The use of clinician-conducted ultrasound examinations has significantly increased in recent decades. While it was initially adopted by only a small number of physicians, it has since become a standard practice in emergency medicine, both in academic and community settings. The FAST protocol, which stands for Focused Assessment with Sonography in Trauma, has shown promise in expediting patient triage, reducing medical expenses, and shortening hospital stays. However, past studies have indicated that the protocol requires extensive training. Now, new research suggests that artificial intelligence (AI) can assist non-physicians in acquiring ultrasound images of diagnostic quality in the emergency department.

In a recent study, researchers at Kaohsiung Chang Gung Memorial Hospital (Kaohsiung, Taiwan) investigating the potential of using AI to assist novices in clinician-performed ultrasound examination found positive initial results. The study involved 30 novice operators with no prior sonography experience, split equally among registered nurses, nurse practitioners, and EMTs. The participants were randomly assigned to either receive AI assistance or not, using an application that captures ultrasound images and provides real-time feedback through deep learning-based guidance. The operators were instructed to obtain images of the body’s Morrison pouch, and three expert echocardiographers evaluated the images on a scale of 1-5. The researchers observed that AI assistance led to higher scores and rates of acceptable image quality than the group that did not receive AI support.

“The diagnostic quality score and the rate of acceptable clips were significantly higher with AI guidance,” corresponding author Chi-Yung Cheng, MD, with the Department of Emergency Medicine, Kaohsiung Chang Gung Memorial Hospital, and colleagues concluded. “Although initially it may take longer to complete an examination with AI guidance, it is expected that the learning curve will be lower for novices practicing FAST.” 

http://www.allmycbs.com/en/index.asp