Search

Intracranial hematoma due to traumatic brain injury is a serious health concern with rates of morbidity and mortality that are increasing worldwide. Manual identification is slow, subject to observer variabilities, and the existing automated techniques for intracranial hematoma detection in non-contrast computed tomography images cannot effectively detect multiple lesions of irregular sizes and shapes. Therefore, a computer-aided system using different window settings, YOLOv5s, cascaded attention module, and spatial pyramid pooling-fast is proposed to detect hematoma types, namely acute intraparenchymal, intraventricular, subdural, epidural, subarachnoid, and chronic subdural. Firstly, the computed tomography images are pre-processed using a window-based stacking approach wherein a three-channel image is generated by stacking grayscale images obtained with the aid of multiple window settings, i.e, brain, bone, and subdural. Secondly, a cascaded attention module is constructed in the neck of the YOLOv5s model to improve its detection performance by placing the convolution block attention module in serial with the efficient channel attention module. The cascaded attention module enriches the feature representation of various hematoma types in complex backgrounds especially when they are small and inconspicuous. The spatial pyramid pooling is replaced by a spatial pyramid pooling-fast to reduce the computational parameters and accelerate the feature fusion ability. The proposed deep learning model is trained, validated, and tested with 15,921 images from the brain haemorrhage extended dataset and it achieved overall precision, recall, F1-score, and mean average precision at 0.5, and mean average precision at 0.5:0.95 of 0.935, 0.908, 0.921, 0.943 and 0.65 respectively. The experimental results show that in comparison to the original YOLOv5s model and state-of-the-art methods, the model was able to localize and classify the acute or chronic instances of five hematoma subtypes in an individual image with improved precision and recall values. Hence the proposed system can be used in hospitals for the early and accurate detection of hematoma.

Copyright of Journal of Guangxi Normal University - Natural Science Edition is the property of Gai Kan Bian Wei Hui and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In sand-dust weather, the influence of sand-dust particles on imaging equipment often results in images with color deviation, blurring, and low contrast, among other issues. These problems making many traditional image restoration methods unable to accurately estimate the semantic information of the images and consequently resulting in poor restoration of clear images. Most current image restoration methods in the field of deep learning are based on supervised learning, which requires pairing and labeling a large amount of data, and the possibility of manual annotation errors. In light of this, we propose an unsupervised sand-dust image restoration network. The overall model adopts an improved CycleGAN to fit unpaired sand-dust images. Firstly, multiscale skip connections in the multiscale cascaded attention module are used to enhance the feature fusion effect after downsampling. Secondly, multi-head convolutional attention with multiple input concatenations is employed, with each head using different kernel sizes to improve the ability to restore detail information. Finally, the adaptive decoder-encoder module is used to achieve adaptive fitting of the model and output the restored image. According to the experiments conducted on the dataset, the qualitative and quantitative indicators of USIR-Net are superior to the selected comparison algorithms, furthermore, in additional experiments conducted on haze removal and underwater image enhancement, we have demonstrated the wide applicability of our model. [ABSTRACT FROM AUTHOR]

Identification of retinal diseases in automated screening methods, such as those used in clinical settings or computer-aided diagnosis, usually depends on the localization and segmentation of the Optic Disc (OD) and fovea. However, this task is difficult since these anatomical features have irregular spatial, texture, and shape characteristics, limited sample sizes, and domain shifts due to different data distributions across datasets. This study proposes a novel Multiresolution Cascaded Attention U-Net (MCAU-Net) model that addresses these problems by optimally balancing receptive field size and computational efficiency. The MCAU-Net utilizes two skip connections to accurately localize and segment the OD and fovea in fundus images. We incorporated a Multiresolution Wavelet Pooling Module (MWPM) into the CNN at each stage of U-Net input to compensate for spatial information loss. Additionally, we integrated a cascaded connection of the spatial and channel attentions as a skip connection in MCAU-Net to concentrate precisely on the target object and improve model convergence for segmenting and localizing OD and fovea centers. The proposed model has a low parameter count of 0.8 million, improving computational efficiency and reducing the risk of overfitting. For OD segmentation, the MCAU-Net achieves high IoU values of 0.9771, 0.945, and 0.946 for the DRISHTI-GS, DRIONS-DB, and IDRiD datasets, respectively, outperforming previous results for all three datasets. For the IDRiD dataset, the MCAU-Net locates the OD center with an Euclidean Distance (ED) of 16.90 pixels and the fovea center with an ED of 33.45 pixels, demonstrating its effectiveness in overcoming the common limitations of state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Infrared small-target detection is now commonly used in maritime surveillance, flight guidance, and other fields. However, extracting small targets from complex backgrounds remains a challenging task due to the small-target scale and complex imaging environment. Many studies are based on designing model structures to enhance the precision of target detection, and the number of Params and FLOPs has been significantly augmented. In this work, a knowledge distillation-based detection method (KDD) is proposed to overcome this challenge. KDD employs the small-target labeling information provided by a large-scale teacher model to refine the training process of students, thereby improving the performance and becoming lightweight. Specifically, we added efficient local attention (ELA), which can accurately identify areas of interest while avoiding dimensionality reduction. In addition, we also added the group aggregation bridge (GAB) module to connect low-level and high-level features for the fusion of different feature scales. Furthermore, a feature fusion loss was introduced to enhance the precision of target detection. Extensive evaluations have demonstrated that KDD performs better compared to several methods, achieving extremely low Params and FLOPs, as well as higher FPS. [ABSTRACT FROM AUTHOR]

RGB sign videos are easily influenced by light, perspective, and clothing, and skeleton data might be an effective complement. The challenge of multi-modal sign language recognition lies in the fusion of different modalities and the feature extraction of each modality. Therefore, we propose a continuous sign language recognition method based on interactive attention and improved graph convolutional networks (GCN). For the skeleton stream, a hand shift decoupling GCN is proposed to improve the graph modeling capabilities by capturing the interaction between the two hand nodes through the shift of hand features and decoupling operation. Then a cascaded attention module progressively in space, time, and channel dimensions effectively extracts high-level semantic features. To make the best of the two streams' features, an interactive attention mechanism is designed which takes the semantic features of the skeletons as queries and RGB features as keys and values, enhancing the spatial–temporal correlation mining ability and recognition performance of the network. The proposed method achieves competitive results on two public sign language datasets CSL and RWTH-PHOENIX-Weather-2014. • An interactive attention mechanism is designed to fuse multi-modal features. • HSD-GCN emphasizes the correlation between hands and achieves better features. • A cascaded attention module is developed in space, time, and channel dimensions. [ABSTRACT FROM AUTHOR]

A growing body of experimental evidence suggests that microRNAs (miRNAs) are closely associated with specific human diseases and play critical roles in their development and progression. Therefore, identifying miRNA related to specific diseases is of great significance for disease screening and treatment. In the early stages, the identification of associations between miRNAs and diseases demanded laborious and time-consuming biological experiments that often carried a substantial risk of failure. With the exponential growth in the number of potential miRNA-disease association combinations, traditional biological experimental methods face difficulties in processing massive amounts of data. Hence, developing more efficient computational methods to predict possible miRNA-disease associations and prioritize them is particularly necessary. In recent years, numerous deep learning-based computational methods have been developed and have demonstrated excellent performance. However, most of these methods rely on external databases or tools to compute various auxiliary information. Unfortunately, these external databases or tools often cover only a limited portion of miRNAs and diseases, resulting in many miRNAs and diseases being unable to match with these computational methods. Therefore, there are certain limitations associated with the practical application of these methods. To overcome the above limitations, this study proposes a multi-view computational model called MVNMDA, which predicts potential miRNA-disease associations by integrating features of miRNA and diseases from local views, global views, and semantic views. Specifically, MVNMDA utilizes known association information to construct node initial features. Then, multiple networks are constructed based on known association to extract low-dimensional feature embedding of all nodes. Finally, a cascaded attention classifier is proposed to fuse features from coarse to fine, suppressing noise within the features and making precise predictions. To validate the effectiveness of the proposed method, extensive experiments were conducted on the HMDD v2.0 and HMDD v3.2 datasets. The experimental results demonstrate that MVNMDA achieves better performance compared to other computational methods. Additionally, the case study results further demonstrate the reliable predictive performance of MVNMDA. [ABSTRACT FROM AUTHOR]

Glaucoma is a serious eye disease and glaucoma optic disc hemorrhage (GODH) is an important diagnostic indicator for glaucoma. Deep-learning-based medical image segmentation methods for automatic optic cup and disc segmentation have made tremendous progress. However, when it comes to the segmentation of GODH, classical deep learning technologies face two main challenges: the difficulties in distinguishing GODH from the end points or bending points of blood vessels, and the imbalance between the pixel classes of the target area and the background area. In this paper, we proposed a deep learning framework integrating expert knowledge (E-Net) for the segmentation of GODH in fundus images. This E-Net consisted of a primary network for GODH segmentation and two auxiliary networks for extraction of optic disc (OD) and blood vessels. The segmentation probability maps from the two auxiliary networks were used to improve the segmentation accuracy of GODH, via expert knowledge loss functions and attention mechanism. Moreover, we designed a weighted segmentation accuracy loss function to balance the segmentation accuracy of the target and background region, thus fully mining the substantial information in the fundus images. The proposed E-Net was verified on a GODH dataset from Beijing Tongren Hospital. The experiments showed that the proposed E-Net achieved state-of-the-art results on this dataset. [ABSTRACT FROM AUTHOR]

Classification of plants based on a multi-organ approach is very challenging. Although additional data provide more information that might help to disambiguate between species, the variability in shape and appearance in plant organs also raises the degree of complexity of the problem. Despite promising solutions built using deep learning enable representative features to be learned for plant images, the existing approaches focus mainly on generic features for species classification, disregarding the features representing plant organs. In fact, plants are complex living organisms sustained by a number of organ systems. In our approach, we introduce a hybrid generic-organ convolutional neural network (HGO-CNN), which takes into account both organ and generic information, combining them using a new feature fusion scheme for species classification. Next, instead of using a CNN-based method to operate on one image with a single organ, we extend our approach. We propose a new framework for plant structural learning using the recurrent neural network-based method. This novel approach supports classification based on a varying number of plant views, capturing one or more organs of a plant, by optimizing the contextual dependencies between them. We also present the qualitative results of our proposed models based on feature visualization techniques and show that the outcomes of visualizations depict our hypothesis and expectation. Finally, we show that by leveraging and combining the aforementioned techniques, our best network outperforms the state of the art on the PlantClef2015 benchmark. The source code and models are available at https://github.com/cs-chan/Deep-Plant. [ABSTRACT FROM AUTHOR]

Ancient murals are important cultural heritages for our exploration of ancient civilizations and are of great research value. Due to long-time exposure to the environment, ancient murals often suffer from damage (deterioration) such as cracks, scratches, corrosion, paint loss, and even large-region falling off. It is an urgent work to protect and restore these damaged ancient murals. Mural inpainting techniques refer to virtually filling the deteriorated regions by reconstructing the structure and texture elements of the mural images. Most existing mural inpainting approaches fail to fill loss contents that contain complex structures and diverse patterns since they neglect the importance of structure guidance. In this paper, we propose a structure-guided two-branch model based on the generative adversarial network (GAN) for ancient mural inpainting. In the proposed model, the mural inpainting process can be divided into two stages: structure reconstruction and content restoration. These two stages are conducted by using a structure reconstruction network (SRN) and a content restoration network (CRN), respectively. In the structure reconstruction stage, SRN employs the Gated Convolution and the Fast Fourier Convolution (FFC) residual block to reconstruct the missing structures of the damaged murals. In the content restoration stage, CRN uses the structures (generated by SRN) to guide the missing content restoration of the murals. We design a two-branch parallel encoder to improve the texture and color restoration quality for the missing regions of the murals. Moreover, we propose a cascade attention module that can capture long-term relevance information in the deep features. It helps to alleviate the texture-blur and color-bias problem. We conduct experiments on both simulated and real damaged murals, and compare our inpainting results with other four competitive approaches. Experimental results show that our proposed model outperforms other approaches in terms of texture clarity, color consistency and structural continuity of the restored mural images. In addition, the mural inpainting results of our model can achieve comparatively high quantitative evaluation metrics. [ABSTRACT FROM AUTHOR]

Facial expression methods play a vital role in human–computer interaction and other fields, but there are factors such as occlusion, illumination, and pose changes in wild facial recognition, as well as category imbalances between different datasets, that result in large variations in recognition rates and low accuracy rates for different categories of facial expression datasets. This study introduces RCL-Net, a method of recognizing wild facial expressions that is based on an attention mechanism and LBP feature fusion. The structure consists of two main branches, namely the ResNet-CBAM residual attention branch and the local binary feature (LBP) extraction branch (RCL-Net). First, by merging the residual network and hybrid attention mechanism, the residual attention network is presented to emphasize the local detail feature information of facial expressions; the significant characteristics of facial expressions are retrieved from both channel and spatial dimensions to build the residual attention classification model. Second, we present a locally improved residual network attention model. LBP features are introduced into the facial expression feature extraction stage in order to extract texture information on expression photographs in order to emphasize facial feature information and enhance the recognition accuracy of the model. Lastly, experimental validation is performed using the FER2013, FERPLUS, CK+, and RAF-DB datasets, and the experimental results demonstrate that the proposed method has superior generalization capability and robustness in the laboratory-controlled environment and field environment compared to the most recent experimental methods. [ABSTRACT FROM AUTHOR]

How to obtain discriminative features has proved to be a core problem for image retrieval. Many recent works use convolutional neural networks to extract features. However, clutter and occlusion will interfere with the distinguishability of features when using convolutional neural network (CNN) for feature extraction. To address this problem, we intend to obtain high-response activations in the feature map based on the attention mechanism. We propose two attention modules, a spatial attention module and a channel attention module. For the spatial attention module, we first capture the global information and model the relation between channels as a region evaluator, which evaluates and assigns new weights to local features. For the channel attention module, we use a vector with trainable parameters to weight the importance of each feature map. The two attention modules are cascaded to adjust the weight distribution for the feature map, which makes the extracted features more discriminative. Furthermore, we present a scale and mask scheme to scale the major components and filter out the meaningless local features. This scheme can reduce the disadvantages of the various scales of the major components in images by applying multiple scale filters, and filter out the redundant features with the MAX-Mask. Exhaustive experiments demonstrate that the two attention modules are complementary to improve performance, and our network with the three modules outperforms the state-of-the-art methods on four well-known image retrieval datasets. [ABSTRACT FROM AUTHOR]

Recovering facial details from dark images has attracted increasing attention due to its potential in various applications such as video surveillance. We propose the first approach to detect and enhance human faces in extremely low-light images. We at first propose an attention module (AM) to detect the facial skin which is relatively robust to the low-quality condition. The AM further locates the landmarks as the prior knowledge to facilitate the reconstruction. Then, with the detected face position, our face hallucination module (FHM) could focus on enhancing the resolution and quality of the face. Moreover, we also introduce a low-light enhancement module to enhance the global image to merge with the hallucinated face from FHM for the final images. Extensive experiments show our method is quantitatively and qualitatively superior to the state-of-the-art in terms of enhancement quality and face hallucination. [ABSTRACT FROM AUTHOR]

Polar codes are closer to the Shannon limit with lower complexity in coding and decoding. As traditional decoding techniques suffer from high latency and low throughput, with the development of deep learning technology, some deep learning-based decoding methods have been proposed to solve these problems. Usually, the deep neural network is treated as a black box and learns to map the polar codes with noise to the original information code directly. In fact, it is difficult for the network to distinguish between valid and interfering information, which leads to limited BER performance. In this paper, a deep residual network based on information refinement (DIR-NET) is proposed for decoding polar-coded short packets. The proposed method works to fully distinguish the effective and interference information in the codewords, thus obtaining a lower bit error rate. To achieve this goal, we design a two-stage decoding network, including a denoising subnetwork and decoding subnetwork. This structure can further improve the accuracy of the decoding method. Furthermore, we construct the whole network solely on the basis of the attention mechanism. It has a stronger information extraction ability than the traditional neural network structure. Benefiting from cascaded attention modules, information can be filtered and refined step-by-step, thus obtaining a low bit error rate. The simulation results show that DIR-Net outperforms existing decoding methods in terms of BER performance under both AWGN channels and flat fading channels. [ABSTRACT FROM AUTHOR]

Detections of the uterus using preoperative magnetic resonance imaging (MRI) data are required for intraoperative navigation in High-Intensity Focused Ultrasound (HIFU) treatment of uterine fibroids. The modelling of the intraoperative uterus between empty and full bladder anatomy deems to be a crucial step to quantify uterus deformation caused by the bladder-and-rectum-filling (BRF) technique. In this paper, a novel two-stage conditional generative adversarial network (cGAN) is proposed to perform intraoperative uterus deformable reconstruction with anatomical constraints using only single preoperative MRI data, viz. ACDeformRec. The highly constrained anatomy properties are further captured by a novel attention network that formulates high-level uterus segmentation task by incorporating correlations of surrounding organs for more anatomically plausible reconstruction. The experimental result demonstrates the robustness of ACDeformRec by evaluating its performance on 181 clinical datasets and has achieved the lowest reconstruction error of 0.735 ± 0.045 mm with Dice Similarity Coefficient of 94.23% and Normalised Cross Correlation of 97.23%. [ABSTRACT FROM AUTHOR]

Pan-sharpening methods based on deep neural network (DNN) have produced state-of-the-art fusion performance. However, DNN-based methods mainly focus on the modeling of the local properties in low spatial resolution multispectral (LR MS) and panchromatic (PAN) images by convolution neural networks. The global dependencies in the images are ignored. To capture the local and global properties of the images concurrently, we propose a multiscale spatial–spectral interaction transformer (MSIT) for pan-sharpening. Specifically, we construct the multiscale sub-networks containing convolution–transformer encoder to extract the local and global features at different scales from LR MS and PAN images, respectively. Then, a spatial–spectral interaction attention module (SIAM) is designed to merge the features at each scale. In SIAM, the interaction attention is used to decouple the spatial and spectral information efficiently for the enhancement of complementarity and the reduction of redundancy in the extracted features. The features from different scales are further integrated into a multiscale reconstruction module (MRM) to generate the desired high spatial resolution multispectral image, in which the spatial and spectral information is fused scale by scale. The experiments on reduced- and full-scale datasets demonstrate that the proposed MSIT can produce better results in terms of visual and numerical analysis when compared with state-of-the-art methods. [ABSTRACT FROM AUTHOR]

BACKGROUND: Convolutional neural network has achieved a profound effect on cardiac image segmentation. The diversity of medical imaging equipment brings the challenge of domain shift for cardiac image segmentation. OBJECTIVE: In order to solve the domain shift existed in multi-modality cardiac image segmentation, this study aims to investigate and test an unsupervised domain adaptation network RA-SIFA, which combines a parallel attention module (PAM) and residual attention unit (RAU). METHODS: First, the PAM is introduced in the generator of RA-SIFA to fuse global information, which can reduce the domain shift from the respect of image alignment. Second, the shared encoder adopts the RAU, which has residual block based on the spatial attention module to alleviate the problem that the convolution layer is insensitive to spatial position. Therefore, RAU enables to further reduce the domain shift from the respect of feature alignment. RA-SIFA model can realize the unsupervised domain adaption (UDA) through combining the image and feature alignment, and then solve the domain shift of cardiac image segmentation in a complementary manner. RESULTS: The model is evaluated using MM-WHS2017 datasets. Compared with SIFA, the Dice of our new RA-SIFA network is improved by 8.4%and 3.2%in CT and MR images, respectively, while, the average symmetric surface distance (ASD) is reduced by 3.4 and 0.8mm in CT and MR images, respectively. CONCLUSION: The study results demonstrate that our new RA-SIFA network can effectively improve the accuracy of whole-heart segmentation from CT and MR images. [ABSTRACT FROM AUTHOR]

The performance of a facial expression recognition network degrades obviously under situations of uneven illumination or partial occluded face as it is quite difficult to pinpoint the attention hotspots on the dynamically changing regions (e.g., eyes, nose, and mouth) as precisely as possible. To address the above issue, by a hybrid of the attention mechanism and pyramid feature, this paper proposes a cascade attention-based facial expression recognition network on the basis of a combination of (i) local spatial feature, (ii) multi-scale-stereoscopic spatial context feature (extracted from the 3-scale pyramid feature), and (iii) temporal feature. Experiments on the CK+, Oulu-CASIA, and RAF-DB datasets obtained recognition accuracy rates of 99.23%, 89.29%, and 86.80%, respectively. It demonstrates that the proposed method outperforms the state-of-the-art methods in both the experimental and natural environment. [ABSTRACT FROM AUTHOR]

Analyzing the surface and bedrock locations in radar imagery enables the computation of ice sheet thickness, which is important for the study of ice sheets, their volume and how they may contribute to global climate change. However, the traditional handcrafted methods cannot quickly provide quantitative, objective and reliable extraction of information from radargrams. Most traditional handcrafted methods, designed to detect ice-surface and ice-bed layers from ice sheet radargrams, require complex human involvement and are difficult to apply to large datasets, while deep learning methods can obtain better results in a generalized way. In this study, an end-to-end multi-scale attention network (MsANet) is proposed to realize the estimation and reconstruction of layers in sequences of ice sheet radar tomographic images. First, we use an improved 3D convolutional network, C3D-M, whose first full connection layer is replaced by a convolution unit to better maintain the spatial relativity of ice layer features, as the backbone. Then, an adjustable multi-scale module uses different scale filters to learn scale information to enhance the feature extraction capabilities of the network. Finally, an attention module extended to 3D space removes a redundant bottleneck unit to better fuse and refine ice layer features. Radar sequential images collected by the Center of Remote Sensing of Ice Sheets in 2014 are used as training and testing data. Compared with state-of-the-art deep learning methods, the MsANet shows a 10% reduction (2.14 pixels) on the measurement of average mean absolute column-wise error for detecting the ice-surface and ice-bottom layers, runs faster and uses approximately 12 million fewer parameters. [ABSTRACT FROM AUTHOR]