Assessing computerized eye tracking technology for gaining insight into expert interpretation of the 12-lead electrocardiogram: an objective quantitative approach

摘要: Abstract Introduction It is well known that accurate interpretation of the 12-lead electrocardiogram (ECG) requires a high degree skill. There also moderate variability among those who interpret ECG. While this case, there are no best practice guidelines for actual ECG process. Hence, study adopts computerized eye tracking technology to investigate whether eye-gaze can be used gain deeper insight into how expert annotators Annotators were recruited in San Jose, California at 2013 International Society Computerised Electrocardiology (ISCE). Methods Each annotator was number ECGs (N = 12) while their gaze recorded using Tobii X60 tracker. The device based on corneal reflection and non-intrusive. With sampling rate 60 Hz, coordinates acquired every 16.7 ms. Fixations determined predefined classification algorithm, which then generate heat maps where looked. form four groups (3 = ST elevation myocardial infarction [STEMI], 3 = hypertrophy, 3 = arrhythmias 3 = exhibiting unique artefacts). an equal distribution difficulty levels (3 = easy interpret, 3 = average 3 = difficult). displayed 4x3 + 1 display format annotations concealed. Results Precisely 252 interpretations (21 annotators × 12 ECGs) recorded. Average duration 58 s (SD = 23). Fleiss' generalized kappa coefficient (Pa = 0.56) indicated inter-rater reliability annotators. 79% agreement STEMI cases, 71% arrhythmia 65% lead misplacement dextrocardia cases only 37% hypertrophy cases. In analyzing total fixation duration, it found average V1 most (4.29 s), followed by leads V2 (3.83 s), rhythm strip (3.47 s), II (2.74 s), V3 (2.63 s), I (2.53 s), aVL (2.45 s), V5 (2.27 s), aVF (1.74 s), aVR (1.63 s), V6 (1.39 s), III (1.32 s) V4 (1.19 s). spends amount time studying frontal plane (15.89 s) when compared transverse (15.70 s). fixated first V2, aVL, V1, II, aVR, V3, strip, III, aVF, V5, V6. We strong correlation (r = 0.67) between each lead. This indicates studied longest. weak negative accuracy (r = − 0.2) age (r = 0.67). Conclusions Eye facilitated As result, authors recommend adopt initial impression/pattern recognition approach conventional systematic protocol interpretation. recommendation observing misdiagnoses given due impression only. summary, research presents results from provides scope future work involves exploiting further science