Improved Detection of Air Trapping on Expiratory Computed Tomography Using Deep Learning

Abstract

BackgroundRadiologic evidence of air trapping (AT) on expiratory CT scans is associated with early pulmonary dysfunction in patients with cystic fibrosis (CF). However, standard techniques for quantitative assessment of AT are highly variable, resulting in limited efficacy for monitoring disease progression.ObjectiveTo investigate the effectiveness of a convolutional neural network (CNN) model for quantifying and monitoring AT, and to compare it with other quantitative AT measures obtained from threshold-based techniques.Materials and MethodsPaired volumetric whole lung inspiratory and expiratory CT scans were obtained at four time points (0, 3, 12 and 24 months) on 36 subjects with mild CF lung disease. A densely connected CNN (DN) was trained using AT segmentation maps generated from a personalized threshold-based method (PTM). Quantitative AT (QAT) values, presented as the relative volume of AT over the lungs, from the DN approach were compared to QAT values from the PTM method. Radiographic assessment, spirometric measures, and clinical scores were correlated to the DN QAT values using a linear mixed effects model.ResultsQAT values from the DN were found to increase from 8.65% ± 1.38% to 21.38% ± 1.82%, respectively, over a two-year period. Comparison of CNN model results to intensity-based measures demonstrated a systematic drop in the Dice coefficient over time (decreased from 0.86 ± 0.03 to 0.45 ± 0.04). The trends observed in DN QAT values were consistent with clinical scores for AT, bronchiectasis, and mucus plugging. In addition, the DN approach was found to be less susceptible to variations in expiratory deflation levels than the threshold-based approach.ConclusionThe CNN model effectively delineated air trapping on expiratory CT scans, which provides an automated and objective approach for assessing and monitoring air trapping in CF patients.

A Computer-based Analysis for Identification and Quantification of Small Airway Disease in Lung Computed Tomography Images

“…Ram and colleagues developed a trained CNN for the quantification of AT, using AT segmentation maps generated from PTM. The results of this study, which are in line with the primary research proposing a CNN model for AT quantification in CT images, revealed that QAT values obtained by the proposed CNN model were correlated with the clinical scores of AT and were less prone to deflation level variations in expiration compared with PTM 55…”

“…Quantitative AT measurements are highly susceptible to the deflation level during expiratory CT. Ram and colleagues found that QAT measured by PTM is highly sensitive to slight deviations from the RV. The results showed a 15% increase in AT measured by PTM at a simulated deflation level of 80% of RV, linearly increasing with the deflation level 55…”

“…It seems that PTM provides a precise estimation of AT in young patients with cystic fibrosis 54. Goris and colleagues showed that the percentage of AT volume, regardless of its definition, is a sensitive discriminator between patients with mild cystic fibrosis and control cases 54,55. Quantitative AT measurements are highly susceptible to the deflation level during expiratory CT. Ram and colleagues found that QAT measured by PTM is highly sensitive to slight deviations from the RV.…”

“…The optimal threshold for each radiologist is recorded and used to quantify AT values in a given expiratory CT scan. Finally, the QAT values measured by all radiologists are averaged for a given expiratory CT scan 55. Clearly, visual assessment of AT is subject to limitations inherent to intra-rater subjectivity and reliability.…”

“…54 Goris and colleagues showed that the percentage of AT volume, regardless of its definition, is a sensitive discriminator between patients with mild cystic fibrosis and control cases. 54,55 Quantitative AT measurements are highly susceptible to the deflation level during expiratory CT. Ram and colleagues found that QAT measured by PTM is highly sensitive to slight deviations from the RV. The results showed a 15% increase in AT measured by PTM at a simulated deflation level of 80% of RV, linearly increasing with the deflation level.…”