Machine Learning Improves the Detection of Misplaced V1 and V2 Electrodes During 12-Lead Electrocardiogram Acquisition

Electrode misplacement during 12-lead
Electrocardiogram (ECG) acquisition can adversely cause
false ECG interpretation, diagnosis and subsequent
incorrect clinical treatment or lack thereof. A common
misplacement errors are the. superior placement of V1
and V2 electrodes. The analysis of ECG signals that were
recorded from ECGs with vertically misplaced leads V1
and V2 can yield a false diagnosis of Brugada syndrome,
myocardial infarction (MI) or left ventricular hypertrophy
(LVH). The aim of the current research was to detect lead
V1 and V2 misplacement using feature engineered
machine learning algorithms to enhance ECG data quality
to improve clinical decision making in cardiac care. In
this particular study, we reasonably assume that V1 and
V2 are concurrently superiorly misplaced together. ECGs
for 450 patients, (normal n=150, LVH n=150, MI n=150)
were extracted from body surface potential maps. ECG
signals were extracted using correct and incorrectly
placed V1 and V2 electrodes, i.e. leads derived from the
fourth intercostal space (ICS) as well as the first ICS,
second ICS, and third ICS. The prevalence for correct and
incorrect leads were 50%. Sixteen features were extracted
including: morphological, statistical and time-frequency
features. Two feature selection approaches (filter method
and wrapper method) were applied to find an optimal set
of features that provide a high accuracy when used with a
machine learning model. To ensure accuracy, six
classifiers were applied including: fine tree, coarse tree,
bagged tree, Linear Support Vector Machine (LSVM),
Quadratic Support Vector Machine (QSVM) and logistic
regression. The accuracy of V1 and V2 misplacement
detection was 94.3% in the first ICS, 92.7% in the second
ICS and 70% in third ICS respectively. Based on
accuracy results, bagged tree was the best classifier in the
first, second and third ICS to detect V1 and V2
misplacement.