Detection of Unicolor ECG Electrode Reversals in Standard 12-Lead ECG

Abstract

This paper presents the performance of a commercial lead quality monitoring library (LQMLib, Schiller AG) for detection of reversals between 12-lead ECG cables with matching colors and proposes methods for improvements, where necessary. The study is performed on a large 12-lead ECG database with 1331 chest pain patients (646 training, 685 test recordings) and ECGs from 29 volunteers with swapped red, yellow, green, black and all unicolor cables. Relying on assessment of inter-lead correlations over continuous 4s episodes, LQMLib without adjustments achieves Se>97% for all unicolor cable reversals, except RC1 (Se=29%) and NC5 (Se=88%). The proposed additional analyses improve RC1 and NC5 swap detection by 63% points and 7% points, respectively.

A lightweight deep learning approach for detecting electrocardiographic lead misplacement

“…In Rjoob et al (2019aRjoob et al ( , 2019b, authors conducted V1 and V2 misplacement detection in three different positions: first, second and third intercostal spaces, which was based on morphological, frequency, and statistical features. Lead correlation features calculate correlation coefficients between leads and use machine learning models to detect misplacement based on coefficient trends (Jekova et al 2013, Jekova et al 2016, Jekova et al 2018. Lead redundancy information utilized the fact that limb leads can be reconstructed through any two limb leads, and changes in the correlation coefficients of the reconstructed leads can be used to detect misplacement (Kors and van Herpen 2001).…”

Simulating Arbitrary Electrode Reversals in Standard 12-Lead ECG

“…Generally, when another precordial lead is substituted for V1, the result is a tall R wave in V1, which could be taken as a sign of right bundle branch block, left ventricular ectopy, right ventricular hypertrophy, acute right ventricular dilation, Type A Wolff-Parkinson-White syndrome, posterior MI, hypertrophic cardiomyopathy, progressive muscular dystrophy or dextrocardia [10]. Reversals between limb and chest electrodes are a possible scenario due to the matching colors of the two ECG cables [11] or the incorrect attachment of the cable connectors to the junction box of the ECG machine [12]. C2/LA (yellow) cable interchange is described in two case reports [13,14] to have produced right axis deviation and Q waves in (III, aVF), accompanied by an inverted T wave in both leads, together with a quick transition in V2 with qR complex and an inverted T wave.…”

“…Reversals between limb and chest electrodes are a possible scenario due to the matching colors of the two ECG cables [11] or the incorrect attachment of the cable connectors to the junction box of the ECG machine [12]. C2/LA (yellow) cable interchange is described in two case reports [13,14] to have produced right axis deviation and Q waves in (III, aVF), accompanied by an inverted T wave in both leads, together with a quick transition in V2 with qR complex and an inverted T wave.…”

“… Limb and chest leads: Interchanges between limb and C2 precordial electrodes specific for a telemonitoring system are detected by correlation to a previously recorded ECG [39]. This early work, together with our recent publication on the unicolor electrode interchange detection [11], are the only studies dealing with recognition of reversals between limb and precordial leads.…”

“…Only a few of the mentioned studies [7,19,27,35,39,40] use real ECG recordings with erroneous electrode placements, which are, however, small-sized and proprietary. Typically, reversal detection algorithms are trained and validated using databases with correctly recorded 12-lead ECGs and simulated reversals within the limb lead set [11,26,28,29,30,31,32,33,34,36,38] or the precordial lead set [11,26,31,33,34,38], where the Wilson’s central terminal (WCT) is not changed. All other reversals modifying the WCT, such as swaps between the limb and precordial electrodes, have not been simulated, although they are quite possible and should be detected due to the distorted morphology of most leads [6,8,21].…”