Search

Purpose: To develop a novel method based on feature selection, combining convolutional neural network (CNN) and ensemble learning (EL), to achieve high accuracy and efficiency of glioma detection and segmentation using multiparametric MRIs. Methods: We proposed an evolutionary feature selection‐based hybrid approach for glioma detection and segmentation on 4 MR sequences (T2‐FLAIR, T1, T1Gd, and T2). First, we trained a lightweight CNN to detect glioma and mask the suspected region to process large batch of MRI images. Second, we employed a differential evolution algorithm to search a feature space, which composed of 416‐dimensions radiomics features extracted from four sequences of MRIs and 128‐dimensions high‐order features extracted by the CNN, to generate an optimal feature combination for pixel classification. Finally, we trained an EL classifier using the optimal feature combination to segment whole tumor (WT) and its subregions including non‐enhancing tumor (NET), peritumoral edema (ED), and enhancing tumor (ET) in the suspected region. Experiments were carried out on 300 glioma patients from the BraTS2019 dataset using fivefold cross‐validation, and the model was independently validated using the rest 35 patients from the same database. Results: The approach achieved a detection accuracy of 98.8% using four MRIs. The Dice coefficients (and standard deviations) were 0.852 ± 0.057, 0.844 ± 0.046, and 0.799 ± 0.053 for segmentation of WT (NET+ET+ED), tumor core (NET+ET), and ET, respectively. The sensitivities and specificities were 0.873 ± 0.074, 0.863 ± 0.072, and 0.852 ± 0.082; the specificities were 0.994 ± 0.005, 0.994 ± 0.005, and 0.995 ± 0.004 for the WT, tumor core, and ET, respectively. The performances and calculation times were compared with the state‐of‐the‐art approaches, our approach yielded a better overall performance with average processing time of 139.5 s per set of four sequence MRIs. Conclusions: We demonstrated a robust and computational cost‐effective hybrid segmentation approach for glioma and its subregions on multi‐sequence MR images. The proposed approach can be used for automated target delineation for glioma patients. [ABSTRACT FROM AUTHOR]

With the increasing popularity and widespread application of embedded devices, lightweight target tracking networks have emerged as a crucial research direction in the field of computer vision. In this paper, we propose a deformable attention object tracking network based on cross-correlation (DAM-LOTNet), which utilizes the lightweight MobileNetV3 network for efficient feature extraction. Aiming at the problem that lightweight networks cannot better capture target features and morphological changes, deformable attention modules are proposed to generate attention maps along the channel and spatial dimensions. Furthermore, a dual cross-correlation operation is proposed, aiming at fusing channel information and pixel-level information. Thus, the model obtains a more comprehensive feature representation and a more accurate similarity calculation capability while maintaining a lightweight framework. Finally, a feature enhancement network constructed by depthwise separable convolution was added to the classification and regression branches. The designed DAM-LOTNet has 0.552M parameters and 0.121G FLOPs. On the UAV123 dataset, DAM-LOTNet achieves state-of-the-art performance compared to other lightweight object tracking algorithms. Extensive experiments on visual tracking benchmarks (including VOT2018, VOT2019, OTB100, etc.) have demonstrated that DAM-LOTNet has extremely low model FLOPs and parameters compared to common deep learning-based trackers, and is essentially on par with these trackers in terms of performance. • It has excellent tracking performance under the premise of lightweight. • Deformable attention module to better focus on the content of the deformed object. • Dual cross-correlation fuses channel information and pixel information. • Improve the model's ability to perceive local and global information. [ABSTRACT FROM AUTHOR]

Accurate polyp segmentation is of great significance for the prevention and diagnosis of early colon cancer. Transformer-based image segmentation models have been proposed for polyp segmentation with good results, however, these methods do not sufficiently consider the negative impact of background noise in different levels of features, resulting in the loss of local details. To alleviate this problem, we propose a GLSNet (A Global Guided Local Feature Stepwise Aggregation Network for polyp segmentation) network, including a spatial feature enhancement (SFE) module, a globally guided local feature enhancement (GLFE) module, and a feature stepwise aggregation (FSA) module. SFE can enhance the spatial feature representation of polyps to better capture the polyp information in the features. GLFE uses high-level features to filter noise in low-level features to capture polyp information hidden in shallow features; FSA fuses the semantic and positional information of polyps across scales to obtain the final segmentation results. Qualitative and quantitative experiments were conducted on 7 benchmark datasets, and the experimental results demonstrated that GLSNet outperforms other existing methods and has stronger generalization performance. In particular, we achieved a mean Dice of 92.9% on the large-scale Kvasir dataset, and mean Dice of 81.3% and 81.6% on CVC-ColonDB and ETIS, respectively, which are significantly higher than those of the competing methods. • We propose a Transformer-based network (GLSNet) for polyp segmentation. • Use a Spatial Feature Enhancement module to highlight regional features of interest. • Use a Global Guided Local Feature Enhancement module to remove noise from features. • Propose a Feature Stepwise Aggregation module to smoothly fuse multi-scale features. • GLSNet achieves the state-of-the-art performance in five polyp segmentation benchmarks. [ABSTRACT FROM AUTHOR]

This study concerns constructing an evolutionary search system to solve the global constrained optimization problems. Firstly, we proposed a hybrid constraint-handling method, called theta-mechanism, which blends two types of constraint-handling functions and alternates use of them in the searching process to balance two competing objectives: seeking as much as possible feasible regions and quickly converging to the optimum point in the found feasible regions. Secondly, to enable the search system to cooperate well with the theta-mechanism, we designed the cluster search algorithm (CSA) and developed the search reachability analysis (SRA) method. Based on SRA, we evaluated the characteristics of several typical search operators in order to assemble them into different operator combinations in CSA to maximize its performance, which enables CSA with theta-mechanism to accomplish the two inconsistent search objectives effectively. We tested the proposed method on 18 benchmark functions from IEEE CEC2010 and 32 real-world constrained optimization problems collected in IEEE CEC2020. Our results show the CSA with theta-mechanism is more competitive than the existing state-of-the-art approaches. • In this work, Cluster Search Algorithm (CSA) with theta constraint-handling mechanism for global constrained optimization is proposed. • The theta-mechanism is designed to blend two types of constraint-handling functions while alternating use of them in the searching process dynamically to balance two inconsistent objectives, which is vital to solve many complex constrained optimization problems. • CSA is developed to cooperate with theta- mechanism, meanwhile, a Search Reachability Analysis (SRA) method is proposed to analyze the characteristics of different search operators so as to form more efficient operator combinations to maximize the search performance of CSA. • Efficacy of CSA is showcased through extensive experiments. [ABSTRACT FROM AUTHOR]

• Self-healing reinforced concrete beams with engineered aggregates were developed and tested. • Mechanical crack-healing is observed for the first time using capsule based healing methods. • Advanced measurement techniques: DIC and UPV were used to confirm self-healing capability. This paper presents a study on the self-healing performance of reinforced concrete (RC) beams using a new approach called engineered aggregates (EA). EA are a type of capsules that release healing agents when the concrete is cracked, without requiring any changes to the conventional concrete preparation methods. In this study, polyurethane (PU) filled EA were added to RC beams with dimensions of 127 mm × 203 mm × 1448 mm and three-point bending tests were conducted. Companion control tests were also conducted on identical beams to compare the self-healing performance of the concrete with and without EA. The development of cracks during the first (initial cracking) and second (after healing) stages was monitored using the digital image correlation (DIC) technique. The addition of EA did not have any impact on the load carrying capacity of the beams, and the DIC results showed that the cracks closed during healing reopened in subsequent testing. Ultrasonic measurements were taken before and after healing to determine the extent of crack healing. The ultrasonic pulse velocity (UPV) measurements and diffusion coefficient calculations showed that the RC beams with PU-filled EA healed the cracks. Water absorption tests were performed to determine the recovery of the original permeability after healing. The study demonstrated the effectiveness of EA in improving the self-healing performance of RC beams and its potential for practical application in the construction industry. This paper is the first member scale study for self-healing concrete with EA, which demonstrated successful filling and bonding of flexural cracks. [ABSTRACT FROM AUTHOR]

This paper presents the results from a new approach for self-healing concrete. Conventional macro-capsules such as those made of glass cannot resist the harsh mixing process of concrete and require changes to conventional preparation methods. Here, a new encapsulation method; namely, engineered aggregates (EAs) were developed and tested, which do not require any changes to concrete preparation methods. EAs were shown to resist the conventional concrete mixing methods and effectively deliver healing agents when cracks appear in concrete. The mechanical properties of the produced self-healing concrete were examined by performing uniaxial compression, split tensile, and four-point bending tests. The self-healing efficiency of the concrete with EAs containing different healing agents (polyurethane, super glue, sodium silicate, magnesium oxidate) was determined based on split tensile, water absorption and water permeability tests. It was seen that more than 50% of the original strength and stiffness could be regained after self-healing. It was also found that the water permeability could be reduced by a factor 102 to 104 due to autonomous crack healing. Although some further research is still necessary concerning functional life of encapsulated healing agents and healing of cracks less than 50 μm, self-healing EA concrete was shown to be a novel approach that could lead to changes in the concrete industry. [ABSTRACT FROM AUTHOR]