Cardiac CT-derived Extracellular Volume as an Alternative to Cardiac MRI for Myocardial Tissue Characterisation

Sri Kousthubha Allampalli1, Iulia Nazarov1, Ronak Rajani2, Martin Bishop1, John Whitaker1
1King's College London, 2Guy's and St Thomas' NHS Foundation


Abstract

Background: Extracellular Volume (ECV) is a promising indicator of myocardial structural remodelling. Cardiac magnetic resonance (CMR)-derived ECV's utility has been validated histologically and clinically, however is often contraindicated in patients with implanted metal devices, renal dysfunction or claustrophobia. Cardiac computed tomography (CCT)-derived ECV offers a cheaper, less time-consuming alternative with fewer contraindications, but a direct comparison is required to evaluate its efficacy. Objective: To develop a registration and artefact removal workflow and evaluate CCT-derived ECV against CMR-derived ECV for myocardial tissue characterisation. Methods: Patients undergoing ventricular tachycardia ablation with pre-procedural CCT and CMR imaging were included. Pre- and post-gadolinium T1 maps were segmented to obtain CMR-ECV across a left ventricular (LV) myocardial slice. For CCT-ECV, pre-contrast and late iodine enhanced scans were registered to a CT angiography, automatically segmented for LV ECV quantification. Image Position Patient (IPP) and Image Orientation Patient (IOP) DICOM tags of the T1 maps were used to create a slice plane intersecting the CT LV volume (Figure 1a), registered to an MR volume by centre of mass. Separate artefact removal strategies were developed for implantable cardioverter defibrillator (ICD)-related artefacts in both modalities. Circumferential universal ventricular coordinates were computed using septal insertion points as reference points to split both slices into 6 American Heart Association (AHA) segments for segmental ECV comparison (Figure 1b). Results: Thirteen patients were included. A mean cosine similarity of 0.986+/-0.013 indicated preservation of the pattern of segmental ECV values between modalities. Segmental ECV correlation was modest (r=0.284, p=0.0118) but improved significantly after artefact removal across all patients (r=0.481, p<0.001) and in those with ICD artefact specifically (r=0.617, p<0.001; Figure 1c). Conclusion: CCT-derived ECV shows meaningful agreement with CMR-derived ECV in both spatial distribution and magnitude. Artefact removal achieved a significant improvement in inter-modality concordance, though further evaluation in larger cohorts is warranted.