# Predictors of late gadolinium enhancement cardiac MRI image quality in patients with cardiac implantable electronic devices

## TL;DR

This study identifies factors affecting the quality of heart MRI scans in patients with cardiac devices, aiming to improve patient selection for diagnostic accuracy.

## Contribution

The study introduces generator-to-lead distance thresholds by device type to optimize LGE-CMR referrals for patients with CIEDs.

## Key findings

- ICD/CRT-D patients had 36.5% non-diagnostic scans, while PM/CRT-P patients had none.
- Generator-to-lead distance thresholds of 10 cm for ICDs and 8 cm for PMs predict acceptable LGE quality.

## Abstract

Cardiac implantable electronic devices (CIEDs) can cause artifacts in late gadolinium enhancement (LGE) cardiac magnetic resonance imaging (CMR), compromising diagnostic accuracy. No consensus exists on optimal CIED patient selection for LGE-CMR.

This study aims to identify predictors of LGE image quality in patients with CIEDS to optimize pre-scan selection.

Patients with CIEDs who underwent conventional 2D-LGE imaging were retrospectively identified from the Amsterdam UMC CMR database. Baseline clinical and device characteristics were collected, and generator-to-lead distance was measured on post-implantation chest X-rays. LGE quality was categorized as fully diagnostic, acceptable, or non-diagnostic. Multivariable regression and receiver operating characteristic (ROC) analysis determined independent predictors and exploratory generator-to-lead distance thresholds using a 90% sensitivity criterion.

Overall, 80 patients (71.3% male, mean age 64 years) were included: 41.3% ICDs, 23.8% pacemakers (PMs), 23.8% cardiac resynchronization therapy defibrillators (CRT-Ds), and 2.5% cardiac resynchronization therapy pacemakers (CRT-Ps). LGE image quality was fully diagnostic in 48.8%, acceptable in 27.5%, and non-diagnostic in 23.8% of patients. PM/CRT-P patients had no non-diagnostic scans (92.9% fully diagnostic). Only 25.0% of ICD/CRT-D scans were fully diagnostic, while 36.5% were non-diagnostic (P < .001). Generator-to-lead distance was significantly associated with LGE quality, with thresholds of 10 cm in ICDs and 8 cm in PMs for acceptable LGE quality.

Device type and positioning significantly impact LGE image quality. ICDs were associated with poorer image quality, while PMs consistently yielded diagnostic-quality images. Generator-to-lead distance emerged as a key predictor, providing a practical tool for optimizing LGE-CMR referrals. This study defines generator-to-lead distance thresholds by device type and proposes a structured pre-scan workflow to support LGE-CMR referral decisions in patients with CIEDs.