Dual-phase coronary MR angiography (CMRA) allows for high-resolution, volumetric visualization of cardiovascular anatomy and function, and has proven particularly useful in patients with congenital heart disease¹. With this approach, high-resolution whole-heart CMRA is acquired during both systolic and diastolic rest period. However, the dual-phase CMRA images are susceptible to degradation due to respiratory motion as well as prolonged scan times due to low gating efficiency. Image-based navigation (iNAV) is an emerging approach for respiratory motion correction which allows tracking and correction of respiratory motion directly on the heart². Nevertheless, the different cardiac motion states of the systolic and diastolic iNAVs require further improvement of this technique to enable accurate motion correction for dual-phase CMRA. The purpose of this study was to implement and evaluate a new iNAV approach for dual-phase CMRA.

The proposed iNAV motion correction strategy for dual-phase CMRA is shown in Figure 1. Separate iNAV references were acquired for the systolic and diastolic phase, to account for differences in cardiac motion state. The iNAV references were defined as the first acquired iNAV in each cardiac phase. Each subsequent iNAV for a certain phase was registered to its corresponding iNAV reference using normalised cross-correlation. Translational motion correction was performed in foot-head and left-right direction. Additionally, respiratory gating was implemented using the diminishing variance algorithm, where a gating efficiency of 50% was assumed. With this approach, the systolic and diastolic iNAVs were separately gated to end-expiration. The iNAV motion correction and gating was implemented inline on the scanner and no post-processing of the CMRA was required.

In a pilot study, the proposed use of independent systolic and diastolic reference iNAVs was compared to a scenario where a single iNAV reference was used for both cardiac phases. To investigate this a diastolic reference iNAV was registered to systolic iNAVs to motion correct a systolic CMRA, and a systolic reference iNAV registered to diastolic iNAVs to motion correct a diastolic CMRA. This was to evaluate the influence of cardiac motion in the iNAV motion estimation and was performed in two healthy volunteers.

The dual-phase CMRA with independent systolic and diastolic iNAV correction and gating was evaluated in 8 healthy subjects. For comparison a dual-phase scan was acquired using conventional 1D diaphragmatic navigator (1D NAV) with a 0.6 tracking factor and 5mm gating window. The CMRA sequence had the following imaging parameters: FOV=300×300×100mm (coronal orientation), Δx=1.5×1.5×1.5mm3, α=70°, SENSE=2.5 (phase encoding direction). All experiments were performed on a 1.5T clinical scanner (Philips Healthcare, Best, The Netherland) using a 32-channel cardiac coil. The dual phase CMRA datasets were reformatted using dedicated software to visualize the right coronary artery (RCA), left anterior descending (LAD) artery and left circumflex (LCX) artery. Vessel sharpness was measured for the RCA, LAD and LCX for both cardiac phases and for both motion correction methods. Furthermore, the scan time was recorded for all dual phase CMRA scans.

The pilot study in two healthy volunteers demonstrates improved image quality obtained when using independent systolic and diastolic iNAVs, shown in Figure 2.

Representative images from two healthy volunteers are shown in Figure 3, where improved visualization of coronary arteries can be seen using iNAV compared to 1D NAV. The vessel sharpness for systole and diastole of the RCA, LAD and LCX using iNAV and 1D NAV, averaged across all 8 healthy subjects are shown in Figure 4. For the systolic phase the LAD was significantly sharper using iNAV compared to 1D NAV (p < 0.05). For the diastolic phase the iNAV provided sharper LAD (p<0.05) and RCA (p<0.05) compared to 1D NAV. The dual phase CMRA scan time using iNAV was 4(min):51(sec)±0:24, while using 1D NAV the scan time was 7:48±2:18, a difference that was statistically significant (p<0.001).

We have implemented and evaluated a new approach for dual-phase CMRA using image-based navigation. This approach maintains a high accuracy of respiratory motion tracking, despite the different cardiac motion states of the data acquisition, by employing two independent navigator references. It compares favorably to the conventional 1D NAV approach in healthy subjects, by providing similar or better vessel sharpness while reducing scan time.