Computer-aided identification of interstitial lung disease based on computed tomography.

Area of Science:

• Pulmonary Medicine
• Radiology
• Medical Imaging Analysis

Background:

• Interstitial lung disease (ILD) diagnosis can be challenging using traditional pulmonary function tests (PFTs) and subjective radiological assessments.
• Objective quantitative imaging analysis is needed to improve diagnostic accuracy for ILD.

Purpose of the Study:

• To evaluate the efficacy of quantitative computed tomography (CT) with 3-dimensional (3D) reconstruction in distinguishing ILD patients from healthy controls.
• To identify quantitative CT parameters that can serve as reliable markers for ILD.

Main Methods:

• Retrospective analysis of chest CT images from 102 ILD patients and 102 healthy controls.
• Measurement of lung parenchymal volume, emphysema indices (LAA910, LAA950, LAA910%, LAA950%), and mean lung density (MLD).
• Statistical analysis including Mann-Whitney U test, ROC curves, discriminant analysis, deep neural networks, and logistic regression.

Main Results:

• ILD patients exhibited lower total lung volume and higher emphysema and MLD values compared to controls (P < 0.001).
• Quantitative CT parameters including LAA910 volume, LAA950 volume, LAA910%, LAA950%, and MLD effectively distinguished ILD patients (AUC > 0.5, P < 0.05).
• High mean lung density (MLD) was identified as a significant marker for ILD (OR = 1.047, P < 0.05).

Conclusions:

• Quantitative CT analysis provides an effective method for identifying ILD patients.
• This approach serves as a valuable alternative to subjective image interpretation and PFTs in ILD diagnosis.