for a precise segmentation of 3D cell datasets, which in turn demands high-quality ground truth for training. Manual annotation, the gold standard for ground truth data, is too time-consuming and thus not feasible for the generation of large 3D training datasets. To address this, we present a framework for generating 3D training data, which integrates biophysical modeling for realistic cell shape and alignment. Our approach allows the in silico generation of coherent membrane and nuclei signals, that enable the training of segmentation models utilizing both channels for improved performance. Furthermore, we present a generative adversarial network (GAN) training scheme that generates not only image data but also matching labels. Quantitative evaluation shows superior performance of biophysical motivated synthetic training data, even outperforming manual annotation and pretrained models. This underscores the potential of incorporating biophysical modeling for enhancing synthetic training data quality.

ModeHunter: A Package for the Reductionist Analysis, Animation, and Application of Elastic Biomolecular Motion.

ModeHunter is a modular Python software package for the simulation of 3D biophysical motion across spatial resolution scales using modal analysis of elastic networks. It has been curated from our in-house Python scripts over the last 15 years, with a focus on detecting similarities of elastic motion between atomic structures, coarse-grained graphs, and volumetric data obtained from biophysical or biomedical imaging origins, such as electron microscopy or tomography. With ModeHunter, normal modes of biophysical motion can be analyzed with various static visualization techniques or brought to life by dynamics animation in terms of single or multimode trajectories or decoy ensembles. Atomic structures can also be refined against volumetric densities with flexible fitting strategies. The software consists of multiple stand-alone programs for the preparation, analysis, visualization, animation, and refinement of normal modes and 3D data sets. At its core, two spatially reductionist elastic motion engines are currently supported: elastic network models (typically for a Cα level of detail and rectangular meshes) and bend-twist stretch (for trigonal or tetrahedral meshes or trees resulting from spatial clustering). The programs have recently been modernized to Python 3, requiring only the common numpy and scipy external libraries for numerical support. The main advantage of our modular design is that the tools can be combined by the end users for specific modeling applications, either standalone or with complementary tools from our C/C++-based Situs modeling package. The modular design and consistent look and feel facilitate the maintenance of individual programs and the development of novel application workflows. Here, we provide the first complete overview of the ModeHunter package as it exists today, with an emphasis on functionality and workflows supported by version 1.4.

Evidential transformer for buried object detection in ground penetrating radar signals and interval‐based bounding box

AbstractThree‐dimensional (3D) buried object detection using ground penetrating radar (GPR) benefits from the powerful capacity of image‐wise deep neural networks. However, it still faces the challenge of information loss from raw GPR signals to two‐ and three‐dimensional images, such as the frequency‐domain information loss when normalizing GPR signals into gray‐scale images and spatial information loss when using stacked B‐ and C‐scan images to replace raw GPR signals as inputs. To solve the challenge, this study has proposed an ENNreg‐transformer model, directly using raw 3D GPR signals to perform buried object detection. In the proposed model, 3D GPR signals are first converted into sequential voxelization to obtain spatiotemporal features. The features are then aggregated by an intuition‐guided feature aggregation layer to simulate the expert behavior to analyze 3D GPR data. Finally, an evidential detection header outputs 3D interval‐based bounding boxes for buried object detection. The experiment on two 3D GPR road datasets demonstrates that the proposed model exceeds other state‐of‐the‐art models on the tasks thanks to raw 3D signals and intuition‐guided feature aggregation. In addition, the interval‐based bounding box represents the spatial bounding‐box uncertainty, which derives from the inherent limitations of GPR and deep networks.

Architectural planning robot driven by unsupervised learning for space optimization.

Space optimization in architectural planning is a crucial task for maximizing functionality and improving user experience in built environments. Traditional approaches often rely on manual planning or supervised learning techniques, which can be limited by the availability of labeled data and may not adapt well to complex spatial requirements. To address these limitations, this paper presents a novel architectural planning robot driven by unsupervised learning for automatic space optimization. The proposed framework integrates spatial attention, clustering, and state refinement mechanisms to autonomously learn and optimize spatial configurations without the need for labeled training data. The spatial attention mechanism focuses the model on key areas within the architectural space, clustering identifies functional zones, and state refinement iteratively improves the spatial layout by adjusting based on learned patterns. Experiments conducted on multiple 3D datasets demonstrate the effectiveness of the proposed approach in achieving optimized space layouts with reduced computational requirements. The results show significant improvements in layout efficiency and processing time compared to traditional methods, indicating the potential for real-world applications in automated architectural planning and dynamic space management. This work contributes to the field by providing a scalable solution for architectural space optimization that adapts to diverse spatial requirements through unsupervised learning.