Acquisition of Cardiac Point-of-Care Ultrasound Images With Deep Learning: A Randomized Trial for Educational...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Medicine
• Point-of-Care Ultrasonography

Background:

• Point-of-care ultrasonography (POCUS) may integrate artificial intelligence (AI) for real-time image enhancement.
• The effect of AI on POCUS learning remains unclear.

Purpose of the Study:

• To investigate if AI-enhanced POCUS devices improve cardiac image acquisition and interpretation in novices.
• To assess the impact of deep learning functionality on POCUS learning outcomes.

Main Methods:

• Internal medicine trainees (N=43) with limited POCUS experience were randomized to AI-enhanced or standard POCUS devices for two weeks.
• The primary outcome was the time to acquire an apical four-chamber (A4C) view on a standardized patient.
• Secondary outcomes included A4C image quality, image quiz performance, and device usage.

Main Results:

• The AI group demonstrated significantly faster A4C scan times (57s vs 85s) and higher image quality scores (4.5 vs 2) at follow-up.
• AI-assisted novices were more accurate in identifying reduced systolic function (85% vs 50%) compared to the non-AI group.
• The AI group utilized the POCUS devices more frequently (median 5.5 vs 2 times).

Conclusions:

• POCUS devices incorporating deep learning AI show potential to enhance cardiac image acquisition and interpretation skills among novice users.
• Further research is warranted to fully elucidate the long-term impact of AI on POCUS education and proficiency.