Premium
Diagnosis of Pneumonia With an Electronic Nose: Correlation of Vapor Signature With Chest Computed Tomography Scan Findings
Author(s) -
Hockstein Neil G.,
Thaler Erica R.,
Torigian Drew,
Miller Wallace T.,
Deffenderfer Olivia,
Hanson C William
Publication year - 2004
Publication title -
the laryngoscope
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.181
H-Index - 148
eISSN - 1531-4995
pISSN - 0023-852X
DOI - 10.1097/00005537-200410000-00005
Subject(s) - medicine , electronic nose , pneumonia , computed tomography , radiology , prospective cohort study , nose , intensive care unit , mechanical ventilation , exhalation , ventilator associated pneumonia , diagnostic accuracy , nuclear medicine , surgery , intensive care medicine , artificial intelligence , computer science
Objectives/Hypothesis: The electronic nose is a sensor of volatile molecules that is useful in the analysis of expired gases. The device is well suited to testing the breath of patients receiving mechanical ventilation and is a potential diagnostic adjunct that can aid in the detection of patients with ventilator‐associated pneumonia. Study Design: A prospective study. Methods: We performed a prospective study of patients receiving mechanical ventilation in a surgical intensive care unit who underwent chest computed tomography (CT) scanning. A single attending radiologist reviewed the chest CT scans, and imaging features were recorded on a standardized form. Within 48 hours of chest CT scan, five sets of exhaled gas were sampled from the expiratory limb of the ventilator circuit. The gases were assayed with a commercially available electronic nose. Both linear and nonlinear analyses were performed to identify correlations between imaging features and the assayed gas signatures. Results: Twenty‐five patients were identified, 13 of whom were diagnosed with pneumonia by CT scan. Support vector machine analysis was performed in two separate analyses. In the first analysis, in which a training set was identical to a prediction set, the accuracy of prediction results was greater than 91.6%. In the second analysis, in which the training set and the prediction set were different, the accuracy of prediction results was at least 80%, with higher accuracy depending on the specific parameters and models being used. Conclusion: The electronic nose is a new technology that continues to show promise as a potential diagnostic adjunct in the diagnosis of pneumonia and other infectious diseases.