Lung Cancer Classification using Optimized Attention-based Convolutional Neural Network with DenseNet-201 Transfer Learning Model on CT image

Biomedical Signal Processing and Control, Journal Year: 2024, Volume and Issue: 95, P. 106330 - 106330

Published: April 25, 2024

Language: Английский

---

Multi-modal deep learning methods for classification of chest diseases using different medical imaging and cough sounds

PLoS ONE, Journal Year: 2024, Volume and Issue: 19(3), P. e0296352 - e0296352

Published: March 12, 2024

Chest disease refers to a wide range of conditions affecting the lungs, such as COVID-19, lung cancer (LC), consolidation (COL), and many more. When diagnosing chest disorders medical professionals may be thrown off by overlapping symptoms (such fever, cough, sore throat, etc.). Additionally, researchers make use X-rays (CXR), cough sounds, computed tomography (CT) scans diagnose disorders. The present study aims classify nine different disorders, including LC, COL, atelectasis (ATE), tuberculosis (TB), pneumothorax (PNEUTH), edema (EDE), pneumonia (PNEU). Thus, we suggested four novel convolutional neural network (CNN) models that train distinct image-level representations for classifications extracting features from images. Furthermore, proposed CNN employed several new approaches max-pooling layer, batch normalization layers (BANL), dropout, rank-based average pooling (RBAP), multiple-way data generation (MWDG). scalogram method is utilized transform sounds coughing into visual representation. Before beginning model has been developed, SMOTE approach used calibrate CXR CT well sound images (CSI) CXR, scan, CSI training evaluating come 24 publicly available benchmark illness datasets. classification performance compared with seven baseline models, namely Vgg-19, ResNet-101, ResNet-50, DenseNet-121, EfficientNetB0, DenseNet-201, Inception-V3, in addition state-of-the-art (SOTA) classifiers. effectiveness further demonstrated results ablation experiments. was successful achieving an accuracy 99.01%, making it superior both SOTA As result, capable offering significant support radiologists other professionals.

Language: Английский

---

Federated learning with deep convolutional neural networks for the detection of multiple chest diseases using chest x-rays

Multimedia Tools and Applications, Journal Year: 2024, Volume and Issue: 83(23), P. 63017 - 63045

Published: Jan. 10, 2024

Language: Английский

---

NSEC-YOLO: Real-time lesion detection on chest X-ray with adaptive noise suppression and global perception aggregation

Journal of Radiation Research and Applied Sciences, Journal Year: 2025, Volume and Issue: 18(1), P. 101281 - 101281

Published: Jan. 7, 2025

Language: Английский

---

Diagnostic Performance of Artificial Intelligence–Based Methods for Tuberculosis Detection: Systematic Review

Journal of Medical Internet Research, Journal Year: 2025, Volume and Issue: 27, P. e69068 - e69068

Published: March 7, 2025

Tuberculosis (TB) remains a significant health concern, contributing to the highest mortality among infectious diseases worldwide. However, none of various TB diagnostic tools introduced is deemed sufficient on its own for pathway, so artificial intelligence (AI)-based methods have been developed address this issue. We aimed provide comprehensive evaluation AI-based algorithms detection across data modalities. Following PRISMA (Preferred Reporting Items Systematic Reviews and Meta-Analysis) 2020 guidelines, we conducted systematic review synthesize current knowledge topic. Our search 3 major databases (Scopus, PubMed, Association Computing Machinery [ACM] Digital Library) yielded 1146 records, which included 152 (13.3%) studies in our analysis. QUADAS-2 (Quality Assessment Diagnostic Accuracy Studies version 2) was performed risk-of-bias assessment all studies. Radiographic biomarkers (n=129, 84.9%) deep learning (DL; n=122, 80.3%) approaches were predominantly used, with convolutional neural networks (CNNs) using Visual Geometry Group (VGG)-16 (n=37, 24.3%), ResNet-50 (n=33, 21.7%), DenseNet-121 (n=19, 12.5%) architectures being most common DL approach. The majority focused model development (n=143, 94.1%) used single modality approach (n=141, 92.8%). AI demonstrated good performance studies: mean accuracy=91.93% (SD 8.10%, 95% CI 90.52%-93.33%; median 93.59%, IQR 88.33%-98.32%), area under curve (AUC)=93.48% 7.51%, 91.90%-95.06%; 95.28%, 91%-99%), sensitivity=92.77% 7.48%, 91.38%-94.15%; 94.05% 89%-98.87%), specificity=92.39% 9.4%, 90.30%-94.49%; 95.38%, 89.42%-99.19%). different biomarker types showed accuracies 92.45% 7.83%), 89.03% 8.49%), 84.21% 0%); AUCs 94.47% 7.32%), 88.45% 8.33%), 88.61% 5.9%); sensitivities 93.8% 6.27%), 88.41% 10.24%), 93% specificities 94.2% 6.63%), 85.89% 14.66%), 0%) radiographic, molecular/biochemical, physiological types, respectively. reference standards 91.44% 7.3%), 93.16% 6.44%), 88.98% 9.77%); 90.95% 7.58%), 94.89% 5.18%), 92.61% 6.01%); 91.76% 7.02%), 93.73% 6.67%), 91.34% 7.71%); 86.56% 12.8%), 93.69% 8.45%), 92.7% 6.54%) bacteriological, human reader, combined standards, transfer (TL) increasing popularity (n=89, 58.6%). Notably, only 1 (0.7%) study domain-shift analysis detection. Findings from underscore considerable promise realm Future research endeavors should prioritize conducting analyses better simulate real-world scenarios PROSPERO CRD42023453611; https://www.crd.york.ac.uk/PROSPERO/view/CRD42023453611.

Language: Английский

---

Enhanced tuberculosis detection using Vision Transformers and explainable AI with a Grad-CAM approach on chest X-rays

BMC Medical Imaging, Journal Year: 2025, Volume and Issue: 25(1)

Published: March 24, 2025

Tuberculosis (TB), caused by Mycobacterium tuberculosis, remains a leading global health challenge, especially in low-resource settings. Accurate diagnosis from chest X-rays is critical yet challenging due to subtle manifestations of TB, particularly its early stages. Traditional computational methods, primarily using basic convolutional neural networks (CNNs), often require extensive pre-processing and struggle with generalizability across diverse clinical environments. This study introduces novel Vision Transformer (ViT) model augmented Gradient-weighted Class Activation Mapping (Grad-CAM) enhance both diagnostic accuracy interpretability. The ViT utilizes self-attention mechanisms extract long-range dependencies complex patterns directly the raw pixel information, whereas Grad-CAM offers visual explanations decisions about highlighting significant regions X-rays. contains Conv2D stem for initial feature extraction, followed many transformer encoder blocks, thereby significantly boosting ability learn discriminative features without any pre-processing. Performance testing on validation set had an 0.97, recall 0.99, F1-score 0.98 TB patients. On test set, has 0.98, which better than existing methods. addition visuals not only improves transparency but also assists radiologists assessing verifying AI-driven diagnoses. These results demonstrate model's higher precision potential application real-world settings, providing massive improvement automated detection TB.

Language: Английский

---

Patch-based U-NET model and MSqueezeNet-PyramidNet for efficient segmentation and classification of tuberculosis, pneumonia, and COVID-19

Multimedia Tools and Applications, Journal Year: 2025, Volume and Issue: unknown

Published: April 8, 2025

Language: Английский

---

Tackling the COVID-19 Detection Problem Using Pre-trained Models

Lecture notes in networks and systems, Journal Year: 2025, Volume and Issue: unknown, P. 225 - 232

Published: Jan. 1, 2025

Language: Английский

---

Detection and Classification of Pneumonia in Chest X-ray Images using Deep Learning Techniques

Published: April 12, 2024

Pneumonia could be a respiratory disease caused by tiny living beings or contagions; it influences wide extend of individuals, especially in creating nations where tall levels defilement, clean conditions, and pressing are sensibly common, expansion to unassuming helpful structures. causes pleural spread, which the lungs fill with fluid, coming about breathing issues. Early distinguishing proof pneumonia is basic for viable treatment expanded odds survival. Chest X-ray imaging foremost commonly utilized strategy diagnosing pneumonia. In any case, looking at chest X-rays can troublesome errand that's subject individual changeability. this ponder, we made computer-supported hypothesis framework modified area utilizing pictures. We advanced compatibility education oversee dissatisfaction open data outlined convolutional neural network model four trade instruction sorts CovXNet, RNN, VGG16. Whereas, within being styles, ResNet 50 move forward those that don't have correct delicacy. Hence, system diverse methods proposed. The proposed was tried on dataset accessible people.

Language: Английский

---

Deep Learning-Based Classification of Chest Diseases Using X-rays, CT Scans, and Cough Sound Images

Diagnostics, Journal Year: 2023, Volume and Issue: 13(17), P. 2772 - 2772

Published: Aug. 26, 2023

Chest disease refers to a variety of lung disorders, including cancer (LC), COVID-19, pneumonia (PNEU), tuberculosis (TB), and numerous other respiratory disorders. The symptoms (i.e., fever, cough, sore throat, etc.) these chest diseases are similar, which might mislead radiologists health experts when classifying diseases. X-rays (CXR), cough sounds, computed tomography (CT) scans utilized by researchers doctors identify such as LC, PNEU, TB. objective the work is nine different types diseases, edema (EDE), pneumothorax (PNEUTH), normal, atelectasis (ATE), consolidation (COL). Therefore, we designed novel deep learning (DL)-based detection network (DCDD_Net) that uses CXR, CT scans, sound images for identification scalogram method used convert sounds into an image. Before training proposed DCDD_Net model, borderline (BL) SMOTE applied balance model trained evaluated on 20 publicly available benchmark datasets scan, images. classification performance compared with four baseline models, i.e., InceptionResNet-V2, EfficientNet-B0, DenseNet-201, Xception, well state-of-the-art (SOTA) classifiers. achieved accuracy 96.67%, precision 96.82%, recall 95.76%, F1-score 95.61%, area under curve (AUC) 99.43%. results reveal outperformed models in terms many evaluation metrics. Thus, can provide significant assistance medical experts. Additionally, was also shown be resilient statistical evaluations using McNemar ANOVA tests.

Language: Английский

---