Inclusion of clinical data can improve accuracy of convolutional neural network-based model predictions
By Elana Gotkine HealthDay Reporter
TUESDAY, Dec. 17, 2024 (HealthDay News) — Convolutional neural network (CNN)-based chronic obstructive pulmonary disease (COPD) diagnosis and staging using single-phase computed tomography (CT) has accuracy comparable to inspiratory-expiratory CT with inclusion of clinical data, according to a study published online Dec. 12 in Radiology: Cardiothoracic Imaging.
Amanda N. Lee, from San Diego State University, and colleagues conducted a retrospective study to measure the benefit of single-phase CT, inspiratory-expiratory CT, and clinical data for CNN-based COPD staging in a retrospective study using images and measurements acquired between November 2007 and April 2011 for 8,893 participants in the COPDGene phase 1 cohort.
The researchers observed moderate to good agreement for CNN-predicted and reference standard spirometry measurements (intraclass correlation coefficient [ICC], 0.66 to 0.79), which improved with inclusion of clinical data (ICC, 0.70 to 0.85) apart from forced expiratory volume in 1 second (FEV1)/forced vital capacity in the inspiratory-phase CNN model with clinical data and FEV1 in the expiratory-phase CNN model with clinical data. Single-phase CNN accuracies for Global Initiative for Chronic Obstructive Lung Disease (GOLD) stage, within-one GOLD stage, and diagnosis (GOLD 0 versus 1 to 4) varied from 59.8 to 84.1 percent, and agreement was moderate to good (ICC, 0.68 to 0.70). Using inspiratory and expiratory images, accuracies of CNN models varied from 60.0 to 86.3 percent, with moderate to good agreement (ICC, 0.72). For both the single-phase CNNs and inspiratory-expiratory CNNs, inclusion of clinical data improved agreement (ICCs: 0.72 and 0.77 to 0.78, respectively) and accuracy (62.5 to 85.8 percent and 67.6 to 88.0 percent, respectively).
“Although inspiratory CT remains the clinical standard, our findings suggest expiratory CT may be a strong candidate for imaging-based COPD staging,” the authors write.
COPDGene is supported by contributions made to an industry advisory board that includes pharmaceutical companies.
Copyright © 2024 HealthDay. All rights reserved.
