Quantitative Evaluation of Airway Lesions in Chronic Obstructive Pulmonary Disease by Applying Deep Learning Reconstruction to Ultra-high-resolution CT Images: Correlation between Wall Area Percentage and Forced Expiratory Volume in One Second Percentage

Abstract

Using ultra-high-resolution images reconstructed with the Advanced intelligent Clear-IQ Engine (AiCE) lung to measure wall area percentage (WA%), we demonstrated that WA% measured in more distal bronchus has a stronger correlation with respiratory function (FEV1%). Furthermore, we also demonstrated that WA% measured from images with the higher spatial resolution has a stronger correlation with FEV1%. The modulation transfer function (MTF) and noise power spectrum (NPS) of the ultra-high-resolution images reconstructed by the AiCE body and the AiCE lung were compared. In addition, WA% from the first- to seventh-generation bronchus was measured for B1 and B10 in the right lung from clinical images obtained with the two reconstruction methods, and the correlation coefficients with FEV1% were evaluated. The MTF was more superior for the AiCE lung than for the AiCE body, and the NPS was lower for the AiCE lung than for the AiCE body in the low-frequency region. The correlation between WA% and FEV1% was slightly stronger in the AiCE lung than in the AiCE body. This study showed that WA% measured from the 7th-generation bronchus using ultra-high-resolution images reconstructed with the AiCE lung strengthens the correlation with FEV1%. Furthermore, the higher the spatial resolution of the measurement images, the stronger the correlation between WA% and FEV1%.