Deep learning methods to generate synthetic CT from MRI in radiotherapy: A literature review

Physica Medica, Journal Year: 2021, Volume and Issue: 89, P. 265 - 281

Published: Aug. 30, 2021

Language: Английский

---

Label-Free Segmentation of COVID-19 Lesions in Lung CT

IEEE Transactions on Medical Imaging, Journal Year: 2021, Volume and Issue: 40(10), P. 2808 - 2819

Published: March 24, 2021

Scarcity of annotated images hampers the building automated solution for reliable COVID-19 diagnosis and evaluation from CT. To alleviate burden data annotation, we herein present a label-free approach segmenting lesions in CT via voxel-level anomaly modeling that mines out relevant knowledge normal lung scans. Our is inspired by observation parts tracheae vessels, which lay high-intensity range where belong to, exhibit strong patterns. facilitate learning such patterns at voxel level, synthesize 'lesions' using set simple operations insert synthesized into scans to form training pairs, learn normalcy-recognizing network (NormNet) recognizes tissues separate them possible lesions. experiments on three different public datasets validate effectiveness NormNet, conspicuously outperforms variety unsupervised detection (UAD) methods.

Language: Английский

---

Federated Neural Architecture Search for Medical Data Security

IEEE Transactions on Industrial Informatics, Journal Year: 2022, Volume and Issue: 18(8), P. 5628 - 5636

Published: Jan. 19, 2022

Medical data widely exist in the hospital and personal life, usually across institutions regions. They have essential diagnostic value therapeutic significance. The disclosure of patient information causes people's panic, therefore, medical security solution is very crucial for intelligent health care. emergence federated learning (FL) provides an effective solution, which only transmits model parameters, breaking through bottleneck sharing, protecting security, avoiding economic losses. Meanwhile, neural architecture search (NAS) has become a popular method to automatically optimal solving complex practical problems. However, few papers combined FL NAS simultaneous privacy protection selection. Convolutional network (CNN) outstanding performance image recognition field. Combining CNN fuzzy rough sets can effectively improve interpretability deep networks. This article aims develop multiobjective convolutional interval type-2 based on (CIT2FR-FL-NAS) with improved evolutionary algorithm. We test proposed framework LC25000 lung colon histopathological dataset. Experimental verification demonstrates that designed CIT2FR-FL-NAS achieve high accuracy superior state-of-the-art models reduce complexity under condition security.

Language: Английский

---

COVID-Transformer: Interpretable COVID-19 Detection Using Vision Transformer for Healthcare

International Journal of Environmental Research and Public Health, Journal Year: 2021, Volume and Issue: 18(21), P. 11086 - 11086

Published: Oct. 21, 2021

In the recent pandemic, accurate and rapid testing of patients remained a critical task in diagnosis control COVID-19 disease spread healthcare industry. Because sudden increase cases, most countries have faced scarcity low rate testing. Chest X-rays been shown literature to be potential source for patients, but manually checking X-ray reports is time-consuming error-prone. Considering these limitations advancements data science, we proposed Vision Transformer-based deep learning pipeline detection from chest X-ray-based imaging. Due lack large sets, collected three open-source sets images aggregated them form 30 K image set, which largest publicly available collection this domain our knowledge. Our transformer model effectively differentiates normal with an accuracy 98% along AUC score 99% binary classification task. It distinguishes COVID-19, normal, pneumonia patient’s 92% Multi-class For evaluation on fine-tuned some widely used models literature, namely, EfficientNetB0, InceptionV3, Resnet50, MobileNetV3, Xception, DenseNet-121, as baselines. outperformed terms all metrics. addition, Grad-CAM based visualization created makes approach interpretable by radiologists can monitor progression affected lungs, assisting healthcare.

Language: Английский

---

Robust and data-efficient generalization of self-supervised machine learning for diagnostic imaging

Nature Biomedical Engineering, Journal Year: 2023, Volume and Issue: 7(6), P. 756 - 779

Published: June 8, 2023

Language: Английский

---

HiFuse: Hierarchical multi-scale feature fusion network for medical image classification

Biomedical Signal Processing and Control, Journal Year: 2023, Volume and Issue: 87, P. 105534 - 105534

Published: Sept. 30, 2023

Language: Английский

---

Applications of convolutional neural networks for intelligent waste identification and recycling: A review

Resources Conservation and Recycling, Journal Year: 2022, Volume and Issue: 190, P. 106813 - 106813

Published: Dec. 14, 2022

Language: Английский

---

Knowledge matters: Chest radiology report generation with general and specific knowledge

Medical Image Analysis, Journal Year: 2022, Volume and Issue: 80, P. 102510 - 102510

Published: June 9, 2022

Automatic chest radiology report generation is critical in clinics which can relieve experienced radiologists from the heavy workload and remind inexperienced of misdiagnosis or missed diagnose. Existing approaches mainly formulate as an image captioning task adopt encoder-decoder framework. However, medical domain, such pure data-driven suffer following problems: 1) visual textual bias problem; 2) lack expert knowledge. In this paper, we propose a knowledge-enhanced approach introduces two types knowledge: General knowledge, input independent provides broad knowledge for generation; Specific dependent fine-grained X-ray generation. To fully utilize both general specific also multi-head attention mechanism. By merging features with proposed model improve quality generated reports. The experimental results on publicly available IU-Xray dataset show that outperforms state-of-the-art methods almost all metrics. And MIMIC-CXR par methods. Ablation studies demonstrate help to performance

Language: Английский

---

Explainable Intrusion Detection for Cyber Defences in the Internet of Things: Opportunities and Solutions

IEEE Communications Surveys & Tutorials, Journal Year: 2023, Volume and Issue: 25(3), P. 1775 - 1807

Published: Jan. 1, 2023

The field of Explainable Artificial Intelligence (XAI) has garnered considerable research attention in recent years, aiming to provide interpretability and confidence the inner workings state-of-the-art deep learning models. However, XAI-enhanced cybersecurity measures Internet Things (IoT) its sub-domains, require further investigation effective discovery attack surfaces, their corresponding vectors, interpretable justification model outputs. Cyber defence involves operations conducted supporting mission objectives identify prevent cyberattacks using various tools techniques, including intrusion detection systems (IDS), threat intelligence hunting, prevention. In cyber defence, especially anomaly-based IDS, emerging applications models interpretation models' architecture explanation prediction examine how would occur. This paper presents a comprehensive review XAI techniques for IoT networks. Firstly, we IDSs focusing on can augment them trust detections. Secondly, AI models, machine (ML) (DL), anomaly ecosystems. Moreover, discuss DL's ability effectively learn from large-scale datasets, accomplishing high performances discovering interpreting security events. Thirdly, demonstrate intersection XAI, IDS IoT. Finally, current challenges solutions perspective networks, revealing future directions. By analysing our findings, new that emerge, assisting decision-makers understanding explaining events compromised

Language: Английский

---

Multi-task deep learning for medical image computing and analysis: A review

Computers in Biology and Medicine, Journal Year: 2022, Volume and Issue: 153, P. 106496 - 106496

Published: Dec. 28, 2022

Language: Английский

---