Revisiting Computer-Aided Tuberculosis Diagnosis DOI
Yun Liu, Yu-Huan Wu, Shichen Zhang

et al.

IEEE Transactions on Pattern Analysis and Machine Intelligence, Journal Year: 2023, Volume and Issue: 46(4), P. 2316 - 2332

Published: Nov. 7, 2023

Tuberculosis (TB) is a major global health threat, causing millions of deaths annually. Although early diagnosis and treatment can greatly improve the chances survival, it remains challenge, especially in developing countries. Recently, computer-aided tuberculosis (CTD) using deep learning has shown promise, but progress hindered by limited training data. To address this, we establish large-scale dataset, namely X-ray (TBX11 K) which contains 11 200 chest (CXR) images with corresponding bounding box annotations for TB areas. This dataset enables sophisticated detectors high-quality CTD. Furthermore, propose strong baseline, SymFormer, simultaneous CXR image classification infection area detection. SymFormer incorporates Symmetric Search Attention (SymAttention) to tackle bilateral symmetry property discriminative features. Since may not strictly adhere property, also Positional Encoding (SPE) facilitate SymAttention through feature recalibration. promote future research on CTD, build benchmark introducing evaluation metrics, evaluating baseline models reformed from existing detectors, running an online challenge. Experiments show that achieves state-of-the-art performance TBX11 K dataset.

Language: Английский

Trends and age-period-cohort effects on the incidence of hand, foot, and mouth disease among children younger than 10 years old from 2009 to 2023 in Quzhou, China DOI Creative Commons
Fei Zhao,

Canya Fu,

Wenjie Xu

et al.

Critical Public Health, Journal Year: 2025, Volume and Issue: 35(1)

Published: May 10, 2025

Language: Английский

Citations

0

Trend analysis and prediction of the incidence and mortality of CKD in China and the US DOI Creative Commons
Wenpeng Zhu,

Mengqi Han,

Yuxin Wang

et al.

BMC Nephrology, Journal Year: 2024, Volume and Issue: 25(1)

Published: March 1, 2024

Currently, limited research is available on the comparative analysis of chronic kidney disease (CKD) incidence and mortality rates between China United States. This study aimed to explore trends in CKD both countries, as well make some future predictions.

Language: Английский

Citations

3

Temporal trends and projections in the global burden of neck pain: findings from the Global Burden of Disease Study 2019 DOI Creative Commons
Siqing Cheng, Jin Cao, Leying Hou

et al.

Pain, Journal Year: 2024, Volume and Issue: 165(12), P. 2804 - 2813

Published: June 19, 2024

Abstract Data were obtained from the Global Burden of Disease study 2019. Joinpoint regression model was used to analyze temporal trends 1990 2019 neck pain burden, focusing on age-standardized incidence rates, prevalence and years lived with disability (YLDs) rates at global, regional, national levels. The age–period–cohort analysis estimate effects age (5-99 years), period (1990-2019), cohort (1893-2012) Future projections for global burden 2020 2044 estimated using nordpred model. From 2019, incidence, cases, YLDs counts have increased by 71.89%, 98.21%, 78.17%, respectively. joinpoint indicated significant shifts in which varied across regions nations. that predominantly concentrated middle-aged older age, showing minimal variation Compared incident prevalent projected increase 134%, 142%, 140% 2044. has persisted a relatively elevated level indicating continuing upward trend. research is urgently needed better understand predictors clinical course enhance prevention management strategies.

Language: Английский

Citations

3

Comparison of microscopic and xpert MTB diagnoses of presumptive mycobacteria tuberculosis infection: retrospective analysis of routine diagnosis at Cape Coast Teaching Hospital DOI Creative Commons
Kwame Kumi Asare,

Daniel E. Azumah,

Czarina Owusua Adu-Gyamfi

et al.

BMC Infectious Diseases, Journal Year: 2024, Volume and Issue: 24(1)

Published: July 2, 2024

Abstract Introduction Tuberculosis is a global health problem that causes 1. 4 million deaths every year. It has been estimated sputum smear-negative diagnosis but culture-positive pulmonary TB contribute to 12.6% of transmission. by smear microscopy minimum detection limit (LOD) 5,000 10,000 bacilli per milliliter (CFU/ml) result in missed cases and false positives. However, GeneXpert technology, with LOD 131–250 CFU/ml samples its implementation believe facilitate early drug-resistant case. Since 2013, Ghana Service (GHS) introduce MTB/RIF diagnostic all regional hospitals Ghana, however no assessment performance between cross the facilities reported. The study compared results routine diagnoses Xpert MTB from 2016 2020 at Cape Coast Teaching Hospital (CCTH). Methods microscopic (CCTH) retrospectively. Briefly, specimens were collected into 20 mL sterile screw-capped containers for each case suspected infection processed within 24 h. decontaminated using NALC-NaOH method final NaOH concentration 1%. supernatants discarded after centrifuge remaining pellets dissolved 1–1.5 ml phosphate buffer saline (PBS) used diagnosis. A fixed smears Ziehl-Neelsen acid-fast stain observed under microscope remainings data analyze GraphPad Prism. Results 50.11% (48.48–51.38%) females an odd ratio (95% CI) 1.004 (0.944–1.069) more likely report clinic smear-positive first 6.6% (5.98–7.25%), second was 6.07% (5.45–6.73%). MTB-RIF detected 2.93% (10/341) (1.42–5.33%) 5.44% (16/294) (3.14–8.69%) samples. prevalence across positive showed males had 56.87% (178/313) 56.15% (137/244) 43.13% (135/313) 43.85% (107/244) sputum. Also, negative 0.18% (10/5607) 1 0.31% (16/5126) 2. Conclusion In conclusion, highlights higher sensitivity assay traditional detecting MTB. identified 10 16 2 which negative.

Language: Английский

Citations

3

Revisiting Computer-Aided Tuberculosis Diagnosis DOI
Yun Liu, Yu-Huan Wu, Shichen Zhang

et al.

IEEE Transactions on Pattern Analysis and Machine Intelligence, Journal Year: 2023, Volume and Issue: 46(4), P. 2316 - 2332

Published: Nov. 7, 2023

Tuberculosis (TB) is a major global health threat, causing millions of deaths annually. Although early diagnosis and treatment can greatly improve the chances survival, it remains challenge, especially in developing countries. Recently, computer-aided tuberculosis (CTD) using deep learning has shown promise, but progress hindered by limited training data. To address this, we establish large-scale dataset, namely X-ray (TBX11 K) which contains 11 200 chest (CXR) images with corresponding bounding box annotations for TB areas. This dataset enables sophisticated detectors high-quality CTD. Furthermore, propose strong baseline, SymFormer, simultaneous CXR image classification infection area detection. SymFormer incorporates Symmetric Search Attention (SymAttention) to tackle bilateral symmetry property discriminative features. Since may not strictly adhere property, also Positional Encoding (SPE) facilitate SymAttention through feature recalibration. promote future research on CTD, build benchmark introducing evaluation metrics, evaluating baseline models reformed from existing detectors, running an online challenge. Experiments show that achieves state-of-the-art performance TBX11 K dataset.

Language: Английский

Citations

8