DL-based Model Research Articles

VSHARP: Variable Splitting Half-quadratic ADMM algorithm for reconstruction of inverse-problems

Medical Imaging (MI) tasks, such as accelerated parallel Magnetic Resonance Imaging (MRI), often involve reconstructing an image from noisy or incomplete measurements. This amounts to solving ill-posed inverse problems, where a satisfactory closed-form analytical solution is not available. Traditional methods such as Compressed Sensing (CS) in MRI reconstruction can be time-consuming or prone to obtaining low-fidelity images. Recently, a plethora of Deep Learning (DL) approaches have demonstrated superior performance in inverse-problem solving, surpassing conventional methods. In this study, we propose vSHARP (variable Splitting Half-quadratic ADMM algorithm for Reconstruction of inverse Problems), a novel DL-based method for solving ill-posed inverse problems arising in MI. vSHARP utilizes the Half-Quadratic Variable Splitting method and employs the Alternating Direction Method of Multipliers (ADMM) to unroll the optimization process. For data consistency, vSHARP unrolls a differentiable gradient descent process in the image domain, while a DL-based denoiser, such as a U-Net architecture, is applied to enhance image quality. vSHARP also employs a dilated-convolution DL-based model to predict the Lagrange multipliers for the ADMM initialization. We evaluate vSHARP on tasks of accelerated parallel MRI Reconstruction using two distinct datasets and on accelerated parallel dynamic MRI Reconstruction using another dataset. Our comparative analysis with state-of-the-art methods demonstrates the superior performance of vSHARP in these applications.

Deep-learning model accurately classifies multi-label lung ultrasound findings, enhancing diagnostic accuracy and inter-reader agreement.

Despite the increasing use of lung ultrasound (LUS) in the evaluation of respiratory disease, operators' competence constrains its effectiveness. We developed a deep-learning (DL) model for multi-label classification using LUS and validated its performance and efficacy on inter-reader variability. We retrospectively collected LUS and labeled as normal, B-line, consolidation, and effusion from patients undergoing thoracentesis at a tertiary institution between January 2018 and January 2022. The development and internal testing involved 7580 images fromJanuary 2018and December2020, andthe model's performance was validated on a temporally separated test set (n = 985 imagescollected after January 2021) and two external test sets (n = 319 and 54 images). Two radiologists interpreted LUS with and without DL assistance and compared diagnostic performance and agreement. The model demonstrated robust performancewith AUCs: 0.93 (95% CI 0.92-0.94) for normal, 0.87 (95% CI 0.84-0.89) for B-line, 0.82 (95% CI 0.78-0.86) for consolidation, and 0.94 (95% CI 0.93-0.95) for effusion.The model improved reader accuracy for binary discrimination (normal vs. abnormal; reader 1: 87.5-95.6%, p = 0.004;reader 2: 95.0-97.5%, p = 0.19), and agreement(k = 0.73-0.83, p = 0.01). In conclusion,the DL-based model may assist interpretation, improving accuracy and overcoming operator competence limitationsin LUS.

Application of machine learning for predicting lymph node metastasis in T1 colorectal cancer: a systematic review and meta-analysis.

We review and analyze research on the application of machine learning (ML) and deep learning (DL) models to lymph node metastasis (LNM) prediction in patients with T1 colorectal cancer (CRC). Predicting LNM before radical surgery is important in patients with T1 CRC. However, current surgical treatment guidelines are limited. LNM prediction using ML or DL may improve predictive accuracy. The diagnostic accuracy of LNM prediction using ML- and DL-based models for patients with CRC was assessed. We performed a comprehensive search of the PubMed, Embase, and Cochrane databases (inception to April 30th of 2022) for studies that applied ML or DL to LNM prediction in T1 CRC patients specifically to compare with histopathological findings and not related to radiological aspects. 33,199 T1 CRC patients enrolled across seven studies with a retrospective design were included. LNM was observed in 3,173 (9.6%) patients. Overall, the ML- and DL-based model exhibited a sensitivity of 0.944 and specificity of 0.877 for the prediction of LNM in patients with T1 CRC. Six different types of ML and DL models were used across the studies included in this meta-analysis. Therefore, a high degree of heterogeneity was observed. The ML and DL models provided high sensitivity and specificity for predicting LNM in patients with T1 CRC, and the heterogeneity between studies was significant. These results suggest the potential of ML or DL as diagnostic tools. However, more reliable algorithms should be developed for predicting LNM before surgery in patients with T1 CRC.

Deep learning–based statistical robustness evaluation of intensity-modulated proton therapy for head and neck cancer

A dense, dilated 3D U-net was trained to predict the 5th and 95th percentile distributions of planning dose plan using the nominal dose and planning CT images. The data set comprised proton therapy plans for 582 H&N cancer patients. Ground truth percentile values were estimated for each patient through 600 dose recalculations, representing randomly sampled uncertainty scenarios. The comprehensive comparisons of different models were conducted for H&N cancer patients, considering those with and without a beam mask (BM) and diverse beam configurations, including varying beam angles, couch angles, and beam numbers. The performance of our model trained based on a mixture of patients with H&N and prostate cancer was also assessed in contrast with models trained based on data specific for patients with cancer at either site. The DL-based model's predictions of percentile dose distributions exhibited excellent agreement with the ground truth dose distributions. The average gamma index with 2 mm/2%, consistently exceeded 97% for both 5th and 95th percentile dose volumes. Mean dose-volume histogram error analysis revealed that predictions from the combined training set yielded mean errors and standard deviations that were generally similar to those in the specific patient training data sets. Our proposed DL-based method for evaluation of the robustness of proton therapy plans provides precise, rapid predictions of percentile dose for a given confidence level regardless of the beam arrangement and cancer site. This versatility positions our model as a valuable tool for evaluating the robustness of proton therapy across various cancer sites.

Construction of University Football Basic Teaching and Training System Based on Object Detection and Tracking Algorithm

Football, often referred to as soccer outside the UK and Europe, is a highly popular activity on many university campuses. As a result, football training programs are offered across many institutions, training students at a range of skill levels. Therefore, this paper investigates a deep learning (DL)-based algorithm for recognizing and tracking targets in sports training videos. The study specifically addresses those videos tailored to identifying small targets, by generating new multi-scale features and modifying anchor point generation rules. Experimental results demonstrate the algorithm's strong performance in tracking football targets. Compared to Histogram of Oriented Gradients, the DL-based model achieved a 29.58% increase in accuracy and a 39.68% decrease in error rates when recognizing football player movement features. This algorithm accurately locates the edge contours of a football player's movements, meaning that universities can, and should, actively reform football teaching and training to enhance teaching effectiveness by utilizing this powerful algorithm.

Decentralized optimal voltage control for wind farm with deep learning-based data-driven modeling

The random fluctuations of wind energy and external grid voltage disturbances can both lead to serious voltage fluctuations and voltage deviations in the wind farm (WF). Voltage/reactive power control is an effective way to improve the voltage stability of WF. The existing research is based on complex physical models with prior parameter information, and the accuracy and calculation speed of the WF model are difficult to guarantee. To address this issue, this paper explores a decentralized optimal voltage control method for WF with deep learning-based (DL) data-driven modeling (DL-DOVC). A hybrid convolutional neural network (CNN)-Transformer architecture is proposed to establish the data-driven model of WF, leveraging its enhanced capabilities for extracting time series correlation, to learn complex patterns and dynamics from time series data. Then, we develop a fully decentralized voltage control method for WF to regulate the terminal bus voltage of wind turbines (WTs) within a feasible range. A DL-based data-driven predictive controller is specially designed to solve the established data-driven voltage optimization problem, it eliminates the necessity for frequent manual model maintenance and is suitable for real-time application. Additionally, the difficulty of WF decentralized optimal voltage control arises from the inability of local controllers to fully consider the impact of all non-local WTs on the local WT states. By designing the DL-based auxiliary state-feedback controller, the effects of non-local WTs are implicitly considered in the auxiliary feedback control law. A WF with 32*6.25 MW WTs is used to test the proposed DL-DOVC method. Simulation results show that the proposed DL-based model can efficiently learn and predict the WF dynamics under different operating scenarios. The near-global optimal voltage control performance is achieved only with local measurements by employing the DL-DOVC method.

ICRBP-LKHA: Large convolutional kernel and hybrid channel-spatial attention for identifying circRNA-RBP interaction sites.

Circular RNAs (circRNAs) play vital roles in transcription and translation. Identification of circRNA-RBP (RNA-binding protein) interaction sites has become a fundamental step in molecular and cell biology. Deep learning (DL)-based methods have been proposed to predict circRNA-RBP interaction sites and achieved impressive identification performance. However, those methods cannot effectively capture long-distance dependencies, and cannot effectively utilize the interaction information of multiple features. To overcome those limitations, we propose a DL-based model iCRBP-LKHA using deep hybrid networks for identifying circRNA-RBP interaction sites. iCRBP-LKHA adopts five encoding schemes. Meanwhile, the neural network architecture, which consists of large kernel convolutional neural network (LKCNN), convolutional block attention module with one-dimensional convolution (CBAM-1D) and bidirectional gating recurrent unit (BiGRU), can explore local information, global context information and multiple features interaction information automatically. To verify the effectiveness of iCRBP-LKHA, we compared its performance with shallow learning algorithms on 37 circRNAs datasets and 37 circRNAs stringent datasets. And we compared its performance with state-of-the-art DL-based methods on 37 circRNAs datasets, 37 circRNAs stringent datasets and 31 linear RNAs datasets. The experimental results not only show that iCRBP-LKHA outperforms other competing methods, but also demonstrate the potential of this model in identifying other RNA-RBP interaction sites.

Assessment of the deep learning-based gamma passing rate prediction system for 1.5 T magnetic resonance-guided linear accelerator.

Measurement-based verification is impossible for the patient-specific quality assurance (QA) of online adaptive magnetic resonance imaging-guided radiotherapy (oMRgRT) because the patient remains on the couch throughout the session. We assessed a deep learning (DL) system for oMRgRT to predict the gamma passing rate (GPR). This study collected 125 verification plans [reference plan (RP), 100; adapted plan (AP), 25] from patients with prostate cancer treated using Elekta Unity. Based on our previous study, we employed a convolutional neural network that predicted the GPRs of nine pairs of gamma criteria from 1%/1mm to 3%/3mm. First, we trained and tested the DL model using RPs (n=75 and n=25 for training and testing, respectively) for its optimization. Second, we tested the GPR prediction accuracy using APs to determine whether the DL model could be applied to APs. The mean absolute error (MAE) and correlation coefficient (r) of the RPs were 1.22 ± 0.27% and 0.29 ± 0.10 in 3%/2mm, 1.35 ± 0.16% and 0.37 ± 0.15 in 2%/2mm, and 3.62 ± 0.55% and 0.32 ± 0.14 in 1%/1mm, respectively. The MAE and r of the APs were 1.13 ± 0.33% and 0.35 ± 0.22 in 3%/2mm, 1.68 ± 0.47% and 0.30 ± 0.11 in 2%/2mm, and 5.08 ± 0.29% and 0.15 ± 0.10 in 1%/1mm, respectively. The time cost was within 3s for the prediction. The results suggest the DL-based model has the potential for rapid GPR prediction in Elekta Unity.

Development and Validation of a Deep Learning Model for Histopathological Slide Analysis in Lung Cancer Diagnosis.

Lung cancer is the leading cause of cancer-related deaths worldwide. Two of the crucial factors contributing to these fatalities are delayed diagnosis and suboptimal prognosis. The rapid advancement of deep learning (DL) approaches provides a significant opportunity for medical imaging techniques to play a pivotal role in the early detection of lung tumors and subsequent monitoring during treatment. This study presents a DL-based model for efficient lung cancer detection using whole-slide images. Our methodology combines convolutional neural networks (CNNs) and separable CNNs with residual blocks, thereby improving classification performance. Our model improves accuracy (96% to 98%) and robustness in distinguishing between cancerous and non-cancerous lung cell images in less than 10 s. Moreover, the model's overall performance surpassed that of active pathologists, with an accuracy of 100% vs. 79%. There was a significant linear correlation between pathologists' accuracy and years of experience (r Pearson = 0.71, 95% CI 0.14 to 0.93, p = 0.022). We conclude that this model enhances the accuracy of cancer detection and can be used to train junior pathologists.

A Fine-Tuned CatBoost-Based Speech Disorder Detection Model

The classification of speech disorders (SDs) is crucial for treating children with speech impairment (SI). An automated SD classification can assist speech therapists in rendering services to children with SI in rural areas. Automated techniques for detecting SDs provide objective assessments of speech attributes, including articulation, fluency, and prosody. Clinical examinations and quantitative assessments provide an in-depth understanding of the patient’s speaking abilities and limitations. Existing deep learning (DL) models for SD detection often lack generalization across diverse populations and speech variations, leading to suboptimal performance when applied to individuals with different linguistic backgrounds or dialects. This study introduces a DL-based model for classifying normal and abnormal speeches using voice samples. To overcome the overfitting and bias, the authors construct convolutional neural network models with the weights of MobileNet V3 and EfficientNet B7 models for feature extraction (FE). To improve performance, they integrate the squeeze and excitation block with the MobileNet V3-based FE model. Similarly, the EfficientNet B7-model-based FE is improved using the structure pruning technique. The enhanced CatBoost model differentiates the normal and abnormal speeches using the extracted features. The experimental analysis is performed using the public dataset that contains 4620 utterances of healthy children and 2178 utterances of children with SI. The comparative study reveals the exceptional performance of the proposed SD classification model. The model outperforms the current SD classification models. It can be employed in clinical settings to support speech therapists. Substantial training with diverse voice samples can improve the generalizability of the proposed model.

Deep Learning based UPoS Tagger for Assamese Religious Text

Religious texts are known to be with specific patterns of writing, and also involve specific vocabularies. These are also known to follow specific style of writing. Thereby these texts are enriched with typical semantic and syntactic characteristics, demanding special attention for Natural Language Processing (NLP) tasks. This research paper focuses on the application of Deep Learning (DL) techniques for Parts of Speech (PoS) tagging focusing on Assamese language religious texts. We have created a specialized dataset comprising approximately 11,000 sentences extracted from various sources including web crawling and filtering religious texts from existing corpora. The dataset was manually validated by linguists to ensure accuracy, errors, and corrections required. A performance matrix was constructed to analyze the performance of the initial tagging using a pre-existing DL-based model trained for Assamese Universal Parts of Speech (UPoS) tagger. Following this, we utilized a subset of the dataset for manual evaluation, and the validated dataset is then considered as a gold standard training dataset for training other DL models using GRU, RNN and Bidirectional LSTM (BiLSTM) architectures. Training accuracies were recorded and presented, demonstrating the effectiveness of the proposed approach. Accuracies, Precision, and Recall were recorded for all the three Models. F1 scores also have been calculated. Comparison of training and testing accuracies are depicted with performance graphs.

Automated prostate gland segmentation in challenging clinical cases: comparison of three artificial intelligence methods.

Automated methods for prostate segmentation on MRI are typically developed under ideal scanning and anatomical conditions. This study evaluates three different prostate segmentation AI algorithms in a challenging population of patients with prior treatments, variable anatomic characteristics, complex clinical history, or atypical MRI acquisition parameters. A single institution retrospective database was queried for the following conditions at prostate MRI: prior prostate-specific oncologic treatment, transurethral resection of the prostate (TURP), abdominal perineal resection (APR), hip prosthesis (HP), diversity of prostate volumes (large ≥ 150cc, small ≤ 25cc), whole gland tumor burden, magnet strength, noted poor quality, and various scanners (outside/vendors). Final inclusion criteria required availability of axial T2-weighted (T2W) sequence and corresponding prostate organ segmentation from an expert radiologist. Three previously developed algorithms were evaluated: (1) deep learning (DL)-based model, (2) commercially available shape-based model, and (3) federated DL-based model. Dice Similarity Coefficient (DSC) was calculated compared to expert. DSC by model and scan factors were evaluated with Wilcox signed-rank test and linear mixed effects (LMER) model. 683 scans (651 patients) met inclusion criteria (mean prostate volume 60.1cc [9.05-329cc]). Overall DSC scores for models 1, 2, and 3 were 0.916 (0.707-0.971), 0.873 (0-0.997), and 0.894 (0.025-0.961), respectively, with DL-based models demonstrating significantly higher performance (p < 0.01). In sub-group analysis by factors, Model 1 outperformed Model 2 (all p < 0.05) and Model 3 (all p < 0.001). Performance of all models was negatively impacted by prostate volume and poor signal quality (p < 0.01). Shape-based factors influenced DL models (p < 0.001) while signal factors influenced all (p < 0.001). Factors affecting anatomical and signal conditions of the prostate gland can adversely impact both DL and non-deep learning-based segmentation models.

A new model using deep learning to predict recurrence after surgical resection of lung adenocarcinoma

This study aimed to develop a deep learning (DL) model for predicting the recurrence risk of lung adenocarcinoma (LUAD) based on its histopathological features. Clinicopathological data and whole slide images from 164 LUAD cases were collected and used to train DL models with an ImageNet pre-trained efficientnet-b2 architecture, densenet201, and resnet152. The models were trained to classify each image patch into high-risk or low-risk groups, and the case-level result was determined by multiple instance learning with final FC layer’s features from a model from all patches. Analysis of the clinicopathological and genetic characteristics of the model-based risk group was performed. For predicting recurrence, the model had an area under the curve score of 0.763 with 0.750, 0.633 and 0.680 of sensitivity, specificity, and accuracy in the test set, respectively. High-risk cases for recurrence predicted by the model (HR group) were significantly associated with shorter recurrence-free survival and a higher stage (both, p < 0.001). The HR group was associated with specific histopathological features such as poorly differentiated components, complex glandular pattern components, tumor spread through air spaces, and a higher grade. In the HR group, pleural invasion, necrosis, and lymphatic invasion were more frequent, and the size of the invasion was larger (all, p < 0.001). Several genetic mutations, including TP53 (p = 0.007) mutations, were more frequently found in the HR group. The results of stages I-II were similar to those of the general cohort. DL-based model can predict the recurrence risk of LUAD and identify the presence of the TP53 gene mutation by analyzing histopathologic features.

A novel adaptive deep transfer learning method towards thermal error modeling of electric spindles under variable conditions

Thermal error modeling (TEM) plays a vital role in maintaining the machining accuracy of electric spindles. Recently, deep learning (DL) techniques have obtained promising achievements in this area. However, DL techniques have certain limitations. The data acquired from variable working conditions present large distribution discrepancies, the DL-based model established on one working condition fails to obtain satisfactory prediction accuracy in another working condition. Moreover, existing studies focus solely on using temperature features to build prediction models, neglecting the full use of multi-sensory information. To address these issues, this paper proposes a novel adaptive deep transfer learning method towards TEM of electric spindles, which takes full advantage of the temperature, current, and power sensory information. Firstly, finite element simulation is employed to analyze the thermal characteristics of the electric spindle and determine the locations for temperature measurement points. Then, the convolutional long short-term memory network (C-LSTMN) is constructed, where the spatial features from multi-sensory information between inputs and prediction patterns are captured by convolutional layers, these features are further processed by long short-term memory network (LSTMN) to extract temporal features. Subsequently, the multi-kernel joint maximum mean discrepancy (MK-JMMD) measure is developed to minimize the distribution discrepancies between the source and target domains, thus the prediction model initially established on the source domain can be adaptive to effectively predict on the target domain. Finally, with unavailable thermal error data in the target domain, the proposed method is validated through 12 transfer tasks using datasets from four working conditions with other comparison methods. The results demonstrate that the proposed method overcomes the challenges of unavailable labeled thermal error samples and outperforms advanced methods.

Development and validation of a deep learning model for predicting postoperative survival of patients with gastric cancer

BackgroundDeep learning (DL), a specialized form of machine learning (ML), is valuable for forecasting survival in various diseases. Its clinical applicability in real-world patients with gastric cancer (GC) has yet to be extensively validated.MethodsA combined cohort of 11,414 GC patients from the Surveillance, Epidemiology and End Results (SEER) database and 2,846 patients from a Chinese dataset were utilized. The internal validation of different algorithms, including DL model, traditional ML models, and American Joint Committee on Cancer (AJCC) stage model, was conducted by training and testing sets on the SEER database, followed by external validation on the Chinese dataset. The performance of the algorithms was assessed using the area under the receiver operating characteristic curve, decision curve, and calibration curve.ResultsDL model demonstrated superior performance in terms of the area under the curve (AUC) at 1, 3, and, 5 years post-surgery across both datasets, surpassing other ML models and AJCC stage model, with AUCs of 0.77, 0.80, and 0.82 in the SEER dataset and 0.77, 0.76, and 0.75 in the Chinese dataset, respectively. Furthermore, decision curve analysis revealed that the DL model yielded greater net gains at 3 years than other ML models and AJCC stage model, and calibration plots at 3 years indicated a favorable level of consistency between the ML and actual observations during external validation.ConclusionsDL-based model was established to accurately predict the survival rate of postoperative patients with GC.

Automated measurement and grading of knee cartilage thickness: a deep learning-based approach.

Knee cartilage is the most crucial structure in the knee, and the reduction of cartilage thickness is a significant factor in the occurrence and development of osteoarthritis. Measuring cartilage thickness allows for a more accurate assessment of cartilage wear, but this process is relatively time-consuming. Our objectives encompass using various DL methods to segment knee cartilage from MRIs taken with different equipment and parameters, building a DL-based model for measuring and grading knee cartilage, and establishing a standardized database of knee cartilage thickness. In this retrospective study, we selected a mixed knee MRI dataset consisting of 700 cases from four datasets with varying cartilage thickness. We employed four convolutional neural networks-UNet, UNet++, ResUNet, and TransUNet-to train and segment the mixed dataset, leveraging an extensive array of labeled data for effective supervised learning. Subsequently, we measured and graded the thickness of knee cartilage in 12 regions. Finally, a standard knee cartilage thickness dataset was established using 291 cases with ages ranging from 20 to 45 years and a Kellgren-Lawrence grading of 0. The validation results of network segmentation showed that TransUNet performed the best in the mixed dataset, with an overall dice similarity coefficient of 0.813 and an Intersection over Union of 0.692. The model's mean absolute percentage error for automatic measurement and grading after segmentation was 0.831. The experiment also yielded standard knee cartilage thickness, with an average thickness of 1.98 mm for the femoral cartilage and 2.14 mm for the tibial cartilage. By selecting the best knee cartilage segmentation network, we built a model with a stronger generalization ability to automatically segment, measure, and grade cartilage thickness. This model can assist surgeons in more accurately and efficiently diagnosing changes in patients' cartilage thickness.

Unleashing the potential of fNIRS with machine learning: classification of fine anatomical movements to empower future brain-computer interface.

In this study, we explore the potential of using functional near-infrared spectroscopy (fNIRS) signals in conjunction with modern machine-learning techniques to classify specific anatomical movements to increase the number of control commands for a possible fNIRS-based brain-computer interface (BCI) applications. The study focuses on novel individual finger-tapping, a well-known task in fNIRS and fMRI studies, but limited to left/right or few fingers. Twenty-four right-handed participants performed the individual finger-tapping task. Data were recorded by using sixteen sources and detectors placed over the motor cortex according to the 10-10 international system. The event's average oxygenated Δ HbO and deoxygenated Δ HbR hemoglobin data were utilized as features to assess the performance of diverse machine learning (ML) models in a challenging multi-class classification setting. These methods include LDA, QDA, MNLR, XGBoost, and RF. A new DL-based model named "Hemo-Net" has been proposed which consists of multiple parallel convolution layers with different filters to extract the features. This paper aims to explore the efficacy of using fNRIS along with ML/DL methods in a multi-class classification task. Complex models like RF, XGBoost, and Hemo-Net produce relatively higher test set accuracy when compared to LDA, MNLR, and QDA. Hemo-Net has depicted a superior performance achieving the highest test set accuracy of 76%, however, in this work, we do not aim at improving the accuracies of models rather we are interested in exploring if fNIRS has the neural signatures to help modern ML/DL methods in multi-class classification which can lead to applications like brain-computer interfaces. Multi-class classification of fine anatomical movements, such as individual finger movements, is difficult to classify with fNIRS data. Traditional ML models like MNLR and LDA show inferior performance compared to the ensemble-based methods of RF and XGBoost. DL-based method Hemo-Net outperforms all methods evaluated in this study and demonstrates a promising future for fNIRS-based BCI applications.

Differential privacy preserved federated learning for prognostic modeling in COVID-19 patients using large multi-institutional chest CT dataset.

Notwithstanding the encouraging results of previous studies reporting on the efficiency of deep learning (DL) in COVID-19 prognostication, clinical adoption of the developed methodology still needs to be improved. To overcome this limitation, we set out to predict the prognosis of a large multi-institutional cohort of patients with COVID-19 using a DL-based model. This study aimed to evaluate the performance of deep privacy-preserving federated learning (DPFL) in predicting COVID-19 outcomes using chest CT images. After applying inclusion and exclusion criteria, 3055 patients from 19 centers, including 1599 alive and 1456 deceased, were enrolled in this study. Data from all centers were split (randomly with stratification respective to each center and class) into a training/validation set (70%/10%) and a hold-out test set (20%). For the DL model, feature extraction was performed on 2D slices, and averaging was performed at the final layer to construct a 3D model for each scan. The DensNet model was used for feature extraction. The model was developed using centralized and FL approaches. For FL, we employed DPFL approaches. Membership inference attack was also evaluated in the FL strategy. For model evaluation, different metrics were reported in the hold-out test sets. In addition, models trained in two scenarios, centralized and FL, were compared using the DeLong test for statistical differences. The centralized model achieved an accuracy of 0.76, while the DPFL model had an accuracy of 0.75. Both the centralized and DPFL models achieved a specificity of 0.77. The centralized model achieved a sensitivity of 0.74, while the DPFL model had a sensitivity of 0.73. A mean AUC of 0.82 and 0.81 with 95% confidence intervals of (95% CI: 0.79-0.85) and (95% CI: 0.77-0.84) were achieved by the centralized model and the DPFL model, respectively. The DeLong test did not prove statistically significant differences between the two models (p-value=0.98). The AUC values for the inference attacks fluctuate between 0.49 and 0.51, with an average of 0.50±0.003 and 95% CI for the mean AUC of 0.500 to 0.501. The performance of the proposed model was comparable to centralized models while operating on large and heterogeneous multi-institutional datasets. In addition, the model was resistant to inference attacks, ensuring the privacy of shared data during the training process.

Segmentation and classification of lungs CT-scan for detecting COVID-19 abnormalities by deep learning technique: U-Net model.

Artificial intelligence (AI) techniques have been ascertained useful in the analysis and description of infectious areas in radiological images promptly. Our aim in this study was to design a web-based application for detecting and labeling infected tissues on CT (computed tomography) lung images of patients based on the deep learning (DL) method as a type of AI. The U-Net architecture, one of the DL networks, is used as a hybrid model with pre-trained densely connected convolutional network 121 (DenseNet121) architecture for the segmentation process. The proposed model was constructed on 1031 persons' CT-scan images from Ibn Sina Hospital of Iran in 2021 and some publicly available datasets. The network was trained using 6000 slices, validated on 1000 slices images, and tested against the 150 slices. Accuracy, sensitivity, specificity, and area under the receiver operating characteristics (ROC) curve (AUC) were calculated to evaluate model performance. The results indicate the acceptable ability of the U-Net-DenseNet121 model in detecting COVID-19 abnormality (accuracy = 0.88 and AUC = 0.96 for thresholds of 0.13 and accuracy = 0.88 and AUC = 0.90 for thresholds of 0.2). Based on this model, we developed the "Imaging-Tech" web-based application for use at hospitals and clinics to make our project's output more practical and attractive in the market. We designed a DL-based model for the segmentation of COVID-19 CT scan images and, based on this model, constructed a web-based application that, according to the results, is a reliable detector for infected tissue in lung CT-scans. The availability of such tools would aid in automating, prioritizing, fastening, and broadening the treatment of COVID-19 patients globally.

Hybrid Deep Learning-Based Model for Intrusion Detection

There is an intensive need for intrusion detection systems (IDSs) due to incremental and frequent cyber-attacks. The first line of defense against online threats is an IDS. Researchers are using deep learning (DL) approaches to detect attackers and preserve user information. In this study, we introduce a hybrid DL-based model. The proposed model integrates LSTM and ResNet to eliminate the vanishing gradient problem and increase the accuracy of the classification model. The proposed model aims to classify between normal or an attack, with each attack either being a DoS, U2R, R2L, or a probe over the NSL-KDD dataset. The proposed model achieves 99.5% according to accuracy. The model was compared with other ML and DL models.