3D late gadolinium enhanced cardiovascular MR with CENTRA-PLUS profile/view ordering: Feasibility of right ventricular myocardial damage assessment using a swine animal model

Abstract

Aims: To develop a high-resolution, 3D late gadolinium enhancement (LGE) cardiovascular magnetic resonance imaging (MRI) technique for improved assessment of myocardial scars, and evaluate its performance against 2D breath-held (BH) LGE MRI using a surgically implanted animal scar model in the right ventricle (RV).

Methods and results: A k-space segmented 3D LGE acquisition using CENTRA-PLUS (Contrast ENhanced Timing Robust Acquisition with Preparation of LongitUdinal Signal; or CP) ordering is proposed. 8 pigs were surgically prepared with cardiac patch implantation in the RV, followed in 60days by 1.5T MRI. LGE with Phase-Sensitive Inversion Recovery (PSIR) were performed as follows: 1) 2DBH using pneumatic control, and 2) navigator-gated, 3D free-breathing (3DFB)-CP-LGE with slice-tracking. The animal heart was excised immediately after cardiac MR for scar volume quantification. RV scar volumes were also delineated from the 2DBH and 3DFB-CP-LGE images for comparison against the surgical standard. Apparent scar/normal tissue signal-to-noise ratio (aSNR) and contrast-to-noise ratio (aCNR) were also calculated. 3DFB-CP-LGE technique was successfully performed in all animals. No difference in aCNR was noted, but aSNR was significantly higher using the 3D technique (p<0.05). Against the surgical reference volume, the 3DFB-CP-LGE-derived delineation yielded significantly less volume quantification error compared to 2DBH-derived volumes (15±10% vs 55±33%; p<0.05).

Conclusion: Compared to conventional 2DBH-LGE, 3DFB-LGE acquisition using CENTRA-PLUS provided superior scar volume quantification and improved aSNR.

Keywords: Breath-hold; Late gadolinium enhancement MRI; Myocardial infarction; Navigator; Viability imaging.

Figure 1

a) Schematics of the 3D PSIR-LGE Sequence. b) Schematic diagram of the CENTRA-PLUS profile ordering over the 3D scan acquisition. The k-space table shows the filled k-space trajectory table over the duration of the scan. Using this scheme, inner (blue) and outer (green) views can be acquired with different navigator gating strategies, such as using a conservative window of 3 mm (inner), and a larger 5 mm window for the remainder of the scan. c) Color scheme of the CENTRA-PLUS ordering within each acquisition table. For each 100 ms acquisition window, ~20-21 consecutively FFEs are acquired from the ky axis towards the outer kz-axis (blue → red). Supplementary animation file shows the time-course of this ky-kz acquisition table.

Figure 2

Gold standard measurements were made using an elliptical shape assumption with heights adjusted across the major axis. In addition to the live measurements immediately in the surgical suite, two additional measurements were performed retrospectively, and the final value was calculated as an average of the 3 measurements.

Figure 3

Acquired images from two cases (Pig #1 in top row; #4 in bottom row). Left column shows the 2DBH-PSIR-LGE cases; Right column shows the 3DFB-CP-PSIR-LGE cases. Identified patch is shown using the red arrow. We also note in swine #1 with the 3D acquisition, where the navigator was placed in the foot-head direction. This resulted in the pencil-beam RF pulse producing an artifact that mimicked LGE and excited a circular region in the chest cavity. The blinded clinician read through the artifact in this case, and the protocol was modified between swine #1 and #2,, to instead prescribe the pencil-beam navigator in the AP direction across the chest wall below the heart position.

Figure 4

Acquired images from Pig #5 with 2.5cm³ directly measured volume of the RV patch scar region. a) 3 slices and the corresponding geometric model view from the 2D (left) and 3D (right), respectively. The implanted patch is shown only in one slice on the 2D image, whereas it is depicted on multiple slices with the 3D sequence. The improved resolution with the 3D sequence allows a clear depiction of the circular patch and scar geometry, despite presence of bright blood infiltrating onto the surface in both models. These geometric visualization employed a maximum intensity projection approach across a 4 mm segment orthogonal to the manually delineated contour on each slice. The yellow points indicate the drawn contours. b) Reformatted view of the 3DFB-LGE sequence at 1.0 mm³ interpolated resolution allowed thickness measurements of the RV patch.

Figure 5

Top row: 3DFB-LGE vs surgical ref volume measurements; Bottom row: 2DBH-LGE vs surgical reference volume measurement comparisons. Pearson's Correlation plot is shown on the left column, Bland-Altman Analysis on the right column. 3DFB-LGE measurements yielded good agreement with the surgical reference measurements, whereas the 2DBH-LGE measurements did not.