Diagnosis of rhinosinusitis with a colorimetric sensor array

Abstract

A variety of electronic nose technologies are currently commercially available or under evaluation at research institutions. Sensors include conducting polymers, polymer composites, acoustic wave, metal oxide and optical fiber arrays. Reactive, metalloporphyrin dyes have also been used as colorimetric sensor arrays (CSAs). We hypothesized that a CSA array could be used to distinguish exhaled gas from patients with chronic bacterial sinusitis versus those without. Exhaled gas was selectively sampled from the nasal passages using a nasal CPAP (continuous positive airway pressure) mask and pumped across an array of 36 metalloporphyrin sensor dots printed on a paper sensor cartridge, which was photographed at 2 min intervals during the sampling period for a total of 12 min. Changes in the red, green and blue values for each of the senor dots were recorded for each time interval. Eleven patients with sinusitis were compared to nine control patients. A leave-one-out classification test was carried out using binary logistic regression, which employed the principal component projections of data as features. The classification rate was as high as 90% accurate at 2, 4 and 6 min, following exposure using 12 principal components. Classification accuracy fell off when more principal components were used and at longer time intervals following exposure. Chronic sinusitis is typically a clinical diagnosis and the etiologic organisms include both Gram-positive and Gram-negative bacteria. Accurate classification at 90% is very encouraging and warrants further investigation.