Coronary Artery Occlusion Detection Using 3-Lead ECG System Suitable for Credit Card-Size Personal Device Integration

Background Early coronary occlusion detection by portable personal device with limited number of electrocardiographic (ECG) leads might shorten symptom-to-balloon time in acute coronary syndromes. Objectives The purpose of this study was to compare the accuracy of coronary occlusion detection using vectorcardgiographic analysis of a near-orthogonal 3-lead ECG configuration suitable for credit card-size personal device integration with automated and human 12 lead ECG interpretation. Methods The 12-lead ECGs with 3 additional leads (“abc”) using 2 arm and 2 left parasternal electrodes were recorded in 66 patients undergoing percutaneous coronary intervention prior to (“baseline”, n = 66), immediately before (“preinflation”, n = 66), and after 90-second balloon coronary occlusion (“inflation”, n = 120). Performance of computer-measured ST-segment shift on vectorcardgiographic loops constructed from “abc” and 12 leads, standard 12-lead ECG, and consensus human interpretation in coronary occlusion detection were compared in “comparative” and “spot” modes (with/without reference to “baseline”) using areas under ROC curves (AUC), reliability, and sensitivity/specificity analysis. Results Comparative “abc”-derived ST-segment shift was similar to two 12-lead methods (vector/traditional) in detecting balloon coronary occlusion (AUC = 0.95, 0.96, and 0.97, respectively, P = NS). Spot “abc” and 12-lead measurements (AUC = 0.72, 0.77, 0.68, respectively, P = NS) demonstrated poorer performance (P < 0.01 vs comparative measurements). Reliability analysis demonstrated comparative automated measurements in “good” agreement with reference (preinflation/inflation), while comparative human interpretation was in “moderate” range. Spot automated and human reading showed “poor” agreement. Conclusions Vectorcardiographic ST-segment analysis using baseline comparison of 3-lead ECG system suitable for credit card-size personal device integration is similar to established 12-lead ECG methods in detecting balloon coronary occlusion.

2][3] Prehospital delays constitute the major contributor to "symptoms to balloon" time with median symptoms to call time over 2.5 hours, by far exceeding in-hospital delays. 4Remote transmission of a prehospital ECG allows early identification of STEMI patients requiring emergent revascularization, which significantly reduces symptoms-to-balloon time and mortality 5 and is a Class I recommendation for STEMI management. 6ehospital 12-lead ECG recording requires large, nonportable, specialized equipment and trained personnel and may not be sufficiently efficient to minimize delays in the delivery of reperfusion.A single ECG tracing at one moment in time has limited accuracy in ischemia detection, while comparison of an ECG to a prior baseline ECG (comparative ECG analysis) significantly improves its accuracy. 7tempts have been made to diagnose acute myocardial ischemia by using personal devices that do not require professional assistance with the goal to shorten prehospital delays in patients with acute coronary syndromes.A "multilead ECG equivalents" using sequential single lead recordings by portable devices (AliveCor and Apple Watch) were capable of myocardial infarction detection in emergency room (ER) patients. 8,9An implantable intracardiac STsegment monitoring device with patient alert system decreased time to presentation in patients with acute coronary syndromes without increasing the number of unnecessary ER visits. 10chemia detection based upon a limited number of simultaneously recorded leads using vectorcardgiographic (VCG) analysis 11,12 or a 12-lead reconstruction 13 has been demonstrated to be feasible.A portable device with 3 wired electrodes suitable for self-application was superior to the standard 12-lead ECG criteria to detect myocardial ischemia using a set of asymptomatic recordings as baseline 14 The presence of wires with disposable electrodes limits the utility of such device.ECG leads can be integrated into a portable device without the need for external electrodes/wires 13 which would improve the ease of application and compliance.

OBJECTIVE OF THE STUDY
To validate the accuracy of automated coronary occlusion detection during balloon coronary angioplasty using vectorcardiographic ST-segment analysis of a 3-lead near-orthogonal ECG system ("abc") and compare it to the accuracy of the standard 12-lead ECG interpreted by automated computer analysis and expert human readers.

METHODS
The study was conducted at the Clinical Center of Serbia, Belgrade, Serbia, and was approved by the institutional Clinical Ethics Committee.All patients provided written consent.Electrodes A, B, C, D were designed to produce near-orthogonal leads "a", "b", "c" corresponding to the lead configuration that will be implemented in a portable device with integrated electrodes where right and left hand fingers produce lead "a" resembling the standard lead I similar to the commercially available ECG rhythm recorders, and together with the chest leads contribute to the vertical ("b") and sagittal ("c") leads (Figure 1B).Because of the positional constraints during PCI procedure, fingertip electrodes to record lead "a" were substituted by electrodes A and B on the shoulders equivalent to standard lead I. Waveforms produced by this location of limb lead placement are equivalent to the classic wrist lead position. 15e following 10-second recording segments were used for analysis: 1. "Baseline" segment (n ¼ 66): obtained in the cath lab holding area before the procedure; ECG ANALYSIS.The initial portion of the ST-segment (starting at J point þ 10 ms) was used to detect coronary artery occlusion (see Supplemental Methods for details).ST-segment analysis of the "abc" leads was performed using VCG approach in a manner similar to previously described in the literature 7 treating leads "a", "b", and "c" as orthogonal ("a" ¼ "x", "b" ¼ "z", "c" ¼ "y") and compared to the standard 12-lead ECG analyzed in 4 different ways.a) vectorcardiographic approach using leads "x", "y", "z" obtained by the Kors regression transformation of the 12-lead ECG 16  All analyses except standard ischemia definition were performed in "spot" (single tracing) and "comparative" (comparison with "baseline" tracing) modes.
V e c t o r c a r d i o g r a p h i c m e a s u r e m e n t .Median QRST complex and QRST vector loops were constructed from 10-second signal segments using "abc" and 12-lead-derived "xyz" leads.Fiducial points were determined automatically by the custom-built software and confirmed by visual analysis.
The following measurements were constructed: 1. S_abc was calculated as the 3-dimensional abcderived ST-segment vector loop segment deviation from isoelectric point averaged across the segment duration and measured in mV ("spot" measurement).
Spot and comparative "xyz" measurements (S_xyz and C_xyz, respectively) were calculated in a similar manner using "xyz"-derived QRST vector loops.
The duration of the ST-segment portion used for all vector measurements was heart rate-corrected using linear dependence on the RR interval to minimize potential confounding effects of heart rate differences 18 ; see Supplemental Methods for details. and "baseline" ECGs presented in Supplemental Figure 1).Three separate readings were performed by each of the 3 readers R1 to R3.
2. "Human spot 2 (HS2)": Repeat interpretation in the same fashion performed 2 to 3 months later to assess for intraobserver variability.
All 12-lead ECG sets were presented to the readers in random order different for each reading.1][22] Two-sided P value of <0.05 was considered significant.All computations were performed using SPSS ver.20,IBM and MatLab ver.2016b, MathWork).

RESULTS
BASELINE CHARACTERISTICS.A total of 66 patients were enrolled in the study.Patients' clinical characteristics are presented in Table 1.The majority of patients (70%) were smokers, and 16 (24%) had prior myocardial infarctions.Only 15 out of 66 patients (23%) had "baseline" ECG interpreted as "normal"; 28 (42%) had ST-segment shift >0.5 mm in at least one of the leads.
BALLOON OCCLUSIONS.Balloon occlusion distribution and duration are presented in Table 2.There were 120 balloon occlusions performed.A total of 33 (50%) patients had angioplasty of more than 1 artery (range 1-6, median 2 per patient) (Supplemental Table 1).There was no difference in occlusion duration across the coronary arteries.
Balloon inflations resulted in significant STsegment shift calculated by all measurements (Table 3).LAD inflations caused larger ST-segment changes compared to the LCX and RCA inflations.

Portable 3 Lead ECG for Coronary Occlusion Detection
There were no significant differences in ST-segment shift observed between inflations in the proximal, mid-, and distal segments of any of the arteries by all methods of measurement.
OCCLUSION DETECTION PERFORMANCE OF THE AUTOMATED MEASUREMENTS.The performance of the automated measurements was evaluated using ROC curves.Areas under the ROC curves were compared, and optimal threshold points were determined (Table 3, Figure 3).Comparative measurements -C_abc, C_xyz, and C_12L (AUC ¼ 0.948; 0.960; 0.965, respectively, P ¼ NS) demonstrated near-uniform performance regardless of the artery, within-artery location, or presence of baseline ECG abnormalities with no significant differences between the measurements in any subset of tracings (Supplemental Table 3).AUC between comparative measurements were not different (P ¼ NS) while significantly higher than AUC of spot measurements (P < 0.05).
Triangles represent sensitivity/specificity data for consensus human readings (HCS1, HCS2, HCC); black square represents standard ischemia definition; solid lines represent comparative measurements; and dashed/dotted lines represent spot measurements.AUC ¼ areas under the

ROC curves
Shvilkin et al

Portable 3 Lead ECG for Coronary Occlusion Detection
A U G U S T 2 0 2 3 : 1 0 0 4 5 4 Spot measurements demonstrated much poorer performance (P < 0.05 compared to comparative ones) with no statistical differences between the individual measurements, S_abc, S_xyz, and S_12L (AUC 0.722; 0.766; 0.683, respectively, P ¼ NS).The LAD distribution conferred numerically better, while LCX and the presence of the baseline ST-segment shiftnumerically worse measurement performance, but the AUC differences did not reach statistical significance (Supplement Table 4).demonstrated even stronger correlations (r ¼ 0.96-0.97)between themselves (Supplemental Figure 3).

CORRELATION OF COMPARATIVE
All automated "abc" and 12-lead measurements (both comparative and spot) correctly classified all tracings meeting standard STEMI and acute ischemia criteria (Supplemental Table 5).
To summarize, the 3-lead "abc" derived measurements performed similarly to the 2 12-lead measurements, "xyz" and "12L" (constructed with and   The advantage of the "abc" electrode configuration is that it can be incorporated into a credit card-size device without separate leads/wires with the limb lead "a" provided by direct finger to device contact improving its ease of use. The information content of the 3 "abc" leads was comparable to that of the 12-lead ECG for the purpose of coronary occlusion detection based on the following findings: 1) there was no statistically significant difference between ROC curves of "abc" and 12-lead ECG-based comparative measurements; 2) coronary occlusion results in numerically similar changes in "abc"and 12-lead-derived measurements that remain consistent across the occlusion locations and demonstrated close correlations; 3) binary decisions by "abc"and 12-lead-derived comparative measurements demonstrate 95% concordance (eg, same true/false positive/negative cases) with no significant differences in validity analyses; 4) none of the measurements missed any tracings matching the standard ischemia definition.ADDRESS FOR CORRESPONDENCE: Dr Alexei Shvilkin, BIDMC Baker4/Cardiology, 185 Pilgrim Road, Boston, Massachusetts 02215, USA.E-mail: ashvilki@ bidmc.harvard.edu.@CMichaelGibson.

PATIENTS.
Consecutive patients (n ¼ 66, men 79%, age 55 AE 9 years) with angiographically proven coronary artery disease (>70% diameter coronary artery stenosis) suitable for percutaneous coronary intervention (PCI) were enrolled.Patients were excluded if they had a pacemaker or implantable cardioverterdefibrillator, a STEMI within the past 7 days, left bundle branch block, allergic reaction to skin electrodes, or other conditions precluding completion of the study protocol.P C I p r o c e d u r e d e s c r i p t i o n .Planned PCI was performed according to the standard Clinical Center of Serbia protocol.Transient coronary artery occlusion with the balloon catheter as part of the PCI procedure for a target duration of 90 seconds was used as a model of myocardial ischemia.Occlusion duration was shortened to the minimum of 60 seconds if patients developed severe chest pain, hemodynamic instability, or ventricular arrhythmias.When inflations were performed in more than one location, sufficient time was allowed between inflations to allow the ECG to return to baseline based on visual analysis.All patients were hemodynamically stable and chest pain-free prior to the catheterization.ELECTROCARDIOGRAPHIC RECORDINGS.Standard 12-lead ECG with additional 4 electrodes in positions "A", "B", "C", and "D" (Figure 1A) were digitally recorded continuously throughout the procedure using self-adhesive radiolucent wet electrodes (Quinton) and a 15-lead ECG recorder with standard specifications (500 Hz sampling rate, 0.05/150 Hz bandpass, and 50 Hz notch filters).

A 3 Portable 3
B B R E V I A T I O N S A N D A C R O N Y M S ECG = electrocardiogram ER = emergency room LAD = left anterior descending coronary artery LCX = left circumflex coronary artery PCI = percutaneous coronary intervention RCA = right coronary artery STEMI = ST-segment elevation myocardial infarction VCG = vectorcardgiographic Shvilkin et al J A C C : A D V A N C E S , V O L . 2 , N O .6 , 2 0 2 Lead ECG for Coronary Occlusion Detection A U G U S T 2 0 2 3 : 1 0 0 4 5 4
the standard ischemia definition17 ; d) consensus expert opinion.

FIGURE 1 Portable 3 Figure 2
FIGURE 1The "abc" Electrode System Electrode Positions and Electrical Schematic

1 2 -
l e a d E C G m e a s u r e m e n t s .Spot and comparative 12-lead-derived measurements S_12L and C_12L were calculated as the average ST-segment shift with segment duration and heart rate correction as above on the ECG lead with maximal ST-segment deviation with and without baseline subtraction, respectively.Standard ischemia (ST-segment elevation and STsegment depression) and STEMI (ST-segment elevation only) criteria were calculated based on the Fourth Universal Definition of Myocardial Infarction definition.17

HUMAN READING OF THE 12 -
LEAD ECG.Human 12-lead ECG interpretation was performed by 3 boardcertified cardiologists (1 interventionalist, 2 electrophysiologists), designated as R1 to R3.The readers were not required to measure whether ST-segment shift conforms to the standard criteria for acute ischemia but rather deliver their expert opinion on whether the acute ischemia is present (Yes/No).Isolated T-wave inversions without concomitant STsegment shift were not considered signs of ischemia.

For
Figure 2.

FIGURE 2 3 Portable 3
FIGURE 2 Calculation of "abc" Measurements C_abc and S_abc in a Patient With Proximal LAD Balloon Occlusion

J
A C C : A D V A N C E S , V O L . 2 , N O .6 "ABC" AND 12-LEAD MEASUREMENTS.Numeric values of comparative "abc" lead-derived measurement C_abc demonstrated strong correlation with the 12-leadderived comparative measurements (C_xyz, C_12L) in all arteries (r ¼ 0.77-0.94)with 95% of readings being concordant (True/True, False/False), located in left lower/right upper quadrant marked by the cutoff threshold lines in Figure 4. Discordant readings were located close to the cutoff thresholds.As expected, the 12-lead-derived measurements C_xyz and C_12L

FIGURE 5 " 3 Portable 3
FIGURE 5 "abc" Measurements and Human Reader Performance by the Occlusion Location ROC Curves

Portable 3
Lead ECG for Coronary Occlusion Detection lead performance in real world and could have artificially decreased the human readers' performance.Balloon inflation was arbitrarily considered as a positive reference for ischemia.It is likely that some of the inflations did not produce ischemic ECG changes due to the presence of extensive collaterals, small area of ischemia, or pre-existing infarction in the artery distribution.All patients had known coronary heart disease, and ST-segment deviation was considered specific sign of ischemia.In unselected populations, there are multiple causes of nonischemic ST-segment changes that can result in false-positive ischemia detection especially in low prevalence groups.24,27There were no patients with left bundle branch block in the study; the performance of "abc" lead measurements in this population is unknown.In this study, we did not include patients with ambulatory asymptomatic recordings to assess the rate of false-positive results in real life.CONCLUSIONSVectorcardiographic analysis of 3-lead "abc" ECG system detects coronary artery occlusion during balloon inflation, similar to the traditional 12-lead ECG.More research is needed to determine whether this method may be useful in the detection of myocardial ischemia in clinical settings.FUNDING SUPPORT AND AUTHOR DISCLOSURES Drs Shvilkin, Vuk cevi c, Vukajlovi c, Bojovi c, Had zievski, and Atanasoskiare are stockholders of HeartBeam, Inc. Drs Mileti c, Zimetbaum, and Gibson have received consulting fees from HeartBeam, Inc. Dr Vajdic is CEO, HeartBeam, Inc.

TABLE 3
ST-Segment Measurements in "Preinflation" and "Inflation" Tracings by the Values are mean AE SEM. a P < 0.05 "Preinflation" vs "Inflation", all measurements.b P < 0.050 LAD vs LCX þ RCA, all measurements.LAD ¼ left anterior descending coronary artery; LCX ¼ left circumflex coronary artery; RCA ¼ right coronary artery; SEM ¼ standard error of the mean.

TABLE 5
Inter observer and Intra observer Reliability of Human Readings Values are n (%) unless otherwise indicated.HC ¼ human comparative; HS ¼ human spot.