Eye-tracking to assess visual attention of medical students and experts during lung ultrasounds in critically ill

Hintergrund Lung ultrasound in the critically ill (LUCI) is a complex neuro-motoric procedure. The cognitive process starts with manual control of the probe, followed by visual perception including gaze motion and fixation to screen and sonogram content and finally its assessment. Until now, visual perception could not be observed directly in LUCI. Eye-tracking offers the possibility of analyzing part of this process. We aim to develop a methodology and prototypical analysis of gaze visualization via eye movements and diagnostically relevant areas of interest (dAOIs) and diagnostically irrelevant areas (iAOIs) during LUCI.

Methoden PoCLUS was divided into four alternatives of clinical context and paired with predefined mandatory positions for gaze fixation. Those were for 1) right and 2) left pleural effusion (PE), Lung sliding, diaphragm, spine sign; for 3) pneumothorax (PTX): lung sliding; for 4) congestion: lung sliding and B-lines. The reference group encompassed n=9 experts. Gaze fixations of medical students (n=40) were recorded after 2.5 hours of training with eye-tracking glasses (Pupil NEON, Pupil Labs GmbH, Berlin, Germany) ([Fig. 1]). All gaze fixations were transferred to a reference image and a heatmap of dAOIs was created (Fig. 2.). The examination time (ET) was measured from the time the probe was put onto the skin until it was taken off.

Fig. 1  Heatmap comparison of A1-A4 experts and B1-B4 novices of 4 LUCI variants (A1/B1) right PE, (A2/B2) left PE, (A3/B3) PTX, (A4/B4) congestion. dAOIs do not differ. C1-C4; Fixated diagnostically irrelevant gaze areas (iAOI) 1) Recessus hepato-renalis or (2) R. spleno-renalis instead of diaphragm, (3) lung sliding onto a rib, (4) B-lines at liver level

Ergebnisse The heatmap of dAOIs of experts met the predefined criteria (Fig. 1No image found for uniqueTag: Fig01No image found for uniqueTag: Fig01; A1-4). Medical students similarly focused dAOIs (87.0% vs. 86.9%) but also iAOIs until the target structure was reached. There were longer ETs of medical students (numbers in seconds mean±SD, experts vs. medical students), PE right; 22±9 vs. 49±23, p=0.001; PE left; 21±10 vs. 55±35, p<0.001; PTX; 58±21 vs. 66±23; n.s., B-lines; 66±29 vs. 97±35, p<0.05).

Schlussfolgerung Eye-tracking offers the analysis of visual attention by real-time gaze visualization when learning LUCI. It has the potential to develop quality criteria and to optimize LUCI training.

Publikationsverlauf

Artikel online veröffentlicht:
16. Oktober 2025