Prediction of multiple drug resistant pulmonary tuberculosis against drug sensitive pulmonary tuberculosis by CT nodular consolidation sign

Huang, Xiuzhen; Skrahin, Aliaksandr; Lu, Pu-Xuan; Alexandru, Sofia; Crudu, Valeriu; Astrovko, Andrei; Skrahina, Alena; Taaffe, Jessica; Harris, Michael A.; Long, Alyssa; Wollenberg, Kurt; Engle, Eric; Hurt, Darrell E.; Akhundova, Irada; Ismayilov, Sharafat; Mammadbayov, Elcan; Gadirova, Hagigat; Abuzarov, Rafik; Seyfaddinova, Mehriban; Avaliani, Zaza; Vashakidze, Sergo; Shubladze, Natalia; Nanava, Ucha; Strâmbu, Irina; Zaharia, Dragoş; Muntean, Alexandru; Ghita, Eugenia; Bogdan, Miron Alexandru; Mindru, Roxana; Spînu, Victor; Sora, Alexandra; Ene, Catalina; Sergueev, Eugene; Kirichenko, V. V.; Lapitski, Vladzimir; Snezhko, Eduard; Kovalev, Vassili; Tuzikov, Alexander V.; Gabrielian, Andrei; Rosenthal, Alex; Tartakovsky, Michael; Wáng, Yì Xiáng J.

doi:10.1101/833954

Cited by 5 publications

(9 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our study confirms that large nodules and large cavities are more common in DR-TB. In addition, we found that multiple nodules and multiple cavities are more common in DR-TB, which confirms the analysis in (5,13).…”

Section: Roles Of Radiological Featuressupporting

confidence: 88%

“…Wáng et al (13) performed a review on available articles before 2018 for radiological signs of MDR-TB and found that thick-walled multiple cavities (particularly with count ≥3 and size ≥30 mm) present the most promising radiological sign for MDR-TB with good specificity but at the cost of low sensitivity. Huang et al (5) reported from 468 DR-TB cases and 223 DS-TB cases that a combination of consolidated nodule number and size can be used to predict the probability of MDR-TB. Flores-Treviño et al (14) found from 144 patients in Mexico that multiple cavities is a predictor for DR-TB.…”

Section: Previous Workmentioning

confidence: 99%

“…Inspired by the work in (5), we hypothesize that the number of affected sextants may serve as important radiological features for DR-TB identification, and that both clinical and radiological features are important factors for machine classification between DR-TB and DS-TB.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Differentiating between drug-sensitive and drug-resistant tuberculosis with machine learning for clinical and radiological features

Yang¹,

Yu²,

Kantipudi³

et al. 2022

Quant Imaging Med Surg

Self Cite

View full text Add to dashboard Cite

Background: Tuberculosis (TB) drug resistance is a worldwide public health problem that threatens progress made in TB care and control. Early detection of drug resistance is important for disease control, with discrimination between drug-resistant TB (DR-TB) and drug-sensitive TB (DS-TB) still being an open problem. The objective of this work is to investigate the relevance of readily available clinical data and data derived from chest X-rays (CXRs) in DR-TB prediction and to investigate the possibility of applying machine learning techniques to selected clinical and radiological features for discrimination between DR-TB and DS-TB. We hypothesize that the number of sextants affected by abnormalities such as nodule, cavity, collapse and infiltrate may serve as a radiological feature for DR-TB identification, and that both clinical and radiological features are important factors for machine classification of DR-TB and DS-TB.Methods: We use data from the NIAID TB Portals program (https://tbportals.niaid.nih.gov), 1,455 DR-TB cases and 782 DS-TB cases from 11 countries. We first select three clinical features and 26 radiological features from the dataset. Then, we perform Pearson's chi-squared test to analyze the significance of the selected clinical and radiological features. Finally, we train machine classifiers based on different features and evaluate their ability to differentiate between DR-TB and DS-TB.Results: Pearson's chi-squared test shows that two clinical features and 23 radiological features are statistically significant regarding DR-TB vs. DS-TB. A ten-fold cross-validation using a support vector machine shows that automatic discrimination between DR-TB and DS-TB achieves an average accuracy of 72.34% and an average AUC value of 78.42%, when combing all 25 statistically significant features. Conclusions:Our study suggests that the number of affected lung sextants can be used for predicting DR-TB, and that automatic discrimination between DR-TB and DS-TB is possible, with a combination of clinical features and radiological features providing the best performance.

show abstract

Section: Roles Of Radiological Featuressupporting

confidence: 88%

Section: Previous Workmentioning

confidence: 99%

See 1 more Smart Citation

Differentiating between drug-sensitive and drug-resistant tuberculosis with machine learning for clinical and radiological features

Yang¹,

Yu²,

Kantipudi³

et al. 2022

Quant Imaging Med Surg

Self Cite

View full text Add to dashboard Cite

show abstract

“…This study concluded that there were substantial differences in findings between the two classes in terms of lesion size and morphology. A slightly larger study [ 6 ], which compared 468 DR-TB cases and 223 DS-TB cases concluded that a combination of the number and size of consolidated nodules is a good predictor for DR-TB. Another, small study [ 7 ], utilized data from 144 patients and found that the presence of multiple cavities is a good predictor for DR-TB.…”

Section: Previous Workmentioning

confidence: 99%

Generalization Challenges in Drug-Resistant Tuberculosis Detection from Chest X-rays

et al. 2022

View full text Add to dashboard Cite

Classification of drug-resistant tuberculosis (DR-TB) and drug-sensitive tuberculosis (DS-TB) from chest radiographs remains an open problem. Our previous cross validation performance on publicly available chest X-ray (CXR) data combined with image augmentation, the addition of synthetically generated and publicly available images achieved a performance of 85% AUC with a deep convolutional neural network (CNN). However, when we evaluated the CNN model trained to classify DR-TB and DS-TB on unseen data, significant performance degradation was observed (65% AUC). Hence, in this paper, we investigate the generalizability of our models on images from a held out country’s dataset. We explore the extent of the problem and the possible reasons behind the lack of good generalization. A comparison of radiologist-annotated lesion locations in the lung and the trained model’s localization of areas of interest, using GradCAM, did not show much overlap. Using the same network architecture, a multi-country classifier was able to identify the country of origin of the X-ray with high accuracy (86%), suggesting that image acquisition differences and the distribution of non-pathological and non-anatomical aspects of the images are affecting the generalization and localization of the drug resistance classification model as well. When CXR images were severely corrupted, the performance on the validation set was still better than 60% AUC. The model overfitted to the data from countries in the cross validation set but did not generalize to the held out country. Finally, we applied a multi-task based approach that uses prior TB lesions location information to guide the classifier network to focus its attention on improving the generalization performance on the held out set from another country to 68% AUC.

show abstract

“…The tool has also been used in scientific publications to advance clinical research on DR-TB. 13 , 20 , 28 , 29 …”

Section: Discussionmentioning

confidence: 99%

The technology behind TB DEPOT: a novel public analytics platform integrating tuberculosis clinical, genomic, and radiological data for visual and statistical exploration

Long

Glogowski

Meppiel

et al. 2020

Journal of the American Medical Informatics Association

Self Cite

View full text Add to dashboard Cite

Objective Clinical research informatics tools are necessary to support comprehensive studies of infectious diseases. The National Institute of Allergy and Infectious Diseases (NIAID) developed the publicly accessible Tuberculosis Data Exploration Portal (TB DEPOT) to address the complex etiology of tuberculosis (TB). Materials and Methods TB DEPOT displays deidentified patient case data and facilitates analyses across a wide range of clinical, socioeconomic, genomic, and radiological factors. The solution is built using Amazon Web Services cloud-based infrastructure, .NET Core, Angular, Highcharts, R, PLINK, and other custom-developed services. Structured patient data, pathogen genomic variants, and medical images are integrated into the solution to allow seamless filtering across data domains. Results Researchers can use TB DEPOT to query TB patient cases, create and save patient cohorts, and execute comparative statistical analyses on demand. The tool supports user-driven data exploration and fulfills the National Institute of Health’s Findable, Accessible, Interoperable, and Reusable (FAIR) principles. Discussion TB DEPOT is the first tool of its kind in the field of TB research to integrate multidimensional data from TB patient cases. Its scalable and flexible architectural design has accommodated growth in the data, organizations, types of data, feature requests, and usage. Use of client-side technologies over server-side technologies and prioritizing maintenance have been important lessons learned. Future directions are dynamically prioritized and key functionality is shared through an application programming interface. Conclusion This paper describes the platform development methodology, resulting functionality, benefits, and technical considerations of a clinical research informatics application to support increased understanding of TB.

show abstract

Prediction of multiple drug resistant pulmonary tuberculosis against drug sensitive pulmonary tuberculosis by CT nodular consolidation sign

Cited by 5 publications

References 11 publications

Differentiating between drug-sensitive and drug-resistant tuberculosis with machine learning for clinical and radiological features

Differentiating between drug-sensitive and drug-resistant tuberculosis with machine learning for clinical and radiological features

Generalization Challenges in Drug-Resistant Tuberculosis Detection from Chest X-rays

The technology behind TB DEPOT: a novel public analytics platform integrating tuberculosis clinical, genomic, and radiological data for visual and statistical exploration

Contact Info

Product

Resources

About