Identifying the most infectious lesions in pulmonary tuberculosis by high-resolution multi-detector computed tomography

Jun Jun Yeh, Solomon Chih Cheng Chen, Wen Bao Teng, Chun Hsiung Chou, Shih Peng Hsieh, Tsung Lung Lee, Ming Ting Wu*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

32 Scopus citations

Abstract

Objective: This study aimed to determine whether characteristics detected by multi-detector computed tomography (MDCT) were predictive of highly infectious, smear-positive, active pulmonary tuberculosis (PTB). Methods: Among 124 patients with active PTB, 84 had positive (group 1) and 40 had negative (group 2) smear results for acid-fast bacilli. Multiplanar MDCT, axial conventional CT and chest X-ray images were analysed retrospectively for morphology, number, and segmental (lobe) distribution of lesions. Results: By multivariate analysis, consolidation over any segment of the upper, middle, or lingual lobes, cavitations, and clusters of nodules were associated with group 1, while centrilobular nodules were predictive of group 2. Using five independent variables associated with risk in group 1, a prediction model was created to distinguish between group 1 and group 2. ROC curve analysis showed an area under the curve of 0.951±0.021 for this prediction model. With the ideal cutoff point score of 1, the sensitivity, specificity, and positive predictive values were 84.5%, 97.5%, and 98.0%, respectively. Conclusions: A model to predict smear-positive active PTB on the basis of findings from MDCT may be a useful tool for clinical decisions about isolating patients pending sputum smear results.

Original languageEnglish
Pages (from-to)2135-2145
Number of pages11
JournalEuropean Radiology
Volume20
Issue number9
DOIs
StatePublished - Sep 2010

Keywords

  • Decision making
  • Patient isolation
  • Receiver-operating characteristic (ROC) curve
  • Sputum
  • Tuberculosis

Fingerprint

Dive into the research topics of 'Identifying the most infectious lesions in pulmonary tuberculosis by high-resolution multi-detector computed tomography'. Together they form a unique fingerprint.

Cite this