Technical Development and Clinical Validation of Novel Cardiovascular Magnetic Resonance Sequences for Simultaneous Coronary Artery Angiography and Vulnerable Plaque Characterisation

Student thesis: Doctoral ThesisDoctor of Philosophy

Abstract

Atherosclerotic coronary artery disease (CAD) is the leading cause of cardiovascular morbidity and mortality worldwide. The early recognition, treatment and longitudinal follow up of patients with CAD could potentially improve long term prognostic outcomes. Coronary magnetic resonance angiography (CMRA) could be a safe, ionising radiation free and noncontrast alternative to invasive coronary angiography (XCA) and coronary computedtomography angiography (CCTA) for the assessment of CAD. However, conventional diaphragmatic navigated (dNAV) 3D whole-heart CMRA frameworks are clinically limited by long, inefficient and unpredictable scan times, image quality degradation due to respiratorymotion artefacts and low spatial resolution.

In this thesis, we describe a novel, highly accelerated, free breathing, image navigated (iNAV), 3D whole-heart CMRA framework that incorporates advanced 2D translational motion correction, 3D non-rigid motion corrected iterative image reconstruction in concert with low-rank patch based denoising, enabling sub-mm isotropic spatial resolution CMRA within a clinically predictable scan time of ≈ 10 minutes with 100% respiratory scan efficiency. We first validated the image quality of this novel iNAV CMRA framework against conventional dNAV CMRA in a cohort of 31 patients with suspected CAD. Average scan time was 9.7 ± 2.3 mins (iNAV CMRA) and 20.9 ± 6.2 mins (dNAV CMRA) (P<0.001), with 100% and 46% respiratory scan efficiency respectively (P<0.001). Visible vessel length was significantly improved with the iNAV CMRA compared with the dNAV CMRA in all three coronary arteries, RCA 12.4 ± 3.7 cm vs. 8.7 ± 4.6 cm (P<0.001); LAD 10.4 ± 3.0 cm vs. 7.1 ± 3.4 cm (P<0.001) and LCx 6.8 ± 2.4 cm vs. 4.8 ± 2.6 cm (P<0.001). Image quality score was significantly improved with the iNAV CMRA for the RCA, LAD, LCx and overall 3D whole-heart dataset (P=0.002, P=0.039, P<0.001 and P=0.005 respectively).

The fast and reliable acquisition time of this novel CMRA sequence enabled the first direct head to head clinical study to assess the impact of spatial resolution on image quality, visualised vessel length and sharpness as well as diagnostic accuracy in a cohort of 40 patients with suspected CAD. The average acquisition time was 10.9 ± 1.4 mins and 10.0 ± 2.2 mins for the 0.9mm3 and 1.2mm3 CMRA respectively (P<0.001). The 0.9mm3 CMRA obtained diagnostic image quality in 99%, 100% and 90% of proximal, middle and distal coronary segments compared with 98%, 94% and 67% of 1.2mm3 CMRA segments (P=1.00, P=0.06 and P<0.001 respectively). The median (IQR) image quality score for the 3D wholeheart dataset, RCA, LAD and LCx were 3.0 (3.0-4.0), 3.0 (3.0-4.0), 3.5 (3.0-4.0) and 3.0 (2.0- 4.0) for the 0.9mm3 CMRA and 3.0 (2.0-3.0), 3.0 (2.0-3.75), 3.0 (1.25-3.0) and 2.5 (1.25-3.0) for the 1.2mm3 CMRA, (P=0.004, P=0.012, P<0.001 and P=0.011 respectively). The average visible vessel lengths for the 0.9mm3 CMRA vs. the 1.2mm3 CMRA are as follows: RCA 14.2 ± 3.2 cm vs. 11.8 ± 4.0 cm (P<0.001); LAD 11.7 ± 2.8 cm vs. 10.7 ± 3.3 cm (P<0.001) and LCx 7.4 ± 2.5 cm vs. 6.4 ± 2.4 cm (P<0.001). The AUC for the 0.9mm3 CMRA vs. 1.2mm3 CMRA were as follows: per patient 0.82 vs. 0.71 (P=0.24), per vessel 0.85 vs. 0.71 (P=0.13) and per segment 0.86 vs. 0.77 (P=0.19) respectively.

We subsequently demonstrated the clinical maturity of this novel CMRA framework in a dedicated clinical study to compare the framework with 0.9mm3 acquired spatial resolution against CCTA in a cohort of 50 patients with suspected CAD. The patients completed their CMRA scan in an average acquisition time of 10.7 ± 1.4 mins. The 0.9mm3 CMRA obtained diagnostic image quality in 95% of all, 97% of proximal, 97% of middle and 90% of distal coronary segments. The sensitivity, specificity, positive predictive value, negative predictive value and diagnostic accuracy were as follows: per patient (100%, 74%, 55%, 100% and 80%), per vessel (81%, 88%, 46%, 97% and 88%) and per segment (76%, 95%, 44%, 99% and 94%) respectively.

We extended the framework to assess the feasibility of end-to-end deep learning non-rigid motion corrected reconstruction (MoCo-MoDL) for highly accelerated free-breathing CMRA to enable significantly higher shorter acquisitions with faster reconstruction times. The acquisition time for ninefold accelerated CMRA was ~2.5 min. The reconstruction time was ~22 s for the proposed MoCo-MoDL and ~35 min for the conventional approach. MoCo-MoDL achieved higher peak SNR (27.86 ± 3.00 vs. 26.71 ± 2.79; P<0.05) and structural similarity (0.78 ± 0.06 vs. 0.75 ± 0.06; P<0.05) than the conventional approach. Similar vessel length and visual image quality score were obtained with the 2 methods, whereas improved vessel sharpness was observed with MoCo-MoDL.

Finally, we extended the framework to enable simultaneous, non-contrast, co-registered bright blood CMRA and dark blood vessel wall/plaque imaging within a clinically feasible time of ≈ 10 minutes. This dual contrast framework was validated in a cohort of 41 patients with suspected acute coronary syndrome (ACS) who also underwent invasive coronary angiography and intravascular imaging. The mean ± standard error of mean (SEM) plaquemyocardial signal ratio (PMR) of culprit segments was significantly higher than non-culprit segments and normal segments (1.01 ± 0.05 vs. 0.67 ± 0.01 vs. 0.35 ± 0.01, P<0.001 respectively). Coronary segments with lipid, calcium and fibroatheroma had a significantly higher PMR compared to normal coronary segments (P<0.001) but lower than segments with plaque-rupture and thrombus (P<0.001). There was a progressive increase in PMR with increasing coronary segment stenosis (P<0.001). Patients with type-2 diabetes, hypertension, hyperlipidaemia and family history of coronary artery disease (CAD) had a significantly higher PMR compared with patients without these risk factors (P<0.001, P=0.02, P=0.04 and P=0.02 respectively).






Date of Award1 Sept 2023
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorRene Botnar (Supervisor) & Claudia Prieto Vasquez (Supervisor)

Cite this

'