Artificial Intelligence for Interstitial Lung Disease Analysis on Chest Computed Tomography: A Systematic Review

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence

Background:

• High-resolution computed tomography (HRCT) is crucial for diagnosing interstitial lung diseases (ILD).
• Inter- and intra-observer variability can challenge HRCT interpretation in ILD.
• Artificial intelligence (AI) is emerging as a transformative tool in medical image analysis.

Purpose of the Study:

• To systematically evaluate the application of AI in analyzing ILD on HRCT scans.
• To assess the potential of AI in advancing patient care for ILD.

Main Methods:

• A systematic literature search was conducted on MEDLINE/PubMed for deep learning studies on ILD analysis using chest CT.
• Studies published up to March 1, 2021, were included.
• Risk of bias was assessed using Quality Assessment of Diagnostic Accuracy Studies and Joanna Briggs Institute Critical Appraisal checklist.

Main Results:

• Nineteen retrospective studies were analyzed, focusing on AI for ILD detection, segmentation, and classification.
• AI achieved accuracies of 78%-91% in classifying ILD patterns.
• Two studies showed near-expert performance for idiopathic pulmonary fibrosis (IPF) diagnosis, but 78.9% of studies had a high risk of bias.

Conclusions:

• AI holds potential for improving radiologic diagnosis and classification of ILD.
• Current AI accuracy is not yet satisfactory, and research is limited by retrospective studies.
• Well-designed prospective studies are needed to reliably assess AI tools for ILD evaluation.