Machine learning assisted evaluation of interstitial lung diseases

Abstract

Interstitial lung disease (ILD) is an inflammatory condition encompassing greater than 200 different chronic disorders that often lead to pulmonary fibrosis and the attendant morbidity and mortality. Distinguishing one ILD from another can be challenging even for experts given similar clinical manifestations. However, firm diagnosis is important for management, counseling, and surveillance. Computer-Aided Lung Informatics for Pathology Evaluation and Ratings (CALIPER, Mayo Clinic, Rochester, MN, USA), a machine learned image analysis tool for characterizing and quantifying diffused lung diseases on CT, has been shown to correlate with ILD mortality such as for idiopathic pulmonary fibrosis. The CALIPER assessment of ILD is not influenced by confounding conditions such as emphysema or pulmonary hypertension that may affect pulmonary function measurements. Moreover, CALIPER can provide more reproducible results with less inter- and intraobserver variability compared to the conventional subjective visual assessment of CT scans by radiologists. Herein, we aim to provide an overview of ILD and discuss CALIPER evaluation of ILD.