The value of radiomic features in chronic obstructive pulmonary disease assessment: a prospective study

To investigate whether radiomic features could supply additional information in the assessment of lung function, health status, and exacerbation of chronic obstructive pulmonary disease (COPD).This prospective study enrolled 197 patients who underwent high-resolution chest computed tomography (CT) between 2009 and 2011. A total of 17 radiomic features from chest CT images were selected to explore the applicability of radiomics in the evaluation of COPD. Spirometric lung function testing, modified Medical Research Council (mMRC), St George's respiratory questionnaire (SGRQ), 6-minute walk test (6MWT), and exacerbations were collected at baseline and 1-year follow-up. The relationships of radiomic features, lung function, health status, and exacerbation were assessed using linear and logistic multivariate analyses.At baseline visit, radiomic features were significantly associated with forced expiratory volume in 1 second percent of predicted (FEV1 % predicted), FEV1/forced vital capacity (FVC), mMRC, and SGRQ after multivariate adjustment. At 1-year follow-up, original_10thpercentile was associated with mMRC change, and original_10thpercentile, original_elongation, original_sphericity, and original_glv were associated with SGRQ change. Original_flatness was associated with exacerbation and can improve the prediction performance of other models for exacerbation. No radiomic features correlated with the 6MWT.Radiomic features were associated with airflow obstruction, health status, and exacerbation of COPD independent of other factors, showing promise in the assessment of COPD.