3D Module Research Articles

3D plants: the impact of integrating science, design, and technology on high school student learning and interests in STEAM subjects and careers

STEAM education is an educational approach of interdisciplinary teaching of science, technology, engineering, art, and mathematics. STEAM education, however, is often viewed as only including art elements into STEM teaching. Without true integration of the disciplines in STEAM curricula, students rarely are exposed to the connection among disciplines, and self-identify as solely scientists, artists, or technophiles. STEAM curricula also infrequently integrate design, which promotes creativity and innovation. Effective STEAM curriculum and practices are needed to prepare students to face 21st century challenges and work demands. We designed a high school STEAM educational module that integrated plant science, design, and emergent technologies through the creation of 3D models of plants and augmented and virtual reality (AVR) experiences and investigated its impact on students’ understanding of the intersection of art and design with science, learning and skills gains, and interests in STEAM subjects and careers. The module used a project-based learning approach that relied on student teamwork and facilitation by educators. In this 3D plant modeling module, students: (1) investigated plants under research at a plant science research center, (2) designed and created 3D models of those plants, (3) learned about the application of 3D modeling in AVR platforms, and (4) disseminated project results. We used qualitative and quantitative research methods both before and after the implementation of the model to assess the impact of the 3D modeling module. Student responses revealed that approximately half of the students had a good understanding of the intersection of art and design with science prior to the implementation of the module, while the other half gained this understanding after completing their projects. Students saw art and design playing a role in science mainly by facilitating communication and further understanding and fostering new ideas. They also reported that science influenced art and design through the artistic creation process. The most common learning gains were in plant science and 3D modeling, with 35% and 20% of the students reporting these themes only after completing their projects, respectively. The skill gains most cited were research, teamwork, and communication skills. Over 25% of the students reported these skill gains only after the completion of their projects. Paired comparisons of survey responses indicated a significant increase in students’ interest in science, mathematics, and design subjects after they completed their projects. At the end of the module, 40% of the students were more interested in STEAM careers. Another 13% of the students indicated they already had an interest in STEAM careers before beginning the module. Our findings indicate that our STEAM module effectively integrated science, art, design, and technology, enhancing student literacy in these fields, and providing students with essential 21st century skills. The module led to interdisciplinary learning and development of interest in STEAM subjects and careers. The combination of pedagogical strategies used in our module for active, collaborative, authentic, and meaningful learning exemplifies an effective STEAM curriculum with valuable instructional tools for educators, inspiring new ways of teaching and learning, contributing to the practice and applications in STEAM education.

Analysis of Students' Critical Thinking Skills Profile and Implementation of PBL on Sound Wave Material

In the 21st century, students are expected to be able to have critical thinking skills to be able to read and solve problems in their daily lives. For this reason, it is important to determine appropriate learning models and media. Problem-Based Learning (PBL) is a learning model that can improve critical thinking skills. This research was conducted to determine students' critical thinking skills and determine the application of learning models in physics subjects, especially in the PBL learning model. The sample consisted of 50 class XI students in one of the high schools in Gresik. The research results from the question instruments show the low level of students' critical thinking skills. The results of interviews with physics educators explained that physics learning activities so far have used several innovative learning models and media. Questionnaire responses showed that students had the perception that they had met the indicators of critical thinking skills, and also showed interest in implementing the 3D digital module. So it can be concluded that there is a need to implement physics learning using the help of interactive learning media such as 3D digital modules to increase students' activeness in learning and train critical thinking skills.

Enhancing students' critical thinking skills through mobile technology: An analysis of problem-based learning implementation in heat material instruction

This study discusses the analysis of students' critical thinking skills profile and the implementation of PBL on heat material. This study aimed to determine the profile of students' critical thinking skills and the implementation of PBL on heat material at SMAN 1 Gondang. This study used pre-research with qualitative descriptive analysis with a sample of 91 SMA Negeri 1 Gondang students. Data collection techniques were written tests, student response questionnaires, and teacher interviews. The study found that students critical thinking skills were in a low category, and inference was the indicator of essential thinking skills with the lowest average. Teachers still used lecture methods and conventional teaching materials during learning, so it was necessary to apply a model, namely a problem-based learning model assisted by a 3D module, which was expected to improve critical thinking skills on heat material. Learning innovations need to be carried out to improve students' critical thinking skills, especially in physics learning, one of which is the application of Problem-Based Learning assisted by a 3D module.

Effect of Corotating Interaction Regions of Solar Wind on GCR Intensity in 2D Modulation Problems

Effect of Corotating Interaction Regions of Solar Wind on GCR Intensity in 2D Modulation Problems

Mambav3d: A mamba-based virtual 3D module stringing semantic information between layers of medical image slices

High-precision medical image segmentation provides a reliable basis for clinical analysis and diagnosis. Researchers have developed various models to enhance the segmentation performance of medical images. Among these methods, two-dimensional models such as Unet exhibit a simple structure, low computational resource requirements, and strong local feature capture capabilities. However, their spatial information utilization is insufficient, limiting their segmentation accuracy. Three-dimensional models, such as 3D Unet, utilize spatial information more fully and are suitable for complex tasks, but they require high computational resources and have limited real-time performance. In this paper, we propose a virtual 3D module (Mambav3d) based on mamba, which introduces spatial information into 2D segmentation tasks to more fully integrate the 3D information of the image and further improve segmentation accuracy under conditions of low computational resource requirements. Mambav3d leverages the properties of hidden states in the state space model, combined with the shift of visual perspective, to incorporate semantic information between different anatomical planes in different slices of the same 3D sample. The voxel segmentation is converted to pixel segmentation to reduce model training data requirements and model complexity while ensuring that the model integrates 3D information and enhances segmentation accuracy. The model references the information from previous layers when labeling the current layer, thereby facilitating the transfer of semantic information between slice layers and avoiding the high computational cost associated with using structures such as Transformers between layers. We have implemented Mambav3d on Unet and evaluated its performance on the BraTs, Amos, and KiTs datasets, demonstrating superiority over other state-of-the-art methods.

Microfluidic Bubble-Templating Multinozzle 3D Printing for Preparations of Macroporous Hydrogels with Ordered Pores of Large Size Ranges.

A strategy for fabrication of macroporous hydrogels through 3D printing assisted by molding and multiple microfluidic bubble-templating nozzles is proposed here. This approach aims to address the challenges faced by methods for 3D printing macroporous hydrogels, such as difficulties in precisely controlling the spatial distribution of macropores, limited porosity, and low resolution of external boundaries due to the poor mechanical properties of hydrogel solutions as printing ink. In this method, fast-switching microfluidic bubble-templating nozzles of varying sizes allowed for precise control of target pore sizes over a wide range. Cavities of polydimethylsiloxane (PDMS) molds were used to define the overall external shape of the hydrogels and enhance the resolution of the boundaries. During the printing process, the laminar shear force generated by a coaxial microfluidic system generated microbubbles which were then injected into the PDMS mold cavities. The spatial distribution of the microbubble clusters within the cavities was controlled by a 3D motion module. Finally, macroporous hydrogels with a pore size range of 238-930 μm were fabricated while avoiding millimeter round corners on the boundaries. A control precision of pore sizes was about 60 μm. Most importantly, this strategy broke through the space occupancy limitations of pores in traditional 3D printing methods for macroporous hydrogels. The space occupancy of pores could reach a maximum of about 87% within a single layer.

Fast reconstruction 3D computed tomography image of stacked cell under faster scanning by dual-branch cross-fusion flat bottom network

Abstract Cone beam computed tomography (CBCT) fast scanning and reconstruction is a key step to achieve rapid detection of internal defects in batteries. In this work, we have achieved a faster CT scanning just in 5 seconds by reducing the X-ray exposure time in sparse view CT. However, the CT data is extremely incomplete by faster scanning; the existing reconstruction methods are difficult to reconstruct a high quality three-dimensional (3D) CT image of stacked cells. To address this issue, we propose a 3D CT image reconstruction network, which can reconstruct higher quality CT images from low quality 3D volume data. The input data of the reconstruction network is not 2D projection data, but 3D volume data. In this network, a high and low resolution dual-branch cross-fusion flat bottom structure is designed. The high resolution flat bottom branch aims to preserve detailed information, while the low resolution flat bottom branch focuses on capturing more semantic information. Cross-fusion between these branches mitigates the loss of semantic details. Additionally, the auxiliary loss function, the main loss function, and the 3D attention module are designed to enhance semantic accuracy and the learning performance of the network. The 3D training data is collected under a fast scanning strategy spanning 5-60 seconds. During the training phase, we use clipping block technology to cut the 3D volume data, enabling direct training on the 3D volume data. Our experimental results demonstrate that our 3D reconstruction network outperforms mainstream algorithms under this faster scanning strategy, which is able to reconstruct higher quality 3D CT images just in 15 seconds. Ablation experiments confirm the positive impact of the dual-branch cross-fusion flat bottom structure, attention module, and loss functions on improving the quality of 3D CT images.

Creation of self-aligning multisatellite planetary gears

In this study, we aim to identify shortcomings in the operation of multisatellite planetary gearboxes and to justify improved designs of planetary self-aligning mechanisms, which allow torque to be transmitted through all installed satellites. To that end, we studied a domestic planetary three-satellite MPZ gearbox installed on a magnetic separator, designed for wet separation of strongly magnetic ores and materials into magnetic and non-magnetic products, and a planetary gearbox of a special structure. 3D models of the classical three-satellite and single-sliding self-aligning gearboxes were constructed in the 3D module of the T-Flex CAD software. The performance of the mechanisms was evaluated by structural analysis and Chebyshev’s mobility formula. The constructed 3D models were used to study the engagement process of the classical three-satellite and single-sliding self-aligning gearbox. The latter is distinguished by the presence of two levers, which are introduced additionally to three satellites. During the research, the lateral clearance between the pairs of satellite teeth and center wheels was accepted according to GOST 1643–81. The conducted analysis of contacts at rotation of the central driving wheel by 30°, 110°, 200°, and 310° established further designing of the applied planetary gears to be inexpedient due to impossibility of realization of power transfer simultaneously through all satellites. The main disadvantage of the operation of multisatellite gearboxes was found to be the requirement for the side clearance selection, which should ensure the operability of the transmission at the moment of motion transfer. When designing multi-satellite single-slide selfaligning planetary mechanisms, additional levers should be introduced into the structure, the number of which should be equal to the number of satellites. Hence, the use of self-aligning planetary gears makes it possible to reduce the dimensions due to the uniform distribution of the transmitted torque, since the calculated value of the torque can be reduced by the number of satellites. This allows the gears to be adapted to the working loading conditions, which significantly improves the performance of the entire machine or unit.

TCANet: Three-dimensional cross-attention mechanism for stereo-matching

Abstract Effective disparity estimation is a current hotspot in stereo vision research, and cost aggregation is an important part of disparity prediction. How to perform more effective cost aggregation is the core step to improve the accuracy of disparity prediction. In previous studies, 3DCNN with a stacked hourglass structure is often used for cost aggregation. In this research, we propose an effective 3D cross-attention stereo network that utilizes the attention mechanism to obtain contextual information for cost aggregation in a more efficient way. Specifically, the 3D cross-attention module in TCANet acquires the geometric information of all pixels on the 3D cross path. By repeating this operation twice, each pixel can eventually obtain global dependencies from all other pixels. Using the 3D cross-attention module in a stacked hourglass-structured 3DCNN only increases the number of parameters by a very small amount, which can effectively improve the performance of the model. Experimental results show that TCANet performs well on virtual dataset Scene Flow and realistic KITTI datasets.

Design of a Simulation and Evaluation System for Athlete Technical and Tactical Training Based on Virtual Reality Technology

In order to avoid the influence of external environmental factors on athletes’ physical fitness and technical tactics, the author proposes the design and research of an integrated simulation training system for athletes’ physical fitness and technical tactics based on virtual reality. The hardware unit of the design system includes a 3D scanner module, VR glasses module, locator module, tracker module, and wireless transceiver device. The software module introduces virtual reality technology and designs a visual simulation module, trained object detection module, trained object tracking module, and overall control module. The simulation experiment results show that compared with existing systems, the training scene output clarity and frame rate of the designed system are better, and the thread switching time and signal mixing time are shorter, which fully proves that the designed system has higher operating efficiency and better application performance.

Enhancing Precision in OLT Measurements and Improving Agreement in Surgical Decision-Making: A Comprehensive Evaluation Using WBCT and Distance Mapping for Preoperative Planning

Category: Ankle; Other Introduction/Purpose: WBCT enables the creation of a three-dimensional model that represents the ankle morphology in a standing position. Distance mapping (DM) is a complementary feature that uses color coding to represent the relative intraarticular distance and can be used to outline intraarticular defects. Consequently, DM offers a novel approach to delineating OLT, allowing for the quantification of its surface area, volume, and depth. The reliability of DM for OLT measurements has yet to be thoroughly evaluated. This study primarily aims to determine the reliability of DM in measuring the surface, depth, and volume of OLT. A secondary objective is to ascertain whether measurements obtained through DM, when integrated with a predefined treatment algorithm, can facilitate consensus among surgeons regarding the optimal surgical intervention Methods: This cohort comprised 36 patients with 40 OLTs evaluated using WBCT and DM. Two raters used DM to determine the lesion boundary (LB) and lesion fundus (LF) and calculate the lesion depth, surface, and volume. The raters were asked to choose between bone marrow stimulation, autologous matrix-induced chondrogenesis, and osteochondral transposition based on the measurement. Inter-rater and intra-rater agreement was measured. Results: Interclass correlation of the lesion's depth surface produced an excellent interrater and intrareader agreement of 0.90 -0.94 p-value < 0.001. Cohen's Kappa agreement analysis of the preferable preoperative plan produced a Kappa = 0.834 P-value< 0.001, indicating a near-perfect agreement Conclusion: WBCT-based 3D modules and DM can be used to measure the OLT surface and depth and to calculate the lesion volume with excellent interrater and interrater reliability; using the OLT measurement and a predetermined treatment algorithm, a near-perfect interrater agreement for the preoperative planning was reached. WBCT in conjunction could help determine the type of surgery needed preoperatively and assess if additional surgeries are needed.

Deep learning-based segmentation for high-dose-rate brachytherapy in cervical cancer using 3D Prompt-ResUNet

Objective.To develop and evaluate a 3D Prompt-ResUNet module that utilized the prompt-based model combined with 3D nnUNet for rapid and consistent autosegmentation of high-risk clinical target volume (HRCTV) and organ at risk (OAR) in high-dose-rate brachytherapy for cervical cancer patients.Approach.We used 73 computed tomography scans and 62 magnetic resonance imaging scans from 135 (103 for training, 16 for validation, and 16 for testing) cervical cancer patients across two hospitals for HRCTV and OAR segmentation. A novel comparison of the deep learning neural networks 3D Prompt-ResUNet, nnUNet, and segment anything model-Med3D was applied for the segmentation. Evaluation was conducted in two parts: geometric and clinical assessments. Quantitative metrics included the Dice similarity coefficient (DSC), 95th percentile Hausdorff distance (HD95%), Jaccard index (JI), and Matthews correlation coefficient (MCC). Clinical evaluation involved interobserver comparison, 4-grade expert scoring, and a double-blinded Turing test.Main results.The Prompt-ResUNet model performed most similarly to experienced radiation oncologists, outperforming less experienced ones. During testing, the DSC, HD95% (mm), JI, and MCC value (mean ± SD) for HRCTV were 0.92 ± 0.03, 2.91 ± 0.69, 0.85 ± 0.04, and 0.92 ± 0.02, respectively. For the bladder, these values were 0.93 ± 0.05, 3.07 ± 1.05, 0.87 ± 0.08, and 0.93 ± 0.05, respectively. For the rectum, they were 0.87 ± 0.03, 3.54 ± 1.46, 0.78 ± 0.05, and 0.87 ± 0.03, respectively. For the sigmoid, they were 0.76 ± 0.11, 7.54 ± 5.54, 0.63 ± 0.14, and 0.78 ± 0.09, respectively. The Prompt-ResUNet achieved a clinical viability score of at least 2 in all evaluation cases (100%) for both HRCTV and bladder and exceeded the 30% positive rate benchmark for all evaluated structures in the Turing test.Significance.The Prompt-ResUNet architecture demonstrated high consistency with ground truth in autosegmentation of HRCTV and OARs, reducing interobserver variability and shortening treatment times.

Brain tumor segmentation by combining MultiEncoder UNet with wavelet fusion.

Accurate segmentation of brain tumors from multimodal magnetic resonance imaging (MRI) holds significant importance in clinical diagnosis and surgical intervention, while current deep learning methods cope with situations of multimodal MRI by an early fusion strategy that implicitly assumes that the modal relationships are linear, which tends to ignore the complementary information between modalities, negatively impacting the model's performance. Meanwhile, long-range relationships between voxels cannot be captured due to the localized character of the convolution procedure. Aiming at this problem, we propose a multimodal segmentation network based on a late fusion strategy that employs multiple encoders and a decoder for the segmentation of brain tumors. Each encoder is specialized for processing distinct modalities. Notably, our framework includes a feature fusion module based on a 3D discrete wavelet transform aimed at extracting complementary features among the encoders. Additionally, a 3D global context-aware module was introduced to capture the long-range dependencies of tumor voxels at a high level of features. The decoder combines fused and global features to enhance the network's segmentation performance. Our proposed model is experimented on the publicly available BraTS2018 and BraTS2021 datasets. The experimental results show competitiveness with state-of-the-art methods. The results demonstrate that our approach applies a novel concept for multimodal fusion within deep neural networks and delivers more accurate and promising brain tumor segmentation, with the potential to assist physicians in diagnosis.

DFEDC: Dual fusion with enhanced deformable convolution for medical image segmentation

Considering the complexity of lesion regions in medical images, current researches relying on CNNs typically employ large-kernel convolutions to expand the receptive field and enhance segmentation quality. However, these convolution methods are hindered by substantial computational requirements and limited capacity to extract contextual and multi-scale information, making it challenging to efficiently segment complex regions. To address this issue, we propose a dual fusion with enhanced deformable convolution network, namely DFEDC, which dynamically adjusts the receptive field and simultaneously integrates multi-scale feature information to effectively segment complex lesion areas and process boundaries. Firstly, we combine global channel and spatial fusion in a serial way, which integrates and reuses global channel attention and fully connected layers to achieve lightweight extraction of channel and spatial information. Additionally, we design a structured deformable convolution (SDC) that structures deformable convolution with inceptions and large kernel attention, and enhances the learning of offsets through parallel fusion to efficiently extract multi-scale feature information. To compensate for the loss of spatial information of SDC, we introduce a hybrid 2D and 3D feature extraction module to transform feature extraction from a single dimension to a fusion of 2D and 3D. Extensive experimental results on the Synapse, ACDC, and ISIC-2018 datasets demonstrate that our proposed DFEDC achieves superior results.

Three-Dimensional Convolutional Vehicle Black Smoke Detection Model with Fused Temporal Features

The growing concern over pollution from vehicle exhausts has underscored the need for effective detection of black smoke emissions from motor vehicles. We believe that the optimal approach for the detection of black smoke is to leverage existing roadway CCTV cameras. To facilitate this, we have collected and publicly released a black smoke detection dataset sourced from roadway CCTV cameras in China. After analyzing the existing detection methods on this dataset, we found that they have subpar performance. As a result, we decided to develop a novel detection model that focuses on temporal information. This model utilizes the continuous nature of CCTV video feeds rather than treating footage as isolated images. Specifically, our model incorporates a 3D convolution module to capture short-term dynamic and semantic features in consecutive black smoke video frames. Additionally, a cross-scale feature fusion module is employed to integrate features across different scales, and a self-attention mechanism is used to enhance the detection of black smoke while minimizing the impact of noise, such as occlusions and shadows. The validation of our dataset demonstrated that our model achieves a detection accuracy of 89.42%,showing around 3% improvement over existing methods. This offers a novel and effective solution for black smoke detection in real-world applications.

ARTEMIS: a method for topology-independent superposition of RNA 3D structures and structure-based sequence alignment.

Non-coding RNAs play a major role in diverse processes in living cells with their sequence and spatial structure serving as the principal determinants of their function. Superposition of RNA 3D structures is the most accurate method for comparative analysis of RNA molecules and for inferring structure-based sequence alignments. Topology-independent superposition is particularly relevant, as evidenced by structurally similar RNAs with sequence permutations such as tRNA and Y RNA. To date, state-of-the-art methods for RNA 3D structure superposition rely on intricate heuristics, and the potential for topology-independent superposition has not been exhausted. Recently, we introduced the ARTEM method for unrestrained pairwise superposition of RNA 3D modules and now we developed it further to solve the global RNA 3D structure alignment problem. Our new tool ARTEMIS significantly outperforms state-of-the-art tools in both sequentially-ordered and topology-independent RNA 3D structure superposition. Using ARTEMIS we discovered a helical packing motif to be preserved within different backbone topology contexts across various non-coding RNAs, including multiple ribozymes and riboswitches. We anticipate that ARTEMIS will be essential for elucidating the landscape of RNA 3D folds and motifs featuring sequence permutations that thus far remained unexplored due to limitations in previous computational approaches.

Deep learning based fine-grained recognition technology for basketball movements

The technical essentials of basketball are very complex, in order to accomplish the identification of fine-grained basketball movements and improve the quality of basketball training, the study innovatively proposes a region screening strategy to obtain the local region information of image frames, and draws on the SlowFast network structure to design the static and dynamic detail models, and finally utilizes the 3D attention feature fusion module to complete the multi-feature relationship processing and recognition. Performance testing experiments have confirmed that the dual stream model based on region filtering has achieved a maximum improvement in average accuracy of 16.35 % and 60.36 % in the recognition of RGB images, preserving relatively complete local effective region information of RGB images. The model based on channel domain attention mechanism improves its average accuracy to over 60 %. The model recognition accuracy calculation time after dual stream fusion is reduced by about 50 %. The model based on 3D attention feature fusion module exhibits stronger robustness when the feature dimension changes. And the loss value of this model rapidly decreases and converges to the minimum loss value, with its accuracy and recall reaching 90.03 % and 88.47 % respectively, and its AUC value reaching a maximum of 0.761. The design of the study contributes to the digital development of teaching and training in basketball and enriches the theoretical body of knowledge on deep learning and action recognition.

Thermal propagation analysis of 2G HTS stacked wires based on a dimensional coupling method

Stacked structures of high-temperature superconducting (HTS) tapes are usually employed to enhance current-carrying capacity in cable or magnet applications, but their multilayer characteristics may result in local hot spots due to uneven cooling. Besides, the inconsistencies along the length of tapes introduced during manufacturing process and the significant transverse Lorentz forces experienced in magnet operations can lead to weak parts, bringing uncertainties to thermal stability. Finite-element methods are widely used to predict thermal propagations, but 3D models encounter challenges such as distorted mesh elements and convergence issues, particularly when combining electromagnetic and heat transfer modules for long-distance wires. In this work, we have developed a Dimensional Coupling Method (DCM) to assess the thermal impact of weak parts in stacked wires applying coupled 1D and 2D models. The 2D model analyzes the electromagnetic and heat characteristics of stacked surfaces, and provides an initial heat source for the 1D model, which evaluates thermal propagation longitudinally. Simulation results of the 1D module are then transferred back to update the 2D outcomes. Models of distinct dimensions are coupled sequentially in physical steps but simultaneously in the time domain. Our approach is verified by 3D model benchmarks and offers a computational cost reduction of approximately 60 % compared to the benchmarks, making it more suitable for applications with large Iop/Ic ratios. Specially, multi-layer stacked wires under low ratios cases are also been analyzed. What’s more, two influencing factors of heat propagation, the weak-part length and position, are also investigated.

Non-uniform critical current and stacking effect remedy for multi-filament REBCO tapes with potential defects

The high aspect ratio of REBCO tapes has a significant impact on several characteristics in high-temperature superconducting applications, such as critical current and AC loss. Narrow filamentary technology can effectively reduce the impact of magnetic field dependence and enhance the electromagnetic performance of REBCO tapes. However, the existing methods are constrained by the trade-off between the narrow degree of REBCO filaments and high current capacity. Meanwhile, when processing REBCO tapes for large-scale magnets, there is a great possibility that local defects are lurking. A striated narrow-stacked (NS) structure is proposed based on the existing narrowing methods to address these challenges. To verify the validity of this structure, it is imperative to explore the non-uniform critical current and stacking effect on performance for multi-filament REBCO tapes with potential defects. This article introduces a magnetic extended network (MEN) model to analyze the electrical characteristics of striated NS structures with different types of potential defects. Then, by coupling with a 3D finite element method electromagnetic module, the calculation results of the MEN model are visualized and used to analyze the electromagnetic characteristics including current sharing mechanism, magnetic field distribution, and critical current compensation due to stacking effect. It is found that stack structures successfully provide the performance remedy for multi-filament REBCO tapes with potential defects. This study aims to promote the narrowing improvement of REBCO tapes in high-field magnets and high-current applications.

EBC-Net: 3D semi-supervised segmentation of pancreas based on edge-biased consistency regularization in dual perturbation space.

Deep learning technology has made remarkable progress in pancreatic image segmentation tasks. However, annotating 3D medical images is time-consuming and requires expertise, and existing semi-supervised segmentation methods perform poorly in the segmentation task of organs with blurred edges in enhanced CT such as the pancreas. To address the challenges of limited labeled data and indistinct boundaries of regions of interest (ROI). We propose Edge-Biased Consistency Regularization (EBC-Net). 3D edge detection is employed to construct edge perturbations and integrate edge prior information into limited data, aiding the network in learning from unlabeled data. Additionally, due to the one-sidedness of a single perturbation space, we expand the dual-level perturbation space of both images and features to more efficiently focus the model's attention on the edges of the ROI. Finally, inspired by the clinical habits of doctors, we propose a 3D Anatomical Invariance Extraction Module and Anatomical Attention to capture anatomy-invariant features. Extensive experiments have demonstrated that our method outperforms state-of-the-art methods in semi-supervised pancreas image segmentation. Moreover, it can better preserve the morphology of pancreatic organs and excel at edges region accuracy. Incorporated with edge prior knowledge, our method mixes disturbances in dual-perturbation space, which shifts the network's attention to the fuzzy edge region using a few labeled samples. These ideas have been verified on the pancreas segmentation dataset.