Evaluation of visual and computer‐based CT analysis for the identification of functional patterns of obstruction and restriction in hypersensitivity pneumonitis

• Published in: Respirology (Carlton, Vic.) Vic.), 2017-11, Vol.22 (8), p.1585-1591
• Main Authors: Jacob, Joseph, Bartholmai, Brian J., Brun, Anne Laure, Egashira, Ryoko, Rajagopalan, Srinivasan, Karwoski, Ronald, Kouranos, Vasileios, Kokosi, Maria, Hansell, David M., Wells, Athol U.
• Format: Article
• Language: English
• Subjects: Adult; Aged; air trapping; Airway Obstruction - diagnosis; Airway Remodeling - physiology; Alveolitis; Alveolitis, Extrinsic Allergic - diagnosis; Alveolitis, Extrinsic Allergic - physiopathology; Carbon monoxide; Carbon Monoxide - analysis; Computed tomography; Female; Humans; Hypersensitivity; hypersensitivity pneumonitis; Image Interpretation, Computer-Assisted - methods; Lung - diagnostic imaging; Lung - pathology; Lung - physiopathology; Lung diseases; Male; Middle Aged; Multivariate analysis; Obstructive lung disease; Outcome Assessment (Health Care); Pneumonitis; pulmonary vessel volume; quantitative computer analysis; Respiratory diseases; Respiratory Function Tests - methods; Tomography, X-Ray Computed - methods; visual computed tomography analysis
• ISSN: 1323-7799; 1440-1843
• DOI: 10.1111/resp.13122

ABSTRACT Background and objective To determine whether computer‐based quantification (CALIPER software) is superior to visual computed tomography (CT) scoring in the identification of CT patterns indicative of restrictive and obstructive functional indices in hypersensitivity pneumonitis (HP). Methods A total of 135 consecutive HP patients had CT parenchymal patterns evaluated quantitatively by both visual scoring and CALIPER. Results were evaluated against: forced vital capacity (FVC), total lung capacity (TLC), diffusing capacity for carbon monoxide (DLCO) and a composite physiological index (CPI) to identify which CT scoring method better correlated with functional indices. Results CALIPER‐derived scores of total interstitial lung disease extent correlated more strongly than visual scores: FVC (CALIPER R = 0.73, visual R = 0.51); DLCO (CALIPER R = 0.61, visual R = 0.48); and CPI (CALIPER R = 0·70, visual R = 0·55). The CT variable that correlated most strongly with restrictive functional indices was CALIPER pulmonary vessel volume (PVV): FVC R = 0.75, DLCO R = 0.68 and CPI R = 0.76. Ground‐glass opacity quantified by CALIPER alone demonstrated strong associations with restrictive functional indices: CALIPER FVC R = 0.65; DLCO R = 0.59; CPI R = 0.64; and visual = not significant. Decreased attenuation lung quantified by CALIPER was a better morphological measure of obstructive lung disease than equivalent visual scores as judged by relationships with TLC (CALIPER R = 0.63 and visual R = 0.12). All results were maintained on multivariate analysis. Conclusion CALIPER improved on visual scoring in HP as judged by restrictive and obstructive functional correlations. Decreased attenuation regions of the lung quantified by CALIPER demonstrated better linkages to obstructive lung physiology than visually quantified CT scores. A novel CALIPER variable, the PVV, demonstrated the strongest linkages with restrictive functional indices and could represent a new automated index of disease severity in HP. See related Editorial In hypersensitivity pneumonitis, computer‐derived (CALIPER) variables are stronger predictors of restriction and obstruction than visual computed tomography (CT) scores as judged by correlations between morphology and functional indices. The pulmonary vessel volume demonstrated the strongest correlations of all CT variables with restrictive indices and may represent a new measure of interstitial damage.