Lung CT to identify COPD is subject to certain technical constraints, particularly to obtain conclusive results in inspiration and expiration. A study published in the Journal Radiology: Cardiothoracic imaging uses a neural network coupled with inspiration-only CT to obtain relevant diagnoses.
Chronic Obstructive Pulmonary Disease (COPD) is commonly diagnosed by a spirometry test, which measures lung function by the amount of air that can be inhaled and exhaled, as well as the speed of exhalation.
CT scan rather than spirometry to identify COPD
But lung computed tomography (CT) has supplanted this practice, with a procedure that generally requires two acquisitions, one at full inspiration, the other at expiration. However, it is possible, depending on a study published in the Journal Radiology: Cardiothoracic imagingto obtain a precise diagnosis from a single inspirational CT and a deep learning model.
« Although studies have recently shown that lung structure, measured quantitatively using lung CT, can complement the staging, diagnosis, and prognosis of COPD severity, many of these studies require the acquisition of two CT seriesspecifies the author of the study, Professor Kyle A. Hasenstab, assistant professor of statistics and data science at San Diego State University (California – USA). However, this type of protocol is not clinically standard across all institutions.»
Significant constraints in certain centers to obtain convincing results in inspiration and expiration
Some hospitals are unable to implement expiratory CT protocols due to additional training requirements for MERMs and radiologists, according to Professor Hasenstab. Additionally, some elderly patients with impaired lung function have difficulty maintaining apnea, as is necessary when acquiring expiratory images. This can affect image quality and diagnostic accuracy.
Dr. Hasenstab and colleagues hypothesized that a single inspiratory CT scan combined with a convolutional neural network (CNN) and clinical data would be sufficient for the diagnosis and staging of COPD. In this retrospective study, inspiratory and expiratory chest CT images, as well as spirometry data, were acquired from 8,893 patients from November 2007 to April 2011. The average age of patients included in the study was 59 years. and all had a history of smoking.
-A neural network coupled with inspiration-only CT as an alternative
The CNN was trained to predict spirometry measurements using clinical data and single-phase or multi-phase lung CT. The spirometry pre-results were then used to predict the severity stage of the disease, according to the Global Initiative for Obstruct Lung Disease (GOLD) classification system, which has four classifications. The study results showed that a CNN model developed using a single CT image of a single respiratory phase accurately diagnosed COPD and was also accurate in identifying a GOLD stage.
A model that provides good staging of COPD according to the GOLD classification
The model performed similarly to COPD diagnoses using combined inspiration and expiration CT measurements. “ Although many imaging protocols for the diagnosis and staging of COPD require two CT acquisitions, our study shows that the diagnosis and staging of COPD is feasible with a single acquisition. », adds Professor Hasenstab. When clinical data was added, the CNN model’s predictions were even more accurate.
« Reducing inspiratory CT to a single acquisition may increase accessibility for this diagnostic approach while reducing cost, discomfort and exposure to ionizing radiation for patients. », concludes Professor Hasenstab.
Paolo Royan
Related News :