Written by Chris Thom
Spoon Feed
Novice ultrasound users were able to acquire diagnostic level lung US (LUS) images at a rate similar to experts.
AI makes lung ultrasound acquisition easy
This study enrolled patients 21 and up with suspected pulmonary edema in 4 different clinical sites (2 EDs, 1 cardiology clinic, and 1 inpatient unit). Each patient underwent two LUS exams, one by a “trained healthcare provider” (THCP) using AI software and one by an ultrasound expert. THCPs included registered nurses and medical assistants with no prior US experience and a small cohort of ultrasound-trained physicians. Both exams followed a standard 8 zone protocol. The AI software provided guidance on image quality and anatomical landmarks, while also providing B-line annotation and auto-saving appropriate clips. Ground truth was the majority determination of the presence or absence of diagnostic imaging quality by a panel of 5 separate, blinded LUS expert readers.
176 participants were recruited across the 4 sites and had a complete set of THCP imaging. Out of these 176, there were 163 who also had complete LUS expert imaging. 98.3% of the LUS studies obtained by THCPs with AI guidance were deemed of diagnostic quality, as compared to 96.6% of studies obtained by US experts, p=0.31. There was one lung zone in the left antero-inferior region where THCPs had a slightly superior rate of adequate imaging quality. The mean time for THCP/AI LUS acquisition was 16.5 minutes.
How will this change my practice?
The identification of B-lines on LUS can be a game-changer in a severely ill patient with undifferentiated dyspnea and is shown to significantly expedite diagnosis (1). AI continues to show promise with B-line identification (2), and software that seeks to quantify B-lines for the user is available from major industry players (3). AI guidance for LUS by novice users is promising. This has the potential to open doors regarding who can acquire these images. However, acquisition time was quite long. Lastly, the additive effect of AI software versus a brief training intervention for the novice user may also need to be explored.
Ultrasound acquisition pro tips
For patients with significant respiratory distress, one or two lung zones are often adequate to identify severe pulmonary edema and direct treatment. Additional lung imaging to cover the anterior and lateral lung fields can be helpful in more subtle cases of pulmonary edema, which is consistent with the current study’s lung zone protocol (4).
One should first seek to recognize the rib shadows and the echogenic pleural line just deep to the ribs. In cases of pulmonary edema, there will be bright vertical lines originating from the pleural line and continuing deep in the imaging field. An increasing number of these B-lines indicates a more severe degree of pulmonary edema. This is one of the easiest and highest impact US pathologies to identify, so I encourage you to give it a try (5).
Source
Artificial Intelligence-Guided Lung Ultrasound by Nonexperts. JAMA Cardiol. 2025 Jan 15. doi: 10.1001/jamacardio.2024.4991. Epub ahead of print. PMID: 39813064
Additional References
- Pivetta E, Goffi A, Nazerian P, et al. Lung ultrasound integrated with clinical assessment for the diagnosis of acute decompensated heart failure in the emergency department: a randomized controlled trial. Eur J Heart Fail. 2019;21(6):754-766.
- Russell FM, Ehrman RR, Barton A, Sarmiento E, Ottenhoff JE, Nti BK. B-line quantification: comparing learners novice to lung ultrasound assisted by machine artificial intelligence technology to expert review. Ultrasound J. 2021;13(1):33. Published 2021 Jun 30.
- Introducing the Auto-B Lines Tool on Venue Family POCUS. 2021.
- Volpicelli G, Mussa A, Garofalo G, et al. Bedside lung ultrasound in the assessment of alveolar-interstitial syndrome. Am J Emerg Med. 2006;24(6):689-696.
- Martindale JL, Noble VE, Liteplo A. Diagnosing pulmonary edema: lung ultrasound versus chest radiography. Eur J Emerg Med. 2013;20(5):356-360.
