AI Enhances Lung Ultrasound Interpretation Across Clinicians with Varying Expertise Levels

Area of Science:

• Medical Imaging
• Artificial Intelligence in Medicine
• Pulmonary Medicine

Background:

• Lung ultrasound (LUS) is a key diagnostic tool for pulmonary conditions.
• LUS accuracy is heavily reliant on the expertise of the interpreting clinician.
• Variability in LUS interpretation can impact patient diagnosis and management.

Purpose of the Study:

• To assess the impact of an artificial intelligence (AI) tool on clinician performance in LUS interpretation.
• To evaluate AI's effectiveness in detecting pleural effusion and consolidation/atelectasis.
• To determine if AI assistance benefits clinicians across different experience levels.

Main Methods:

• A multi-reader, multi-case study involving 14 clinicians reviewing 374 LUS scans.
• Clinicians initially reviewed scans without AI, followed by a second review with AI-generated bounding boxes.
• Performance was measured using area under the curve (AUC), sensitivity, specificity, and Fleiss' Kappa.

Main Results:

• AI significantly improved AUC, sensitivity, and specificity for detecting pleural effusion and consolidation/atelectasis.
• Fleiss' Kappa values substantially increased with AI assistance, indicating improved inter-reader agreement.
• Performance gains were more pronounced in less experienced clinicians.

Conclusions:

• AI-powered tools can enhance diagnostic accuracy in lung ultrasound interpretation.
• AI assistance democratizes expertise, improving LUS performance for all clinician levels.
• This technology holds promise for improving the reliability of LUS diagnostics in clinical practice.