Boundary Character Research Articles

Cross co-teaching for semi-supervised medical image segmentation

Excellent performance has been achieved on semi-supervised medical image segmentation, but existing algorithms perform relatively poorly for objects with variable topologies and weak boundaries. In this paper, we propose a novel cross co-teaching framework, called Cross-structure-task Collaborative Teaching (CroCT), which not only can effectively handle variable topologies, but also strengthens the learning for weak boundaries of unlabeled data. Specifically, a new cross-structure-task collaborative teaching mechanism is developed based on our designed “E-Net” structure composed of a shared encoder and two decoder branches with distinct learning paradigms, which asks these two branches to regress topology-aware signed distance functions and densely-predicted segmentation masks for each other. Powered by the collaboration across different structural biases and sequence-related tasks, our CroCT can extract more discriminative yet complementary representations from abundant raw medical data to promote the consistency learning generalization, further boosting the performance for tackling highly diverse shapes and topological changes intra-/inter-slices. Besides, it guarantees the diversities from multi-levels, i.e., structure and task perspectives, to preclude prediction uncertainty. In addition, a novel adaptive boundary enhancing (ABE) module is proposed to introduce compact annularly enhanced boundary features into semi-supervised training, which significantly improves weak boundary perception ability for unlabeled data while facilitating collaborative teaching for efficiently propagating complementary knowledge across different branches. The extensive experiments on three challenging medical benchmarks, employing different labeled settings, demonstrate the superiority of our CroCT over recent state-of-the-art competitors.

Multi-guided-based image matting via boundary detection

Existing automatic matting methods tend to directly obtain the alpha mattes from the RGB image using semantic segmentation networks, and relying solely on segmentation to achieve high-quality estimation is usually unrealistic. To address this issue, we propose a multi-guided-based image matting (MGBMatting) model that utilizes boundary information and semantic features as comprehensive and sufficient guidance, increasing attention to the unknown regions in the trimap, which is often a challenging aspect of matting tasks. The boundary-extracted module we introduced in MGBMatting effectively enhances the boundary features of the foreground. In addition, the boundary optimization module effectively achieves spatial consistency in features as well as enhances the representational power of the features. Further, the dual-stream encoder increases the possibility of capturing both local features and long-range feature dependencies. We use two widely used image matting datasets, Composition-1k and Distinctions-646, to evaluate our MGBMatting. Extensive experimentation demonstrates that the proposed MGBMatting yields significant performance.

DBL-Net: A dual-branch learning network with information from spatial and frequency domains for tumor segmentation and classification in breast ultrasound image

Breast cancer is a common malignancy, ranking as one of the most prevalent cancers among women globally. Improving treatment success rates for breast cancer are crucially dependent on early detection and accurate diagnosis. Ultrasound imaging, as a non-invasive and low-radiation imaging technique, plays a significant role in early breast cancer detection. However, due to the characteristics of fuzzy tumor boundaries and irregular shapes of the tumor, accurate breast tumor classification and segmentation remain challenging tasks. This paper introduces a novel approach for breast ultrasound image segmentation and classification tasks. Firstly, we study the potential value of frequency domain information in breast image analysis and propose the DBL-Net, a multi-task learning network for simultaneous tumor segmentation and tumor classification. Secondly, we customize a Spatial-Frequency Domain Feature Encoding module (S-FEM) to better encode and fuse spatial and frequency domain features of breast ultrasound images. Furthermore, we design a Representational Perception Enhancer (RPE) to improve the feature representation capabilities and prediction accuracy of the model. Finally, We have carried out a comprehensive series of tests on the BUS dataset, and the findings clearly indicate that our method outperforms the current state-of-the-art methods.

Advancing Temporal Action Localization with a Boundary Awareness Network

Temporal action localization (TAL) is crucial in video analysis, yet presents notable challenges. This process focuses on the precise identification and categorization of action instances within lengthy, raw videos. A key difficulty in TAL lies in determining the exact start and end points of actions, owing to the often unclear boundaries of these actions in real-world footage. Existing methods tend to take insufficient account of changes in action boundary features. To tackle these issues, we propose a boundary awareness network (BAN) for TAL. Specifically, the BAN mainly consists of a feature encoding network, coarse pyramidal detection to obtain preliminary proposals and action categories, and fine-grained detection with a Gaussian boundary module (GBM) to get more valuable boundary information. The GBM contains a novel Gaussian boundary pooling, which serves to aggregate the relevant features of the action boundaries and to capture discriminative boundary and actionness features. Furthermore, we introduce a novel approach named Boundary Differentiated Learning (BDL) to ensure our model’s capability in accurately identifying action boundaries across diverse proposals. Comprehensive experiments on both the THUMOS14 and ActivityNet v1.3 datasets, where our BAN model achieved an increase in mean Average Precision (mAP) by 1.6% and 0.2%, respectively, over existing state-of-the-art methods, illustrate that our approach not only improves upon the current state of the art but also achieves outstanding performance.

The Application of Geographic Information System in Analysing the Voting Patterns among Dayak Voters in the Sarawak State Elections of 2016 and 2021

Geographic information system (GIS) and election-related research focuses on the geographical characteristics of election boundaries, the election information management system and the application of GIS analysis to election boundaries. This article aims to acquire a deeper understanding of the issues affecting Dayak voters during the state elections of 2016 and 2021, geographically. This study employs three spatial analyses: Thematic Map-Density, Spatial Autocorrelation Moran’s I Index and Hot Spot Analysis (Getis-Ord Gi*). In select constituencies with a Dayak-majority, interviews and observations supplemented the spatial data. This article examines Dayak-majority constituencies’ geographical distribution and spatial concentration in the 2016 and 2021 elections. It demonstrates that there are significant differences in the density of voter participation, majorities obtained by candidates, and total votes cast by the electorate across the state in both state elections, confirming the assertion made by researchers of electoral politics in Sarawak that election issues are primarily location-specific and that understanding the problems faced by communities at the local level is essential. It is anticipated that the outcome of this study will aid in providing a spatial overview and projection of state election results and in strategising the future management of state elections.

In-situ analysis of SCC initiation on Alloy 600 surface in primary water environment using digital image correlation with electron backscatter diffraction

This study examined the stress corrosion cracking (SCC) phenomenon of Alloy 600 in a high-temperature primary water environment. The investigation involved the observation of local plastic deformation and SCC initiation on the specimen surface using an observation apparatus, comprising a laser confocal microscope (LCM) and an autoclave equipped with a diamond window view-port. Prior to conducting the SCC initiation test, the crystalline structure of the specimen, including grain boundary (GB) characteristics, Schmid factor, and trace inclination angle, underwent examination using an electron backscatter diffraction (EBSD) technique. The local strain distribution was subsequently analyzed using a digital image correlation (DIC) method, employing LCM images obtained during the SCC experiments. The ex-situ EBSD analysis indicated that SCC mainly initiated at random high-angle grain boundaries (HAGB) with trace inclination angles close to 90° relative to the tensile axis. DIC analysis further revealed that most GBs in the vicinity of the highly strained area cracked among random HAGBs with a high trace inclination angle to the tensile axis. This confirms that the local strain gradient at GBs plays an important role in SCC initiation, serving as a more suitable indicator of an additional stress component to effective normal stress acting on GBs compared to the Schmid factor mismatch.

Novel role of thermomechanical grain boundary engineering in the microstructure evolution of austenitic stainless steel

The effect of special grain boundary development via the grain boundary engineering process (GBE) was explored in a bimodal microstructured 316L austenitic stainless steel with an average grain size of 1.92 μm. The GBE process was carried out by the rolling-annealing method via two routes of low and medium applied strain, followed by a short annealing period in a single-step and iterative manner. The grain boundary character distribution was obtained by Electron Backscatter Diffraction (EBSD) analysis, and the intergranular corrosion resistance and the tensile behavior of the samples were examined. Microstructural characterization showed that applying low strain repetitively increased the coincidence site lattice (CSL) and Σ3 boundaries percentage and created large twin-related domains (TRD). The reduced sensitization was attributed to a high proportion of CSLs and large-sized TRDs by applying a low-strain route. In the early stages of the low-strain route, the grain boundary character distribution showed a high number of primary twin nuclei in coarse grains and a continuous network of single twin nuclei in smaller grains, creating a high percentage of Σ3-grain boundaries. The low applied strain and the absence of the necessary driving force for the recrystallization and GBs migration led to the formation of twin nuclei, which would assist with the reduction of microstructure energy. A high percentage of Σ3 boundaries and an increase in the percentage of triple points consisting of low energy boundaries were found to be influential factors in increasing elongation.

A Novel Boundary-Guided Global Feature Fusion Module for Instance Segmentation

The task of instance segmentation is widely acknowledged as being one of the most formidable challenges in the field of computer vision. Current methods have low utilization of boundary information, especially in dense scenes with occlusion and complex shapes of object instances, the boundary information may become ineffective. This results in coarse object boundary masks that fail to cover the entire object. To address this challenge, we are introducing a novel method called boundary-guided global feature fusion (BGF) which is based on the Mask R-CNN network. We designed a boundary branch that includes a Boundary Feature Extractor (BFE) module to extract object boundary features at different stages. Additionally, we constructed a binary image dataset containing instance boundaries for training the boundary branch. We also trained the boundary branch separately using a dedicated dataset before training the entire network. We then input the Mask R-CNN features and boundary features into a feature fusion module where the boundary features provide shape information needed for detection and segmentation. Finally, we use a global attention module (GAM) to further fuse features. Through extensive experiments, we demonstrate that our approach outperforms state-of-the-art instance segmentation algorithms, producing finer and more complete instance masks while also improving model capability.

Effect of Mn addition in W-Ni-Cu heavy alloy processed through hot press sintering techniques on microstructure, microtexture and grain boundary character evolution

In this current study, the impact of manganese addition to 93 W-4.9Ni-2.1Cu-xMn (where x = 1, 2, 3, and 4 Wt%) on the microstructure, texture development, and grain boundary character distribution of tungsten heavy alloy (WHAs) was examined. The samples were produced through a vacuum hot press sintering, with parameters 50 MPa pressure, 10 °C/min heating rate, and 1450 °C sintering temperature for 20 min holding duration. To study the influence of manganese on the texture development and microstructure in conventional WHAs, electron backscatter diffraction (EBSD) was employed. The important findings from the EBSD analysis reveal that fibre texture (001) 〈111〉 and a higher fraction of low-angle grain boundaries in the Mn unalloyed compact resulted from extensive atomic jumps between the particles at elevated temperatures. Further texture analysis on the Mn-added compacts reveals that manganese oxide (MnO) formation during liquid phase sintering led to lower densification and randomized weak fibre texture development in manganese-alloyed compacts. In addition, the texture degradation in Mn-added compacts is mainly caused by recrystallization followed by sub-structural recovery during liquid phase sintering. This study indicates that increasing manganese content in WHAs stoichiometry led to a heterogeneous distribution of binder matrix (NiCu) in the sintered compact. As a result, W-4.9Ni-2.1Cu-4Mn alloy revealed decreased relative density (90.70%), microhardness (389.5 HV0.5), and electrical conductivity (15.74% of IACS).

Prehistoric and Early Medieval Settlement Features at Ravelrig Road and Newmills Road, Balerno, City of Edinburgh

Evaluation and excavation works in advance of housing development at two sites on Ravelrig Road and Newmills Road, Balerno, City of Edinburgh, revealed significant archaeological features evidencing multi-phase settlement. At Ravelrig Road, two groups of pits were dated to the Early and Middle to Late Neolithic periods with a cluster of Copper Age and Early Bronze Age features that included two possible storage features. At Newmills Road, alongside limited evidence for late prehistoric occupation in the form of pits dating to the last centuries BC and the first centuries AD, the most significant remains were of early medieval date, including a ring-groove representing a putative roundhouse and other features, including a hearth and linear boundary features, generally placed between the 7th and 9th centuries AD according to radiocarbon dating.

Evolution of Grain Boundary Character Distribution in B10 Alloy from Friction Stir Processing to Annealing Treatment.

In this study, the grain boundary character distribution (GBCD) of a B10 alloy was optimized, employing thermomechanical processing consisting of friction stirring processing (FSP) and annealing treatment. Using electron backscatter diffraction, the effects of rotational speed of FSP and annealing time on the evolution of GBCD were systematically investigated. The GBCD evolution was analyzed concerning various parameters, such as the fraction of low-Σ coincidence site lattice (CSL) boundaries, the average number of grains per twin-related domain (TRD), the length of longest chain (LLC), and the triple junction distribution. The experimental results revealed that the processing of a 1400 rpm rotational speed of FSP followed by annealing at 750 °C for 60 min resulted in the optimum grain boundary engineering (GBE) microstructure with the highest fraction of low-Σ CSL boundaries being 82.50% and a significantly fragmented random boundary network, as corroborated by the highest average number of grains per TRD (14.73) with the maximum LLC (2.14) as well as the highest J2/(1 - J3) value (12.76%). As the rotational speed of FSP increased from 600 rpm to 1400 rpm, the fraction of low-Σ CSL boundaries monotonously increased. The fraction of low-Σ CSL boundaries first increased and then decreased with an increase in annealing time. The key to achieving GBE lies in inhibiting the recrystallization phenomenon while stimulating abundant multiple twinning events through strain-induced boundary migration.

Effect of Solution Annealing Time on the Microstructure and Mechanical Properties of Selective-Laser-Melted 2205 Duplex Stainless Steel

The 2205 duplex stainless steel (DSS) produced by selective laser melting (SLM) exhibits high strength (1078.8 MPa) but poor plasticity (15.2%) owing to the high cooling rate during SLM, which inhibits the formation of austenite and creates a nearly entirely ferritic microstructure. The dual-phase nature can be restored through solution annealing, which enables well-matched strength and plasticity, but which has not been extensively studied. We investigate the effects of 5 min, 30 min, and 120 min of solution annealing at 1000 °C on the dual-phase ratio, grain size, texture strength, inclusions, grain boundary characteristics, and mechanical properties of SLM-manufactured 2205 DSS. After 30 min of solution annealing, the elongation increased to 32.2% owing to the restoration of the dual-phase structure, the reduction in dislocation density, the weakening of texture, and the decrease in grain size. Increasing solution annealing time also corresponded to a decrease in the ultimate tensile strength (from 831.7 to 787.5 MPa) and yield strength (from 610.3 to 507.8 MPa) due to grain coarsening and the gradual transformation of ferrite to austenite. Furthermore, the mechanism of the transformation from ferrite to austenite was proposed, and it was observed that the transformation of MnSiO3 to MnCrO4 provided nucleation sites for austenite.

Effects of riser height and pipeline length on the transition boundary and frequency of severe slugging in offshore pipelines

Investigating the transition boundary and frequency characteristics of severe slugging under different riser heights and pipeline lengths provides important support for the expansion of offshore oil and gas to deeper waters. The visualization research on air-water two-phase flow is carried out in a system with a horizontal pipeline length of 1687 m and riser heights of 16 m, 22 m, and 29 m, and flow pattern maps are plotted. Based on existing experimental data, the database of the flow pattern and slugging frequency in pipeline-riser systems with riser heights of 3–29 m, pipeline lengths of 10–1687 m, and pipe diameters of 25.4–50.8 mm is established. The long pipeline makes the gas pressure growth rate slower than that of the liquid pressure in the riser, which causes severe slugging occurring at higher superficial gas velocity. On the contrary, increasing the riser height causes the boundary of severe slugging to shift to a lower superficial gas velocity. The severe slugging length is greater in the higher riser and longer pipeline system, causing a smaller severe slugging frequency. A correlation for predicting severe slugging frequency at various riser heights, pipe lengths, and pipe diameters is established by employing gas and liquid Froude numbers.

Dugs-UNet: A Novel Deep Semantic Segmentation Approach to Convection Detection Based on FY-4A Geostationary Meteorological Satellite

Severe convection is a disastrous mesoscale weather system. The early detection of such systems is very important for saving peoples’ lives and properties. Previous studies address the issue mainly based on thresholding methods, which are not robust and accurate enough. In this paper, we propose a novel semantic segmentation method (Dugs-UNet) to solve the problem. Our method is based on the well-known U-Net framework. As convective clouds mimic fluids, its detection faces two important challenges. First, the shape and boundary features of clouds need to be carefully exploited. Second, the positive and negative samples for convection detection are very imbalanced. To address the two challenges, our method was carefully developed. Regarding the importance of the shape and boundary features for convective target detection, we introduce a shape stream module to extract these features. Also, a data-dependent upsample operation is adopted in the decoder of U-Net to effectively utilize the features. This is one of our contributions. To address the imbalance issue for convective target detection, the a focal loss function is employed to train our method, which is another contribution. Experimental results of 2018 Fengyun-4A satellite observations in China demonstrate the effectiveness of the proposed method. Compared to conventional thresholding-based methods and deep semantic segmentation algorithms such as SegNet, PSPNet, DeepLav-v3+ and U-Net, the proposed approach performs the best.

New insights into the character of austenite-ferrite boundaries in an additively manufactured duplex stainless steel

The crystallography and chemistry of interfaces between austenite and ferrite in duplex steels control many important materials properties but remain poorly understood. In this study, we experimentally show that in an additively manufactured and heat treated duplex stainless steel, the majority of austenite-ferrite interfaces terminate on {111}A/{110}F planes, and this behaviour is more pronounced for rational interfaces with the Kurdjomov-Sachs orientation relationship. Interface segregation was found to be controlled by not only the interface crystallography but also the bonding properties of solute atoms. Solute elements showed higher interfacial excess at irrational interfaces. Furthermore, a heterogeneous distribution of selected solute elements in austenite-ferrite interfaces planes was observed. Our findings reinforce the importance and, in fact, necessity to consider five independent crystallographic parameters and chemical architecture of interphase boundaries for advanced control of mechanical and other critical properties in duplex materials.

Effect of nitrogen content on grain boundary engineering and corrosion resistance of 316LN stainless steel

Nitrogen solid solution can effectively enhance the strength of steels, but its effect on grain boundary engineering and the corrosion mechanism of steels is still unclear and systematic investigation is needed. In this study, 316LN steel specimens with five nitrogen concentrations were melted and subjected to compression and annealing. The grain boundary character distribution was investigated by electron backscatter diffraction. The corrosion behavior was investigated by the electrochemical polarization curves, electrical impedance spectroscopy, and X-ray photoelectron spectroscopy. The results showed that nitrogen promotes the formation of low-Σ boundaries when the pre-strain is not above 10%. The polarization resistance increases with increasing nitrogen content. The quantitative relationship between the nitrogen content and the pitting potential was obtained. Nitrogen favors the formation of oxides, hydroxides, and NH3/NH4+ on the passivated surface. The effect of planar slip on grain boundary engineering and the influence of NH4+ on the corrosion resistance are discussed.

Investigation of Microstructures and Tensile Properties of 316L Stainless Steel Fabricated via Laser Powder Bed Fusion.

In this study, a thorough investigation of the microstructures and tensile properties of 316L stainless steel fabricated via laser powder bed fusion (L-PBF) was done. 316L stainless steel specimens with two different thicknesses of 1.5 mm and 4.0 mm fabricated under similar conditions were utilized. Microstructural characterization was performed using optical microscopy (OM) and scanning electron microscopy (SEM) equipped with electron backscatter diffraction (EBSD). Melt pools and cellular structures were observed using OM, whereas EBSD was utilized to obtain the grain size, grain boundary characteristics, and crystallographic texture. The 1.5 mm thick sample demonstrated a yield strength (YS) of 538.42 MPa, ultimate tensile strength (UTS) of 606.47 MPa, and elongation to failure of 69.88%, whereas the 4.0 mm thick sample had a YS of 551.21 MPa, UTS of 619.58 MPa, and elongation to failure of 73.66%. These results demonstrated a slight decrease in mechanical properties with decreasing thickness, with a 2.4% reduction in YS, 2.1% reduction in UTS, and 5.8% reduction in elongation to failure. In addition to other microstructural features, the cellular structures were observed to be the major contributors to the high mechanical properties. Using the inverse pole figure (IPF) maps, both thicknesses depicted a crystallographic texture of {001} <101> in their as-built state. However, when subjected to tensile loads, texture transitions to {111} <001> and {111} <011> were observed for the 1.5 mm and 4.0 mm samples, respectively. Additionally, EBSD analysis revealed the pre-existence of high-density dislocation networks and a high fraction of low-angle grain boundaries. Interestingly, twinning was observed, suggesting that the plastic deformation occurred through dislocation gliding and deformation twinning.

Multi-branch residual image semantic segmentation combined with inverse weight gated-control

The loss of pixel-level information in the multi-class segmentation task based on the U-net model results in unclear boundaries and low semantic segmentation accuracy. Aiming at this, a deep multi-branch residual Unet (IWG-MRUN) with fused inverse weight gated-control is proposed to improve the quality of image semantic segmentation. Specifically, we first introduce a deep multi-branch residual module, which used parallel convolution mode to capture the contextual feature to extract the detailed features of the input image at a deeper level. Then, we adopt an inverse weight gated-control module to enhance the diversity of up-sampling information by counterclockwise transmitting attention horizontally to improve the restoration accuracy of up-sampled image pixels. Finally, to obtain finer granularity features from low spatial resolution images, we adopt the different receptive field pyramid attention mechanisms at the highest level of the U-shaped encoder to capture high-level context information at different scales, thereby improving the accuracy of semantic segmentation. The experimental results show that the segmentation accuracy of the proposed algorithm reaches 91.80% and the CCE loss is reduced to 0.21. When compared to the Unet, BiSeNet, DeeplabV3+ and U-net + BLR model, the pixel accuracy of semantic segmentation is improved by 15.0%, 1.98%, 0.9% and 6.5%, respectively. The semantic segmentation model proposed in this paper provides an end-to-end semantic segmentation capability with the enriched finer granularity features of the target boundary and realizes the accurate segmentation of the objects in different categories.

MRI- and DWI-Based Radiomics Features for Preoperatively Predicting Meningioma Sinus Invasion.

The aim of this study was to use multimodal imaging (contrast-enhanced T1-weighted (T1C), T2-weighted (T2), and diffusion-weighted imaging (DWI)) to develop a radiomics model for preoperatively predicting venous sinus invasion in meningiomas. This prediction would assist in selecting the appropriate surgical approach and forecasting the prognosis of meningiomas. A retrospective analysis was conducted on 331 participants who had been pathologically diagnosed with meningiomas. For each participant, 3948 radiomics features were acquired from the T1C, T2, and DWI images. Minimum redundancy maximum correlation, rank sum test, and multi-factor recursive elimination were used to extract the most significant features of different models. Then, multivariate logistic regression was used to build classification models to predict meningioma venous sinus invasion. The diagnostic capabilities were assessed using receiver operating characteristic (ROC) analysis. In addition, a nomogram was constructed by incorporating clinical and radiological characteristics and a radiomics signature. To assess the clinical usefulness of the nomogram, a decision curve analysis (DCA) was performed.Tumor shape, boundary, and enhancement features were independent predictors of meningioma venous sinus invasion (p = 0.013, p = 0.013, p = 0.005, respectively). Eleven (T2:1, T1C:4, DWI:6) of the 3948 radiomics features were screened for strong association with meningioma sinus invasion. The areas under the ROC curves for the training and external test sets were 0.946 and 0.874, respectively. The clinicoradiomic model showed excellent predictive performance for invasive meningioma, which may help to guide surgical approaches and predict prognosis.

Overcoming the Challenge of Accurate Segmentation of Lung Nodules: A Multi-crop CNN Approach.

Lung nodules are generated based on the growth of small and round- or oval-shaped cells in the lung, which are either cancerous or non-cancerous. Accurate segmentation of these nodules is crucial for early detection and diagnosis of lung cancer. However, lung nodules can have various shapes, sizes, and densities, making their accurate segmentation a difficult task. Moreover, they can be easily confused with other structures in the lung, including blood vessels and airways, further complicating the segmentation process. To address this challenge, this paper proposes a novel multi-crop convolutional neural network (multi-crop CNN) model that utilizes different sized cropped regions of CT scan images for accurate segmentation of lung nodules. The model consists of three modules, namely the feature representation module, boundary refinement module, and segmentation module. The feature representation module captures features from the lung CT scan image using cropped regions of different sizes, while the boundary refinement module combines the boundary maps and feature maps to generate a final feature map for the segmentation process. The segmentation module produces a high-resolution segmentation map that shows improved accuracy in segmenting cancerous lung nodules. The proposed multi-crop CNN model is evaluated on two segmentation datasets namely LUNA 16 and LIDC-IDRI with an accuracy of 98.3% and 98.5%, respectively. The performances are measured in terms of accuracy, recall, precision, dice coefficient, specificity, AUC/ROC, Hausdorff distance, Jaccard index, and average Hausdorff. Overall, the proposed multi-crop CNN model demonstrates the potential to enhance the lung nodule segmentation accuracy, which could lead to earlier detection and diagnosis of lung cancer and ultimately reduce mortality rates associated with the disease.