Dconformer: A denoising convolutional transformer with joint learning strategy for intelligent diagnosis of bearing faults

Mechanical Systems and Signal Processing, Год журнала: 2024, Номер 210, С. 111142 - 111142

Опубликована: Янв. 23, 2024

Язык: Английский

---

ViTO: Vision Transformer-Operator

Computer Methods in Applied Mechanics and Engineering, Год журнала: 2024, Номер 428, С. 117109 - 117109

Опубликована: Июнь 11, 2024

Язык: Английский

---

Comparison of Vision Transformers and Convolutional Neural Networks in Medical Image Analysis: A Systematic Review

Journal of Medical Systems, Год журнала: 2024, Номер 48(1)

Опубликована: Сен. 12, 2024

Язык: Английский

---

Deep learning for crack detection on masonry façades using limited data and transfer learning

Journal of Building Engineering, Год журнала: 2023, Номер 76, С. 107105 - 107105

Опубликована: Июнь 25, 2023

Crack detection in masonry façades is a crucial task for ensuring the safety and longevity of buildings. However, traditional methods are often time-consuming, expensive, labour-intensive. In recent years, deep learning techniques have been applied to detect cracks images, but these models require large amounts annotated data achieve high accuracy, which can be difficult obtain. this article, we propose approach crack on brickwork using transfer with limited data. Our uses pre-trained convolutional neural network model as feature extractor, then optimised specifically detection. To evaluate effectiveness our proposed method, created curated dataset 700 façade used 500 images training, 100 validation, remaining testing. Results showed that very effective detecting cracks, achieving an accuracy F1-score up 100% when following end-to-end training network, thus being promising solution building inspection maintenance, particularly situations where limited. Moreover, easily adapted different types façades, making it versatile tool maintenance.

Язык: Английский

---

ViTs as backbones: Leveraging vision transformers for feature extraction

Information Fusion, Год журнала: 2025, Номер unknown, С. 102951 - 102951

Опубликована: Янв. 1, 2025

Язык: Английский

---

Enhancing furcation involvement classification on panoramic radiographs with vision transformers

BMC Oral Health, Год журнала: 2025, Номер 25(1)

Опубликована: Янв. 29, 2025

The severity of furcation involvement (FI) directly affected tooth prognosis and influenced treatment approaches. However, assessing, diagnosing, treating molars with FI was complicated by anatomical morphological variations. Cone-beam computed tomography (CBCT) enhanced diagnostic accuracy for detecting measuring defects. Despite its advantages, the high cost radiation dose associated CBCT equipment limited widespread use. aim this study to evaluate performance Vision Transformer (ViT) in comparison several commonly used traditional deep learning (DL) models classifying or without on panoramic radiographs. A total 1,568 images obtained from 506 radiographs were construct database models. This developed assessed a ViT model radiographs, compared models, including Multi-Layer Perceptron (MLP), Visual Geometry Group (VGG)Net, GoogLeNet. Among evaluated outperformed all others, achieving highest precision (0.98), recall (0.92), F1 score (0.95), along lowest cross-entropy loss (0.27) (92%). also recorded area under curve (AUC) (98%), outperforming other statistically significant differences (p < 0.05), confirming classification capability. gradient-weighted class activation mapping (Grad-CAM) analysis revealed key areas that focused during predictions. DL algorithms can automatically classify using readily accessible images. These findings demonstrate outperforms tested highlighting potential transformer-based approaches significantly advance image classification. approach is expected reduce both financial burden patients while simultaneously improving precision.

Язык: Английский

---

Novel Approach for Osteoporosis Classification Using X-ray Images

Biomedical & Pharmacology Journal, Год журнала: 2025, Номер 18(December Spl Edition), С. 203 - 216

Опубликована: Янв. 20, 2025

This research delves into the technical advancements of image segmentation and classification models, specifically refined Pix2Pix Vision Transformer (ViT) architectures, for crucial task osteoporosis detection using X-ray images. The improved model demonstrates noteworthy strides in segmentation, achieving a specificity 97.24% excelling reduction false positives. Simultaneously, modified ViT especially MViT-B/16 variant, exhibit superior accuracy at 96.01% classifying cases, showcasing their proficiency identifying critical medical conditions. These models are poised to revolutionize diagnosis, providing clinicians with accurate tools early intervention. synergies between open avenues nuanced approaches automated diagnostic systems, potential significantly improve clinical results contribute broader landscape analysis. As remains prevalent often undiagnosed condition, insights from this study hold substantial importance advancing field, emphasizing role improving patient care health outcomes.

Язык: Английский

---

A COVID-19 medical image classification algorithm based on Transformer

Scientific Reports, Год журнала: 2023, Номер 13(1)

Опубликована: Апрель 1, 2023

Abstract Coronavirus 2019 (COVID-19) is a new acute respiratory disease that has spread rapidly throughout the world. This paper proposes novel deep learning network based on ResNet-50 merged transformer named RMT-Net. On backbone of ResNet-50, it uses Transformer to capture long-distance feature information, adopts convolutional neural networks and depth-wise convolution obtain local features, reduce computational cost acceleration detection process. The RMT-Net includes four stage blocks realize extraction different receptive fields. In first three stages, global self-attention method adopted important information construct relationship between tokens. fourth stage, residual are used extract details feature. Finally, average pooling layer fully connected perform classification tasks. Training, verification testing carried out self-built datasets. model compared with VGGNet-16, i-CapsNet MGMADS-3. experimental results show Test_ acc 97.65% X-ray image dataset, 99.12% CT which both higher than other models. size only 38.5 M, speed 5.46 ms 4.12 per image, respectively. It proved can detect classify COVID-19 accuracy efficiency.

Язык: Английский

---

Deep learning-based assessment of knee septic arthritis using transformer features in sonographic modalities

Computer Methods and Programs in Biomedicine, Год журнала: 2023, Номер 237, С. 107575 - 107575

Опубликована: Май 3, 2023

Язык: Английский

---

A Multichannel CT and Radiomics-Guided CNN-ViT (RadCT-CNNViT) Ensemble Network for Diagnosis of Pulmonary Sarcoidosis

Diagnostics, Год журнала: 2024, Номер 14(10), С. 1049 - 1049

Опубликована: Май 18, 2024

Pulmonary sarcoidosis is a multisystem granulomatous interstitial lung disease (ILD) with variable presentation and prognosis. The early accurate detection of pulmonary may prevent progression to fibrosis, serious potentially life-threatening form the disease. However, lack gold-standard diagnostic test specific radiographic findings poses challenges in diagnosing sarcoidosis. Chest computed tomography (CT) imaging commonly used but requires expert, chest-trained radiologists differentiate from malignancies, infections, other ILDs. In this work, we develop multichannel, CT radiomics-guided ensemble network (RadCT-CNNViT) visual explainability for vs. cancer (LCa) classification using chest images. We leverage hand-crafted radiomics features as input channels, 3D convolutional neural (CNN) vision transformer (ViT) feature extraction fusion before head. CNN sub-network captures localized spatial information lesions, while ViT long-range, global dependencies between features. Through multichannel fusion, our model achieves highest performance accuracy, sensitivity, specificity, precision, F1-score, combined AUC 0.93 ± 0.04, 0.94 0.08, 0.95 0.05, 0.97, respectively, five-fold cross-validation study (n = 126) LCa 93) cases. A detailed ablation showing impact + compared or alone, input, also presented work. Overall, AI developed work offers promising potential triaging patients timely diagnosis treatment CT.

Язык: Английский

---