Now you see it. Now you don’t. Visual perceptual learning occurs during ultrasound training but is quickly lost without continued experience.

Abstract

Purpose : Ultrasound (US) training during medical school engages students in visual perceptual learning as they begin to identify anatomic structures based on associated features and patterns. Accurate processing of the “gist” of a complex visual scene (including US images) occurs in a sub-second timescale (150-200 msec) provided the observer is familiar with the scene or object being observed. The purpose of this study was to determine whether one year of US training enables 2^nd-year osteopathic medical students (OMS-II) to correctly identify anatomic structures in US images presented for just 200 msec.

Methods : OMS-II and US-naïve pre-medical students (PMs) were tested using a PowerPoint presentation containing five conventional US images, one 3-D reconstructed US image (fetal face), and four images of commonplace objects as controls (ex: book). When ready, participants left clicked while gazing at a fixation cross, initiating the following sequence: 1) blank screen (500 msec), 2) image (US or object; 200 msec), 3) a multiple-choice question asking the participant what they had just seen. As a control, each US image was presented at random twice throughout the experiment. Responses were only marked correct if participants accurately identified the anatomic structure both times. As an additional control, only participants that correctly identified the fetal face and control objects were included in our analysis. All OMS-II participants were tested six months after the completion of their formal US training. Prior to the task, OMS-II participants indicated whether they engaged in extracurricular US experiences during this six-month window.

Results : OMS-IIs (n = 9) with one year of formal US training performed better than US-naïve PMs (n = 4) on this task (48.3% vs 10.0%; respectively). However, splitting the OMS-II group based on whether they pursued extracurricular US experience, resulted in a significant difference in task performance (90.0% ± 5.8% “yes”, vs 27.5% ± 6.6% “no”; p < 0.0001).

Conclusions : Our preliminary results suggest that one year of US training during medical school enables rapid, automatic visual processing of US images. However, this ability wanes significantly within six months unless students continue to engage with US. Future studies will be necessary to determine how quickly this skill is lost following formal training.

This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.