CMR-Analysis (including machine learning)

1350132 - Quantification of dyssynchrony requires identification of mechanical activation and cannot rely on electrocardiogram triggering

Friday, January 27, 2023

12:20 PM – 1:20 PM

Location: Pacific Jewel

Presenting Author(s)

DL

Daniel E. Loewenstein, MD

PhD student
Karolinska Institute
Stockholm, Stockholms Lan, Sweden

Primary Author(s)

DL

Daniel E. Loewenstein, MD

PhD student
Karolinska Institute
Stockholm, Stockholms Lan, Sweden

Co-Author(s)

BW

Björn Wieslander, MD, PhD

MD PhD
Karolinska Institutet
Stockholm, Stockholms Lan, Sweden
PS

Pavol Sareny, MD

MD
Malacky Hospital, Slovakia
IK

Igor Klem, MD

Md
Duke University
Durham, North Carolina, United States
ES

Erik B. Schelbert, MD

Cardiologist
Minneapolis Heart Institute at Abbott Northwestern Hospital
Saint Paul, Minnesota, United States
RJ

Robert Jablonowski, MD, PhD

MD, PhD
Lund University
Lund, Sweden
Henrik Engblom, MD, PhD

Professor
Lund University
Lund, Skane Lan, Sweden
BA

Brett D. Atwater, MD

Director of Electrophysiology
Inova Fairfax Hospital, North Carolina, United States
LB

Ljuba Bacharova, MD, MBA

senior researcher / MD, DSc, MBA
International Laser Center CVTI
Bratislava, Bratislava, Slovakia
Martin Ugander, MD, PhD

Professor of Cardiac Imaging
University of Sydney

Background: Parameters used for quantifying mechanical dyssynchrony in cardiac imaging often utilize an electrocardiogram (ECG) lead as a temporal reference for identifying the start of ventricular electrical activation as a surrogate for the onset of mechanical activation. However, both the maximum QRS amplitude and the onset of the QRS can vary between different leads, in particular in left bundle branch block (LBBB). We hypothesized that in LBBB, the magnitude of inter-lead differences in QRS onset or maximum QRS amplitude could be large enough to introduce clinically meaningful errors in the quantification of mechanical dyssynchrony.

Methods: Patients (n=308) with an ECG demonstrating LBBB by Strauss’ criteria that had undergone CMR were included from four centers. Mechanical dyssynchrony analysis derived from CMR cine images was performed with the reference timeframe determined by ECG triggering (image timeframe 1), or by visual identification of the start of mechanical activation in short-axis and long-axis views regardless of timeframe in relation to the ECG trigger. ECGs were analyzed for differences between global QRS onset and both individual lead QRS onset and maximum QRS amplitude in all 12 leads.

Results: In mechanical dyssynchrony analysis of CMR cine images with a temporal resolution of (median [interquartile range]) 31 [27–37] ms, the reference timeframe identifying the start of mechanical activation occurred 56 [39–71] ms before the first timeframe identified by ECG triggering. Group differences between the global QRS onset and individual lead QRS onset varied between the lowest for V3 (0 [0–3] ms) and highest in aVF (8 [2–15] ms). Intra-subject lead-to-lead differences between in QRS onset varied between the lowest between V2 and V3 (0 [0–3] ms) and highest between aVF and aVL (8 [3–15] ms). Group differences between the global QRS onset and maximum QRS amplitude varied between the lowest for V1 (63 [57–69] ms) and highest in III (99 [84–112] ms). Intra-subject lead-to-lead differences in QRS onset varied between the lowest between aVF and III (2 [0–20] ms) and highest between III and V1 (38 [22–48] ms).

Conclusion: In cardiac imaging of LBBB, electrocardiogram triggering on the maximum QRS amplitude can introduce clinically meaningful errors in the quantification of mechanical dyssynchrony. Quantification of dyssynchrony in LBBB should be based on the identification of mechanical activation as a temporal reference.