Two- Versus 8-Zone Lung Ultrasound in Heart Failure: Analysis of a Large Data Set Using a Deep Learning Algorithm

Area of Science:

• Medical Imaging
• Cardiology
• Pulmonology

Background:

• Lung ultrasound (LUS) is crucial for diagnosing heart failure.
• Standard LUS protocols use 8 or more zones, potentially limiting clinical efficiency.
• Deep learning (DL) algorithms can analyze LUS for B-line artifacts.

Purpose of the Study:

• To compare a simplified 2-zone LUS protocol with the standard 8-zone protocol.
• To evaluate the efficacy of a DL algorithm in assessing B-line severity across different LUS protocols.
• To determine if a 2-zone protocol is sufficient for heart failure assessment.

Main Methods:

• Prospective observational study of adult patients with suspected heart failure.
• Daily LUS using an 8-zone protocol and a hand-held ultrasound system.
• A DL algorithm adapted to rate B-line severity (0-4 scale) on video loops.
• Bland-Altman analysis to compare 2-zone and 8-zone DL-derived severity scores.

Main Results:

• Excellent agreement between 2-zone and 8-zone protocols in Bland-Altman analyses.
• No significant difference in average B-line severity scores between the two protocols.
• Primary dataset (n=205) showed a mean difference of 0.03; subgroup (n=147) showed 0.01.

Conclusions:

• A 2-zone anterior-superior thoracic ultrasound protocol provides comparable B-line severity information to an 8-zone approach.
• This simplified protocol can enhance the clinical utility of lung ultrasound in heart failure management.
• DL-assisted LUS with a 2-zone protocol offers an efficient diagnostic tool.