Search

PROteolysis TArgeting Chimeras (PROTACs) has recently emerged as a promising technology. However, the design of rational PROTACs, especially the linker component, remains challenging due to the absence of structure-activity relationships and experimental data. Leveraging the structural characteristics of PROTACs, fragment-based drug design (FBDD) provides a feasible approach for PROTAC research. Concurrently, artificial intelligence-generated content has attracted considerable attention, with diffusion models and Transformers emerging as indispensable tools in this field. In response, we present a new diffusion model, DiffPROTACs, harnessing the power of Transformers to learn and generate new PROTAC linkers based on given ligands. To introduce the essential inductive biases required for molecular generation, we propose the O(3) equivariant graph Transformer module, which augments Transformers with graph neural networks (GNNs), using Transformers to update nodes and GNNs to update the coordinates of PROTAC atoms. DiffPROTACs effectively competes with existing models and achieves comparable performance on two traditional FBDD datasets, ZINC and GEOM. To differentiate the molecular characteristics between PROTACs and traditional small molecules, we fine-tuned the model on our self-built PROTACs dataset, achieving a 93.86% validity rate for generated PROTACs. Additionally, we provide a generated PROTAC database for further research, which can be accessed at https://bailab.siais.shanghaitech.edu.cn/service/DiffPROTACs-generated.tgz. The corresponding code is available at https://github.com/Fenglei104/DiffPROTACs and the server is at https://bailab.siais.shanghaitech.edu.cn/services/diffprotacs., (© The Author(s) 2024. Published by Oxford University Press.)

Objectives: To evaluate the performance of automatic deep learning (DL) algorithm for size, mass, and volume measurements in predicting prognosis of lung adenocarcinoma (LUAD) and compared with manual measurements., Methods: A total of 542 patients with clinical stage 0-I peripheral LUAD and with preoperative CT data of 1-mm slice thickness were included. Maximal solid size on axial image (MSSA) was evaluated by two chest radiologists. MSSA, volume of solid component (SV), and mass of solid component (SM) were evaluated by DL. Consolidation-to-tumor ratios (CTRs) were calculated. For ground glass nodules (GGNs), solid parts were extracted with different density level thresholds. The prognosis prediction efficacy of DL was compared with that of manual measurements. Multivariate Cox proportional hazards model was used to find independent risk factors., Results: The prognosis prediction efficacy of T-staging (TS) measured by radiologists was inferior to that of DL. For GGNs, MSSA-based CTR measured by radiologists ( R MSSA%) could not stratify RFS and OS risk, whereas measured by DL using 0HU ( 2D-AI MSSA 0HU %) could by using different cutoffs. SM and SV measured by DL using 0 HU ( AI SM 0HU % and AI SV 0HU %) could effectively stratify the survival risk regardless of different cutoffs and were superior to 2D-AI MSSA 0HU %. AI SM 0HU % and AI SV 0HU % were independent risk factors., Conclusion: DL algorithm can replace human for more accurate T-staging of LUAD. For GGNs, 2D-AI MSSA 0HU % could predict prognosis rather than R MSSA%. The prediction efficacy of AI SM 0HU % and AI SV 0HU % was more accurate than of 2D-AI MSSA 0HU % and both were independent risk factors., Clinical Relevance Statement: Deep learning algorithm could replace human for size measurements and could better stratify prognosis than manual measurements in patients with lung adenocarcinoma., Key Points: • Deep learning (DL) algorithm could replace human for size measurements and could better stratify prognosis than manual measurements in patients with lung adenocarcinoma (LUAD). • For GGNs, maximal solid size on axial image (MSSA)-based consolidation-to-tumor ratio (CTR) measured by DL using 0 HU could stratify survival risk than that measured by radiologists. • The prediction efficacy of mass- and volume-based CTRs measured by DL using 0 HU was more accurate than of MSSA-based CTR and both were independent risk factors., (© 2023. The Author(s), under exclusive licence to European Society of Radiology.)

Background: Deep learning (DL) prediction models hold great promise in the triage of COVID-19., Objective: We aimed to evaluate the diagnostic test accuracy of DL prediction models for assessing and predicting the severity of COVID-19., Methods: We searched PubMed, Scopus, LitCovid, Embase, Ovid, and the Cochrane Library for studies published from December 1, 2019, to April 30, 2022. Studies that used DL prediction models to assess or predict COVID-19 severity were included, while those without diagnostic test accuracy analysis or severity dichotomies were excluded. QUADAS-2 (Quality Assessment of Diagnostic Accuracy Studies 2), PROBAST (Prediction Model Risk of Bias Assessment Tool), and funnel plots were used to estimate the bias and applicability., Results: A total of 12 retrospective studies involving 2006 patients reported the cross-sectionally assessed value of DL on COVID-19 severity. The pooled sensitivity and area under the curve were 0.92 (95% CI 0.89-0.94; I 2 =0.00%) and 0.95 (95% CI 0.92-0.96), respectively. A total of 13 retrospective studies involving 3951 patients reported the longitudinal predictive value of DL for disease severity. The pooled sensitivity and area under the curve were 0.76 (95% CI 0.74-0.79; I 2 =0.00%) and 0.80 (95% CI 0.76-0.83), respectively., Conclusions: DL prediction models can help clinicians identify potentially severe cases for early triage. However, high-quality research is lacking., Trial Registration: PROSPERO CRD42022329252; https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD 42022329252., (©Changyu Wang, Siru Liu, Yu Tang, Hao Yang, Jialin Liu. Originally published in the Journal of Medical Internet Research (https://www.jmir.org), 21.07.2023.)

Introduction: Early detection of lung cancer is one way to improve outcomes. Improving the detection of nodules on chest CT scans is important. Previous artificial intelligence (AI) modules show rapid advantages, which improves the performance of detecting lung nodules in some datasets. However, they have a high false-positive (FP) rate. Its effectiveness in clinical practice has not yet been fully proven. We aimed to use AI assistance in CT scans to decrease FP. Materials and methods: CT images of 60 patients were obtained. Five senior doctors who were blinded to these cases participated in this study for the detection of lung nodules. Two doctors performed manual detection and labeling of lung nodules without AI assistance. Another three doctors used AI assistance to detect and label lung nodules before manual interpretation. The AI program is based on a deep learning framework. Results: In total, 266 nodules were identified. For doctors without AI assistance, the FP was 0.617-0.650/scan and the sensitivity was 59.2-67.0%. For doctors with AI assistance, the FP was 0.067 to 0.2/scan and the sensitivity was 59.2-77.3% This AI-assisted program significantly reduced FP. The error-prone characteristics of lung nodules were central locations, ground-glass appearances, and small sizes. The AI-assisted program improved the detection of error-prone nodules. Conclusions: Detection of lung nodules is important for lung cancer treatment. When facing a large number of CT scans, error-prone nodules are a great challenge for doctors. The AI-assisted program improved the performance of detecting lung nodules, especially for error-prone nodules., Competing Interests: Competing Interests: The authors have declared that no competing interest exists., (© The author(s).)

Background Separate noncontrast CT to quantify the coronary artery calcium (CAC) score often precedes coronary CT angiography (CTA). Quantifying CAC scores directly at CTA would eliminate the additional radiation produced at CT but remains challenging. Purpose To quantify CAC scores automatically from a single CTA scan. Materials and Methods In this retrospective study, a deep learning method to quantify CAC scores automatically from a single CTA scan was developed on training and validation sets of 292 patients and 73 patients collected from March 2019 to July 2020. Virtual noncontrast scans obtained with a spectral CT scanner were used to develop the algorithm to alleviate tedious manual annotation of calcium regions. The proposed method was validated on an independent test set of 240 CTA scans collected from three different CT scanners from August 2020 to November 2020 using the Pearson correlation coefficient, the coefficient of determination, or r , and the Bland-Altman plot against the semiautomatic Agatston score at noncontrast CT. The cardiovascular risk categorization performance was evaluated using weighted κ based on the Agatston score (CAC score risk categories: 0-10, 11-100, 101-400, and >400). Results Two hundred forty patients (mean age, 60 years ± 11 [standard deviation]; 146 men) were evaluated. The positive correlation between the automatic deep learning CTA and semiautomatic noncontrast CT CAC score was excellent (Pearson correlation = 0.96; 2 = 0.92). The risk categorization agreement based on deep learning CTA and noncontrast CT CAC scores was excellent (weighted κ = 0.94 [95% CI: 0.91, 0.97]), with 223 of 240 scans (93%) categorized correctly. All patients who were miscategorized were in the direct neighboring risk groups. The proposed method's differences from the noncontrast CT CAC score were not statistically significant with regard to scanner ( r 2 = 0.92). The risk categorization agreement based on deep learning CTA and noncontrast CT CAC scores was excellent (weighted κ = 0.94 [95% CI: 0.91, 0.97]), with 223 of 240 scans (93%) categorized correctly. All patients who were miscategorized were in the direct neighboring risk groups. The proposed method's differences from the noncontrast CT CAC score were not statistically significant with regard to scanner ( P = .67). Conclusion A deep learning automatic calcium scoring method accurately quantified coronary artery calcium from CT angiography images and categorized risk. © RSNA, 2021 See also the editorial by Goldfarb and Cao et al in this issue.P = .051), and section thickness ( P = .67). Conclusion A deep learning automatic calcium scoring method accurately quantified coronary artery calcium from CT angiography images and categorized risk. © RSNA, 2021 See also the editorial by Goldfarb and Cao et al in this issue.

Data-driven deep learning has been considered a promising method for building powerful models for medical data, which often requires a large amount of diverse data to be sufficiently effective. However, the expensive cost of collecting and the privacy constraints lead to the fact that existing medical datasets are small-scale and distributed. Federated learning via model distillation is a data-private collaborative learning where the model can leverage all available data without direct sharing. The data knowledge is shared by distillation through the multi-site average prediction scores on the public dataset. However, the average consensus is suboptimal to individual client due to data domain shift in MRI data caused by acquisition protocols, recruitment criteria, etc. In this work, we propose a federated conditional mutual learning (FedCM) to improve the performance by considering the clients' local performance and the similarity between clients. This work is the first federated learning on multi-dataset Alzheimer's disease classification by 3DCNN using T1w MRI. Our method achieves the best recognition rates comparing with FedMD and other frameworks. Further visualization and relevance ranking on the region of interests (ROI) in human brains implies that the left hemisphere may have greater relevance than the right hemisphere does. Several potential regions are listed for future investigation.

Purpose: Microvascular invasion (MVI) is a valuable predictor of survival in hepatocellular carcinoma (HCC) patients. This study developed predictive models using eXtreme Gradient Boosting (XGBoost) and deep learning based on CT images to predict MVI preoperatively., Methods: In total, 405 patients were included. A total of 7302 radiomic features and 17 radiological features were extracted by a radiomics feature extraction package and radiologists, respectively. We developed a XGBoost model based on radiomics features, radiological features and clinical variables and a three-dimensional convolutional neural network (3D-CNN) to predict MVI status. Next, we compared the efficacy of the two models., Results: Of the 405 patients, 220 (54.3%) were MVI positive, and 185 (45.7%) were MVI negative. The areas under the receiver operating characteristic curves (AUROCs) of the Radiomics-Radiological-Clinical (RRC) Model and 3D-CNN Model in the training set were 0.952 (95% confidence interval (CI) 0.923-0.973) and 0.980 (95% CI 0.959-0.993), respectively (p = 0.14). The AUROCs of the RRC Model and 3D-CNN Model in the validation set were 0.887 (95% CI 0.797-0.947) and 0.906 (95% CI 0.821-0.960), respectively (p = 0.83). Based on the MVI status predicted by the RRC and 3D-CNN Models, the mean recurrence-free survival (RFS) was significantly better in the predicted MVI-negative group than that in the predicted MVI-positive group (RRC Model: 69.95 vs. 24.80 months, p < 0.001; 3D-CNN Model: 64.06 vs. 31.05 months, p = 0.027)., Conclusion: The RRC Model and 3D-CNN models showed considerable efficacy in identifying MVI preoperatively. These machine learning models may facilitate decision-making in HCC treatment but requires further validation.

With the development of image-guided surgery and radiotherapy, the demand for medical image registration is stronger and the challenge is greater. In recent years, deep learning, especially deep convolution neural networks, has made excellent achievements in medical image processing, and its research in registration has developed rapidly. In this paper, the research progress of medical image registration based on deep learning at home and abroad is reviewed according to the category of technical methods, which include similarity measurement with an iterative optimization strategy, direct estimation of transform parameters, etc. Then, the challenge of deep learning in medical image registration is analyzed, and the possible solutions and open research are proposed.

Background: Recently, deep learning technologies have rapidly expanded into medical image analysis, including both disease detection and classification. As far as we know, migraine is a disabling and common neurological disorder, typically characterized by unilateral, throbbing and pulsating headaches. Unfortunately, a large number of migraineurs do not receive the accurate diagnosis when using traditional diagnostic criteria based on the guidelines of the International Headache Society. As such, there is substantial interest in developing automated methods to assist in the diagnosis of migraine., Methods: To the best of our knowledge, no studies have evaluated the potential of deep learning technologies in assisting with the classification of migraine patients. Here, we used deep learning methods in combination with three functional measures (the amplitude of low-frequency fluctuations, regional homogeneity and regional functional correlation strength) based on rs-fMRI data to distinguish not only between migraineurs and healthy controls, but also between the two subtypes of migraine. We employed 21 migraine patients without aura, 15 migraineurs with aura, and 28 healthy controls., Results: Compared with the traditional support vector machine classifier, which has an accuracy of 83.67%, our Inception module-based convolutional neural network approach showed a significant improvement in classification output (over 86.18%). Our data also indicate that the Inception module-based CNN performs better than the AlexNet-based CNN (Inception module-based CNN reached an accuracy of 99.25%). Finally, we also found that regional functional correlation strength (RFCS) could be regarded as the optimum input out of the three indices (ALFF, ReHo, RFCS)., Conclusions: Overall, our study shows that combining the three functional measures of rs-fMRI with deep learning classification is a powerful method to distinguish between migraineurs and healthy individuals. Our data also highlight that deep learning-based frameworks could be used to develop more complicated models or systems to aid in clinical decision making in the future.

Short-term precipitation forecasting is essential for agriculture, transportation, urban management, and tourism. The radar echo extrapolation method is widely used in precipitation forecasting. To address issues like forecast degradation, insufficient capture of spatiotemporal dependencies, and low accuracy in radar echo extrapolation, we propose a new model: MS-DD3D-RSTN. This model employs spatiotemporal convolutional blocks (STCBs) as spatiotemporal feature extractors and uses the spatial-temporal loss (STLoss) function to learn intra-frame and inter-frame changes for end-to-end training, thereby capturing the spatiotemporal dependencies in radar echo signals. Experiments on the Sichuan dataset and the HKO-7 dataset show that the proposed model outperforms advanced models in terms of CSI and POD evaluation metrics. For 2 h forecasts with 20 dBZ and 30 dBZ reflectivity thresholds, the CSI metrics reached 0.538, 0.386, 0.485, and 0.198, respectively, representing the best levels among existing methods. The experiments demonstrate that the MS-DD3D-RSTN model enhances the ability to capture spatiotemporal dependencies, mitigates forecast degradation, and further improves radar echo prediction performance. [ABSTRACT FROM AUTHOR]

Crop leaf chlorophyll content (LCC) and fractional vegetation cover (FVC) are crucial indicators for assessing crop health, growth development, and maturity. In contrast to the traditional manual collection of crop trait parameters, unmanned aerial vehicle (UAV) technology rapidly generates LCC and FVC maps for breeding materials, facilitating prompt assessments of maturity information. This study addresses the following research questions: (1) Can image features based on pretrained deep learning networks and ensemble learning enhance the estimation of remote sensing LCC and FVC? (2) Can the proposed adaptive normal maturity detection (ANMD) algorithm effectively monitor maize maturity based on LCC and FVC maps? We conducted the following tasks: (1) Seven phases (tassel initiation to maturity) of maize canopy orthoimages and corresponding ground-truth data for LCC and six phases of FVC using UAVs were collected. (2) Three features, namely vegetation indices (VI), texture features (TF) based on Gray Level Co-occurrence Matrix, and deep features (DF), were evaluated for LCC and FVC estimation. Moreover, the potential of four single-machine learning models and three ensemble models for LCC and FVC estimation was evaluated. (3) The estimated LCC and FVC were combined with the proposed ANMD to monitor maize maturity. The research findings indicate that (1) image features extracted from pretrained deep learning networks more accurately describe crop canopy structure information, effectively eliminating saturation effects and enhancing LCC and FVC estimation accuracy. (2) Ensemble models outperform single-machine learning models in estimating LCC and FVC, providing greater precision. Remarkably, the stacking + DF strategy achieved optimal performance in estimating LCC (coefficient of determination (R2): 0.930; root mean square error (RMSE): 3.974; average absolute error (MAE): 3.096); and FVC (R2: 0.716; RMSE: 0.057; and MAE: 0.044). (3) The proposed ANMD algorithm combined with LCC and FVC maps can be used to effectively monitor maize maturity. Establishing the maturity threshold for LCC based on the wax ripening period (P5) and successfully applying it to the wax ripening-mature period (P5–P7) achieved high monitoring accuracy (overall accuracy (OA): 0.9625–0.9875; user's accuracy: 0.9583–0.9933; and producer's accuracy: 0.9634–1). Similarly, utilizing the ANMD algorithm with FVC also attained elevated monitoring accuracy during P5–P7 (OA: 0.9125–0.9750; UA: 0.878–0.9778; and PA: 0.9362–0.9934). This study offers robust insights for future agricultural production and breeding, offering valuable insights for the further exploration of crop monitoring technologies and methodologies. [ABSTRACT FROM AUTHOR]

Crack detection plays a critical role in ensuring road safety and maintenance. Traditional, manual, and semi-automatic detection methods have proven inefficient. Nowadays, the emergence of deep learning techniques has opened up new possibilities for automatic crack detection. However, there are few methods with both localization and segmentation abilities, and most perform poorly. The consistent nature of pavement over a small mileage range gives us the opportunity to make improvements. A novel data-augmentation strategy called CrackMover, specifically tailored for crack detection methods, is proposed. Experiments demonstrate the effectiveness of CrackMover for various methods. Moreover, this paper presents a new instance segmentation method for crack detection. It adopts a redesigned backbone network and incorporates a cascade structure for the region-based convolutional network (R-CNN) part. The experimental evaluation showcases significant performance improvements achieved by these approaches in crack detection. The proposed method achieves an average precision of 33.3%, surpassing Mask R-CNN with a Residual Network 50 backbone by 8.6%, proving its effectiveness in detecting crack distress. [ABSTRACT FROM AUTHOR]

Motivation The prediction of binding affinity between drug and target is crucial in drug discovery. However, the accuracy of current methods still needs to be improved. On the other hand, most deep learning methods focus only on the prediction of non-covalent (non-bonded) binding molecular systems, but neglect the cases of covalent binding, which has gained increasing attention in the field of drug development. Results In this work, a new attention-based model, A Transformer Encoder and Fingerprint combined Prediction method for Drug–Target Affinity (TEFDTA) is proposed to predict the binding affinity for bonded and non-bonded drug–target interactions. To deal with such complicated problems, we used different representations for protein and drug molecules, respectively. In detail, an initial framework was built by training our model using the datasets of non-bonded protein–ligand interactions. For the widely used dataset Davis, an additional contribution of this study is that we provide a manually corrected Davis database. The model was subsequently fine-tuned on a smaller dataset of covalent interactions from the CovalentInDB database to optimize performance. The results demonstrate a significant improvement over existing approaches, with an average improvement of 7.6% in predicting non-covalent binding affinity and a remarkable average improvement of 62.9% in predicting covalent binding affinity compared to using BindingDB data alone. At the end, the potential ability of our model to identify activity cliffs was investigated through a case study. The prediction results indicate that our model is sensitive to discriminate the difference of binding affinities arising from small variances in the structures of compounds. Availability and implementation The codes and datasets of TEFDTA are available at https://github.com/lizongquan01/TEFDTA. [ABSTRACT FROM AUTHOR]

With the development of cloud computing and deep learning, an increasing number of artificial intelligence models have been applied to reality. Such as videos on cell phones can be uploaded to the cloud for storage, which is detected by cloud arithmetic. Nevertheless, achieving this goal requires frequent consideration of the security of the model, since videos or images that go to the cloud, it is very likely to receive an adversarial attack. Regarding object detection, there has however been slow advancement in robustness research in this area. This is because training a target detection model requires a lot of arithmetic and time. Moreover, the current research has only slightly reduced the gap between clean and adversarial samples. To alleviate this problem, we propose a uniform perspective object detection robustness model based on bilinear interpolation that can accurately identify clean and adversarial samples. We propose the robustness optimization based on uniform metric perspective (RPU) for feature learning of clean and adversarial samples, drawing on the fine-grained idea. Following this, we analyze the fragility of the adversarial samples and consequently use the proposed perturbation filtering verification (PVB) based on bilinear interpolation. With slightly degraded clean sample detection performance, it substantially improves the robustness of object detection. Extensive experiments on PASCAL VOC and MS COCO show that our model guarantees the detection performance of clean samples and increases the detection performance of adversarial samples. The work we did has been open-sourced on GitHub: https://github.com/KujouRiu/RPU-PVB. [ABSTRACT FROM AUTHOR]

Accurate wind speed prediction is significantly important for the full utilization of wind energy resources and the improvement in the economic benefits of wind farms. Because the ensemble forecast takes into account the uncertainty of information about the atmospheric motion, domestic and foreign weather service forecast centers often choose to use the ensemble numerical forecast to achieve the fine forecast of wind speed. However, due to the unavoidable systematic errors of the ensemble numerical forecast model, it is necessary to correct the deviation in the ensemble numerical forecast wind speed. Considering the typical spatio-temporal characteristics of the grid prediction data of the wind field, based on Convolutional Long–Short Term Memory (ConvLSTM) units and attention mechanism, this paper takes the complex and representative North China region as the research area, aiming to reveal the shortcomings of existing deep learning integrated prediction correction models in extracting temporal features of grid prediction data. We propose a new ensemble prediction wind field correction model integrating multi-factor and spatio-temporal characteristics. This model uses reanalyzed land data provided by the European Center for Medium-Range Weather Forecasts as the real data to correct the deviation in the near-surface 10 m wind field data predicted by the regional ensemble numerical prediction model of the China Meteorological Administration. We used the reanalyzed land data provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) as the live data to correct the deviation in the near-surface 10 m wind field data predicted by the regional ensemble numerical forecast model of the China Meteorological Administration (CMA). At the same time, Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) were used as the scoring indicators, and the results of the China Meteorological Administration–Regional Ensemble Prediction System (CMA–REPS) ensemble average, multiple linear regression method correction, Long–Short Term Memory (LSTM) method correction, and U-net (UNET) method correction were compared. Compared with the UNET model method, the experimental results show that when processing the 10 m zonal wind data, 10 m meridional wind data, and 10 m average wind speed data of CMA–REPS 24 h forecasts, the correction results of our model can reduce the RMSE score index by 9.15%, 4.83%, and 7.79%. At the same time, when processing the 48 h and 72 h near-surface 10 m wind field data of the CMA–REPS forecast, our model can improve the prediction accuracy of CMA–REPS near-surface wind forecast data. Therefore, the correction effect of the proposed model in a complex terrain area is evidently better compared to other methods. [ABSTRACT FROM AUTHOR]

Ensuring road safety, structural stability and durability is of paramount importance, and detecting road cracks plays a critical role in achieving these goals. We propose a GM-ResNet-based method to enhance the precision and efficacy of crack detection. Leveraging ResNet-34 as the foundational network for crack image feature extraction, we consider the challenge of insufficient global and local information assimilation within the model. To overcome this, we incorporate the global attention mechanism into the architecture, facilitating comprehensive feature extraction across the channel and the spatial width and height dimensions. This dynamic interaction across these dimensions optimizes feature representation and generalization, resulting in a more precise crack detection outcome. Recognizing the limitations of ResNet-34 in managing intricate data relationships, we replace its fully connected layer with a multilayer fully connected neural network. We fashion a deep network structure by integrating multiple linear, batch normalization and activation function layers. This construction amplifies feature expression, stabilizes training convergence and elevates the performance of the model in complex detection tasks. Moreover, tackling class imbalance is imperative in road crack detection. Introducing the focal loss function as the training loss addresses this challenge head-on, effectively mitigating the adverse impact of class imbalance on model performance. The experimental outcomes on a publicly available crack dataset emphasize the advantages of the GM-ResNet in crack detection accuracy compared to other methods. It is worth noting that the proposed method has better evaluation indicators in the detection results compared with alternative methodologies, highlighting its effectiveness. This validates the potency of our method in achieving optimal crack detection outcomes. [ABSTRACT FROM AUTHOR]

Compressed sensing (CS) MRI has shown great potential in enhancing time efficiency. Deep learning techniques, specifically generative adversarial networks (GANs), have emerged as potent tools for speedy CS-MRI reconstruction. Yet, as the complexity of deep learning reconstruction models increases, this can lead to prolonged reconstruction time and challenges in achieving convergence. In this study, we present a novel GAN-based model that delivers superior performance without the model complexity escalating. Our generator module, built on the U-net architecture, incorporates dilated residual (DR) networks, thus expanding the network's receptive field without increasing parameters or computational load. At every step of the downsampling path, this revamped generator module includes a DR network, with the dilation rates adjusted according to the depth of the network layer. Moreover, we have introduced a channel attention mechanism (CAM) to distinguish between channels and reduce background noise, thereby focusing on key information. This mechanism adeptly combines global maximum and average pooling approaches to refine channel attention. We conducted comprehensive experiments with the designed model using public domain MRI datasets of the human brain. Ablation studies affirmed the efficacy of the modified modules within the network. Incorporating DR networks and CAM elevated the peak signal-to-noise ratios (PSNR) of the reconstructed images by about 1.2 and 0.8 dB, respectively, on average, even at 10× CS acceleration. Compared to other relevant models, our proposed model exhibits exceptional performance, achieving not only excellent stability but also outperforming most of the compared networks in terms of PSNR and SSIM. When compared with U-net, DR-CAM-GAN's average gains in SSIM and PSNR were 14% and 15%, respectively. Its MSE was reduced by a factor that ranged from two to seven. The model presents a promising pathway for enhancing the efficiency and quality of CS-MRI reconstruction. [ABSTRACT FROM AUTHOR]

Background: Identifying cervical lymph node metastasis (LNM) in primary thyroid cancer preoperatively using ultrasound is challenging. Therefore, a non-invasive method is needed to assess LNM accurately. Purpose: To address this need, we developed the Primary Thyroid Cancer Lymph Node Metastasis Assessment System (PTC-MAS), a transfer learning-based and B-mode ultrasound images-based automatic assessment system for assessing LNM in primary thyroid cancer. Methods: The system has two parts: YOLO Thyroid Nodule Recognition System (YOLOS) for obtaining regions of interest (ROIs) of nodules, and LMM assessment system for building the LNM assessment system using transfer learning and majority voting with extracted ROIs as input. We retained the relative size features of nodules to improve the system's performance. Results: We evaluated three transfer learning-based neural networks (DenseNet, ResNet, and GoogLeNet) and majority voting, which had the area under the curves (AUCs) of 0.802, 0.837, 0.823, and 0.858, respectively. Method III preserved relative size features and achieved higher AUCs than Method II, which fixed nodule size. YOLOS achieved high precision and sensitivity on a test set, indicating its potential for ROIs extraction. Conclusions: Our proposed PTC-MAS system effectively assesses primary thyroid cancer LNM based on preserving nodule relative size features. It has potential for guiding treatment modalities and avoiding inaccurate ultrasound results due to tracheal interference. [ABSTRACT FROM AUTHOR]

The acquisition of maize tassel phenotype information plays a vital role in studying maize growth and improving yield. Unfortunately, detecting maize tassels has proven challenging because of the complex field environment, including image resolution, varying sunlight conditions, plant varieties, and planting density. To address this situation, the present study uses unmanned aerial vehicle (UAV) remote sensing technology and a deep learning algorithm to facilitate maize tassel identification and counting. UAVs are used to collect maize tassel images in experimental fields, and RetinaNet serves as the basic model for detecting maize tassels. Small maize tassels are accurately identified by optimizing the feature pyramid structure in the model and introducing attention mechanisms. We also study how mapping differences in image resolution, brightness, plant variety, and planting density affect the RetinaNet model. The results show that the improved RetinaNet model is significantly better at detecting maize tassels than the original RetinaNet model. The average precision in this study is 0.9717, the precision is 0.9802, and the recall rate is 0.9036. Compared with the original model, the improved RetinaNet improves the average precision, precision, and recall rate by 1.84%, 1.57%, and 4.6%, respectively. Compared with mainstream target detection models such as Faster R-CNN, YOLOX, and SSD, the improved RetinaNet model more accurately detects smaller maize tassels. For equal-area images of differing resolution, maize tassel detection becomes progressively worse as the resolution decreases. We also analyze how detection depends on brightness in the various models. With increasing image brightness, the maize tassel detection worsens, especially for small maize tassels. This paper also analyzes the various models for detecting the tassels of five maize varieties. Zhengdan958 tassels prove the easiest to detect, with R2 = 0.9708, 0.9759, and 0.9545 on 5, 9, and 20 August 2021, respectively. Finally, we use the various models to detect maize tassels under different planting densities. At 29,985, 44,978, 67,466, and 89,955 plants/hm2, the mean absolute errors for detecting Zhengdan958 tassels are 0.18, 0.26, 0.48, and 0.63, respectively. Thus, the detection error increases gradually with increasing planting density. This study thus provides a new method for high-precision identification of maize tassels in farmland and is especially useful for detecting small maize tassels. This technology can be used for high-throughput investigations of maize phenotypic traits. [ABSTRACT FROM AUTHOR]

Purpose: The majority of existing studies about named entity recognition (NER) concentrate on the prediction enhancement of deep neural network (DNN)-based models themselves, but the issues about the scarcity of training corpus and the difficulty of annotation quality control are not fully solved, especially for Chinese ancient corpora. Therefore, designing a new integrated solution for Chinese historical NER, including automatic entity extraction and man-machine cooperative annotation, is quite valuable for improving the effectiveness of Chinese historical NER and fostering the development of low-resource information extraction. Design/methodology/approach: The research provides a systematic approach for Chinese historical NER with a three-stage framework. In addition to the stage of basic preprocessing, the authors create, retrain and yield a high-performance NER model only using limited labeled resources during the stage of augmented deep active learning (ADAL), which entails three steps—DNN-based NER modeling, hybrid pool-based sampling (HPS) based on the active learning (AL), and NER-oriented data augmentation (DA). ADAL is thought to have the capacity to maintain the performance of DNN as high as possible under the few-shot constraint. Then, to realize machine-aided quality control in crowdsourcing settings, the authors design a stage of globally-optimized automatic label consolidation (GALC). The core of GALC is a newly-designed label consolidation model called simulated annealing-based automatic label aggregation ("SA-ALC"), which incorporates the factors of worker reliability and global label estimation. The model can assure the annotation quality of those data from a crowdsourcing annotation system. Findings: Extensive experiments on two types of Chinese classical historical datasets show that the authors' solution can effectively reduce the corpus dependency of a DNN-based NER model and alleviate the problem of label quality. Moreover, the results also show the superior performance of the authors' pipeline approaches (i.e. HPS + DA and SA-ALC) compared to equivalent baselines in each stage. Originality/value: The study sheds new light on the automatic extraction of Chinese historical entities in an all-technological-process integration. The solution is helpful to effectively reducing the annotation cost and controlling the labeling quality for the NER task. It can be further applied to similar tasks of information extraction and other low-resource fields in theoretical and practical ways. [ABSTRACT FROM AUTHOR]

The adversarial attack is crucial to improving the robustness of deep learning models; they help improve the interpretability of deep learning and also increase the security of the models in real-world applications. However, existing attack algorithms mainly focus on image classification tasks, and they lack research targeting object detection. Adversarial attacks against image classification are global-based with no focus on the intrinsic features of the image. In other words, they generate perturbations that cover the whole image, and each added perturbation is quantitative and undifferentiated. In contrast, we propose a global-to-local adversarial attack based on object detection, which destroys important perceptual features of the object. More specifically, we differentially extract gradient features as a proportion of perturbation additions to generate adversarial samples, as the magnitude of the gradient is highly correlated with the model's point of interest. In addition, we reduce unnecessary perturbations by dynamically suppressing excessive perturbations to generate high-quality adversarial samples. After that, we improve the effectiveness of the attack using the high-frequency feature gradient as a motivation to guide the next gradient attack. Numerous experiments and evaluations have demonstrated the effectiveness and superior performance of our from global to Local gradient attacks with high-frequency momentum guidance (GLH), which is more effective than previous attacks. Our generated adversarial samples also have excellent black-box attack ability. [ABSTRACT FROM AUTHOR]

To solve slow speed and low accuracy of traditional detection methods of tunnel lining cracks, especially under the complicated situation of tunnel in operation, this work proposed an improved You Only Look Once version X (YOLOX) tunnel lining crack image detection algorithm. First, Mobilenetv3 was used to replace YOLOX's CSPDarknet network. The Efficient Channel Attention (ECA) module was then added to the enhanced feature extraction network, and the IOU loss function was replaced by the generalised IOU (GIOU) loss function. A tunnel crack image data set was constructed and used to compare the performance of the improved YOLOX algorithm with that of five other algorithms. The improved YOLOX algorithm solves the shortcomings of the other five algorithms. The results showed that the improved YOLOX algorithm had 82.48% F1 score and 87.28% AP value, which is higher than that of the other five algorithms at varying degrees. In addition, the data size of the improved YOLOX model was 51.2 M, which is 75.27% compressed compared to the YOLOX model. The time was 16.52 ms, and the FPS was 60.52 frames/s. Therefore, the proposed improved YOLOX algorithm can realise the high-speed, high-precision, real-time dynamic detection of tunnel lining cracks in complicated environments. [ABSTRACT FROM AUTHOR]

The monitoring and prevention of cracks in tunnel structures is an essential problem and the traditional methods are arduous to suffice the rapid increasing of tunnel mileage. Many researches, nowadays, mainly focus on the automatic identification and intelligent detection of cracks. However, crack identification and localization of tunnel structures are merely preliminary for effective prevention. Therefore, how to intelligently analyze and predict the further evolution of cracks will be the top priority of tunnel disaster prevention. In this article, a novel method is proposed which combinates deep learning (DL) techniques and finite element method to detect and analyze the cracks which is significant for the intelligent prevention. The innovation of this article lies that the position of cracks is identified automatically with the DL based on Mask R-CNN method and the evolution of cracks is predicted efficiently with FEM model. [ABSTRACT FROM AUTHOR]

In this study, a deep learning method called Lightning-SN was developed and used for cloud-to-ground (CG) lightning identification. Based on artificial scenarios, this network model selects radar products that exhibit characteristic factors closely related to lightning. Advanced time of arrival and direction lightning positioning data were used as the labeling factors. The Lightning-SN model was constructed based on an encoder–decoder structure with 25 convolutional layers, five pooling layers, five upsampling layers, and a sigmoid activation function layer. Additionally, the maximum pooling index method was adopted in Lightning-SN to avoid characteristic boundary information loss in the pooling process. The gradient harmonizing mechanism was used as the loss function to improve the model performance. The evaluation results showed that the Lightning-SN improved the segmentation accuracy of the CG lightning location compared with the traditional threshold method, according to the 6-minute operating period of the current S-band Doppler radar, exhibiting a better performance in terms of lightning location identification based on high-resolution radar data. The model was applied to the Ningbo area of Zhejiang Province, China. It was applied to the lightning hazard prevention in the hazardous chemical park in Ningbo. The composite reflectivity and radial velocity were the two dominant factors, with a greater influence on the model performance than other factors. [ABSTRACT FROM AUTHOR]

Deep reinforcement learning (DRL) targets to train an autonomous agent to interact with a pre-defined environment and strives to achieve specific goals through deep neural networks (DNN). Recurrent neural network (RNN) based DRL has demonstrated superior performance, as RNNs can effectively capture the temporal evolution of the environment and respond with proper agent actions. However, apart from the outstanding performance, little is known about how RNNs understand the environment internally and what has been memorized over time. Revealing these details is extremely important for deep learning experts to understand and improve DRLs, which in contrast, is also challenging due to the complicated data transformations inside these models. In this article, we propose Deep Reinforcement Learning Interactive Visual Explorer (DRLIVE), a visual analytics system to effectively explore, interpret, and diagnose RNN-based DRLs. Having focused on DRL agents trained for different Atari games, DRLIVE accomplishes three tasks: game episode exploration, RNN hidden/cell state examination, and interactive model perturbation. Using the system, one can flexibly explore a DRL agent through interactive visualizations, discover interpretable RNN cells by prioritizing RNN hidden/cell states with a set of metrics, and further diagnose the DRL model by interactively perturbing its inputs. Through concrete studies with multiple deep learning experts, we validated the efficacy of DRLIVE. [ABSTRACT FROM AUTHOR]

The accurate intelligent identification and detection of road cracks is a key issue in road maintenance, and it has become popular to perform this task through the field of computer vision. In this paper, we proposed a deep learning-based crack detection method that initially uses the idea of image sparse representation and compressed sensing to preprocess the datasets. Only the pixels that represent the crack features remain, while most pixels of non-crack features are relatively sparse, which can significantly improve the accuracy and efficiency of crack identification. The proposed method achieved good results based on the limited datasets of crack images. Various algorithms were tested, namely, linear smooth, median filtering, Gaussian smooth, and grayscale threshold, where the optimal parameters of the various algorithms were analyzed and trained with faster regions with convolutional neural network features (faster R-CNN). The results of the experiments showed that the proposed method has good robustness, with higher detection efficiency in the presence of, for example, road markings, shallow cracks, multiple cracks, and blurring. The result shows that the improvement of mean average precision (mAP) can reach 5% compared with the original method. [ABSTRACT FROM AUTHOR]

Stochastic Neural Networks (SNNs) have emerged as a promising tool for improving model adversarial robustness by injecting uncertainty into model activations or weights. However, the current implementations are dominated by injecting fixed Gaussian noises. Despite its ease of use, Gaussian distribution also has several limitations, e.g., inaccurate uncertainty estimation due to its unimodality. In this paper, we propose a non-informative noise-enhanced Stochastic Neural Network (NINE-SNN), which relaxes the Gaussian distribution to a non-informative prior arbitrary distribution and encourages the model to learn an appropriate uncertainty. We provide theoretical insights showing that by adding increasing levels of stochastic noise to a DNN, the model naturally becomes more resistant to input perturbations. We evaluate the proposed method on various benchmarks of architectures and well-known white-box and black-box attacks. The results show that the proposed method achieves state-of-the-art performances without adversarial training, demonstrating the superiority and efficiency of NINE-SNN. Compared to adversarial training, NINE-SNN has saved about 7x computation time cost, and has nearly no accuracy loss for clean data accuracy. Moreover, it achieves SOTA results in SNNs network architectures and Non-stochastic network architectures in robust data accuracy. [Display omitted] • Introduce a novel stochastic neural network. • Extend beyond Gaussian by placing an arbitrary distribution over a non-informative prior. • Augment model uncertainty through network training. • The proposed method as an implicit regularization for robustness. • SOTA performance without the need for adversarial training. [ABSTRACT FROM AUTHOR]

With the significant increase of e-commerce, freight transportation demand has surged significantly over the past decade. Most of the demand has been served by trucks in the United States. One major problem commonly identified across the country is the worsening truck parking availability because the increase of truck parking facilities has lagged behind the growth of trucking activities. The lack of parking spaces and real-time parking availability information greatly exacerbate the uncertainty of trips, and often results in illegal and potentially dangerous parking or overtime driving. This paper elaborates on pilot research on improving truck parking facilities cooperated with the Washington State Department of Transportation (WSDOT), building and testing the advanced Truck Parking Information and Management System (TPIMS) with the real-time user visualization and prediction function empowered by artificial intelligence. Furthermore, by analyzing the activities of truck drivers, the researchers aggregated the regularity of truck parking patterns by a customized sequential similarity methodology. A Truck Parking Occupancy Prediction (TPOP) neural network for time-variant occupancy prediction by deep learning and attributes embedding is proposed and integrated into the TPIMS. The TPOP achieves 5.82%, 5.07%, 4.84%, and 4.19% mean average percentage error (MAPE) for 16, 8, 4, and 2 minutes ahead of occupancy prediction respectively, significantly outperforms other state-of-the-art methods. Clearly, the proposed solutions can benefit both the truck drivers and government agencies by a more efficient and smart TPIMS. [ABSTRACT FROM AUTHOR]

Managing and estimating the availability information of parking lots is of great importance to travelers and managers. However, the task is very challenging since the occupancy rate is affected by various factors, including spatial‐temporal features, parking lot attributes features, and environmental changes. Previous studies mostly focus on the short‐term prediction by capturing the historical sequential dependencies among inputs and outputs, which leads to low estimation accuracy for long‐term prediction and limited scalability for real deployment in parking lots. To address the challenges, a comprehensive framework for real‐time Smart Parking Data Management and Prediction (SPDMP) system is proposed. Three types of data sources, including historical sequential data, real‐time sequential data, and attributes category data, are sufficiently integrated into a customized Parking Availability Prediction (PAP) neural network by representation learning and heterogeneous feature embedding. Specifically, instead of using parking lot property information and environmental data directly, the authors design an Attributes Tensor Embedding Component (ATEC) to integrate the intra‐class affection and interclass representations and correlations by two steps of customized embedding process. To balance the impact of the indiscriminate features for various prediction targets, this study proposes a multi‐factor attention mechanism to learn the weights and help the PAP achieve a stable performance for time‐various tasks. From extensive experiments on two real‐world large‐scale data sets collected in China and USA (including 19 urban parking and 2 truck parking lots), PAP achieves superior prediction results on both urban parking (6.69% average MAPE) and truck parking prediction (7.63% average MAPE) for five prediction time slots (5 min, 10 min, 30 min, 1 h, and 2 h). Furthermore, even with limited training data, PAP still shows better transferability and adaptability for various types of lots. The proposed SPDMP is selected and used as a pilot test bed by the Washington State Department of Transportation (WSDOT) for truck parking information promotion. [ABSTRACT FROM AUTHOR]

Video-based dynamic facial emotion recognition (FER) is a challenging task, as one must capture and distinguish tiny facial movements representing emotional changes while ignoring the facial differences of different objects. Recent state-of-the-art studies have usually adopted more complex methods to solve this task, such as large-scale deep learning models or multimodal analysis with reference to multiple sub-models. According to the characteristics of the FER task and the shortcomings of existing methods, in this paper we propose a lightweight method and design three attention modules that can be flexibly inserted into the backbone network. The key information for the three dimensions of space, channel, and time is extracted by means of convolution layer, pooling layer, multi-layer perception (MLP), and other approaches, and attention weights are generated. By sharing parameters at the same level, the three modules do not add too many network parameters while enhancing the focus on specific areas of the face, effective feature information of static images, and key frames. The experimental results on CK+ and eNTERFACE'05 datasets show that this method can achieve higher accuracy. [ABSTRACT FROM AUTHOR]

To alleviate the social contradiction between limited medical resources and increasing medical needs, the medical image-assisted diagnosis based on deep learning has become the research focus in Wise Information Technology of med. Most of the existing medical segmentation models based on Convolution or Transformer have achieved relatively sound effects. However, the Convolution-based model with a limited receptive field cannot establish long-distance dependencies between features as the Network deepens. The Transformer-based model produces large computation overhead and cannot generalize the bias of local features and perceive the position feature of medical images, which are essential in medical image segmentation. To address those issues, we present Triple Gate MultiLayer Perceptron U-Net (TGMLP U-Net), a medical image segmentation model based on MLP, in which we design the Triple Gate MultiLayer Perceptron (TGMLP), composed of three parts. Firstly, considering encoding the position information of features, we propose the Triple MLP module based on MultiLayer Perceptron in this model. It uses linear projection to encode features from the high, wide, and channel dimensions, enabling the model to capture the long-distance dependence of features along the spatial dimension and the precise position information of features in three dimensions with less computational overhead. Then, we design the Local Priors and Global Perceptron module. The Global Perceptron divides the feature map into different partitions and conducts correlation modelling for each partition to establish the global dependency between partitions. The Local Priors uses multi-scale Convolution with high local feature extraction ability to explore further the relationship of context feature information within the structure. At last, we suggest a Gate-controlled Mechanism to effectively solves the problem that the dependence of position embeddings between Patches and within Patches in medical images cannot be well learned due to the relatively small number of samples in medical images segmentation data. Experimental results indicate that the proposed model outperforms other state-of-the-art models in most evaluation indicators, demonstrating its excellent performance in segmenting medical images. [ABSTRACT FROM AUTHOR]

Customized path-based speed prediction is an eventful tool for congestion avoidance, route optimization and travel time prediction for navigation apps, cab-hailing companies and autonomous vehicles. Traditionally, the speed prediction algorithms are based on road segments and can only support several main roads. Path-based speed prediction is very challenging since the speed is always changing in different path locations and is jointly affected by lots of complicated factors. This article presents a novel deep learning framework for customized path-based speed prediction. A Path-based Speed Prediction Neural Network (PSPNN) is designed to achieve speed predictions for a given path and attributes information. A hierarchical Convolutional Neural Network (CNN) and deep Bidirectional Long Short-Term Memory (Bi-LSTM) structure for different kinds of feature extraction are applied for multiple levels: the path cell, sub-path and the whole path. The method narrows down the prediction unit from road segments to customized path cells (mean length: 59.52m) and achieves a mean absolute error (MAE) of 1.94 m/s and Mean Absolute Percentage Error (MAPE) of 18.14%, showing the potential of serving rigorous data-driven applications. So far, PSPNN is the first made-to-order path-based speed prediction algorithm and can help both travelers and managers to obtain large-scale bespoke paths speed information in advance. [ABSTRACT FROM AUTHOR]

In the field of deep learning, deep neural networks (DNNs) have shown good performance on classification applications. However, a DNN model is vulnerable to adversarial examples, which is formed by adding tiny perturbations on a normal example and can mislead the DNN model to make a wrong estimate during the prediction. In this brief, for adversarial attacks in radio signals field, we propose a novel adversarial example detection strategy based on multifeature fusion and provide a framework which includes generating adversarial examples, extracting the local intrinsic dimensionality (LID) features and the constellation diagram (CD) features, detecting adversarial examples. We obtain the output values of normal examples and adversarial examples in each layer of the model respectively, and then, calculate the LID features values of examples by the maximum likelihood estimate based on a certain neighborhood range. Meanwhile, we calculate the CD features values by the range feature and density feature of the constellation diagram distribution. Finally, a logistic regression classifier is trained based on multifeature fusion values to detect adversarial examples. The experimental results across two benchmark datasets demonstrate that the proposed multifeature fusion method could accurately detect adversarial examples of radio signals. The detection accuracy is up to 98.7% when the perturbation reached 10%. [ABSTRACT FROM AUTHOR]

Multi-contrast magnetic resonance (MR) image registration is useful in the clinic to achieve fast and accurate imaging-based disease diagnosis and treatment planning. Nevertheless, the efficiency and performance of the existing registration algorithms can still be improved. In this paper, we propose a novel unsupervised learning-based framework to achieve accurate and efficient multi-contrast MR image registration. Specifically, an end-to-end coarse-to-fine network architecture consisting of affine and deformable transformations is designed to improve the robustness and achieve end-to-end registration. Furthermore, a dual consistency constraint and a new prior knowledge-based loss function are developed to enhance the registration performances. The proposed method has been evaluated on a clinical dataset containing 555 cases, and encouraging performances have been achieved. Compared to the commonly utilized registration methods, including VoxelMorph, SyN, and LT-Net, the proposed method achieves better registration performance with a Dice score of 0.8397± 0.0756 in identifying stroke lesions. With regards to the registration speed, our method is about 10 times faster than the most competitive method of SyN (Affine) when testing on a CPU. Moreover, we prove that our method can still perform well on more challenging tasks with lacking scanning information data, showing the high robustness for the clinical application. [ABSTRACT FROM AUTHOR]