Free-breathing myocardial T2* mapping using GRE-EPI and automatic non-rigid motion correction

Abstract

Background: Measurement of myocardial T2* is becoming widely used in the assessment of patients at risk for cardiac iron overload. The conventional breath-hold, ECG-triggered, segmented, multi-echo gradient echo (MGRE) sequence used for myocardial T2* quantification is very sensitive to respiratory motion and may not be feasible in patients who are unable to breath-hold. We propose a free-breathing myocardial T2* mapping approach that combines a single-shot gradient-echo echo-planar imaging (GRE-EPI) sequence for T2*-weighted image acquisition with automatic non-rigid motion correction (MOCO) of respiratory motion between single-shot images.

Methods: ECG-triggered T2*-weighted images at different echo times were acquired by a black-blood, single-shot GRE-EPI sequence during free-breathing. A single image at a single TE is acquired in each heartbeat. Automatic non-rigid MOCO was applied to correct for in-plane respiratory motion before pixel-wise T2* mapping. In a total of 117 patients referred for clinical cardiac magnetic resonance exams, the free-breathing MOCO GRE-EPI sequence was compared to the breath-hold segmented MGRE approach. Image quality was scored independently by 2 experienced observers blinded to the particular image acquisition strategy. T2* measurements in the interventricular septum and in the liver were compared for the two methods in all cases with adequate image quality.

Results: T2* maps were acquired in all 117 patients using the breath-hold MGRE and the free-breathing MOCO GRE-EPI approaches, including 8 patients with myocardial iron overload and 25 patients with hepatic iron overload. The mean image quality of the free-breathing MOCO GRE-EPI images was scored significantly higher than that of the breath-hold MGRE images by both reviewers. Out of the 117 studies, 21 breath-hold MGRE studies (17.9% of all the patients) were scored to be less than adequate or very poor by both reviewers, while only 2 free-breathing MOCO GRE-EPI studies were scored to be less than adequate image quality. In a comparative evaluation of the images with at least adequate quality, the intra-class correlation coefficients for myocardial and liver T2* were 0.868 and 0.986 respectively (p < 0.001), indicating that the T2* measured by breath-hold MGRE and free-breathing MOCO GRE-EPI were in close agreement. The coefficient of variation between the breath-hold and free-breathing approaches for myocardial and liver T2* were 9.88% and 9.38% respectively. Bland-Altman plots demonstrated good absolute agreement of T2* in the interventricular septum and the liver from the free-breathing and breath-hold approaches (mean differences -0.03 and 0.16 ms, respectively).

Conclusion: The free-breathing approach described for T2* mapping using MOCO GRE-EPI enables accurate myocardial and liver T2* measurements and is insensitive to respiratory motion.

Fig. 1

Sequence diagram for a black-blood, single-shot GRE-EPI which acquires a series of ECG-triggered T2*-weighted images at different echo times (TEs) during free-breathing

Fig. 2

The workflow to generate the T2* map from free-breathing, T2*-weighted images acquired using the ECG-triggered, dark-blood, single-shot GRE-EPI sequence

Fig. 3

Representative examples of T2* maps in four patients (a–d) acquired using breath-hold MGRE and free-breathing MOCO GRE-EPI. All four of these patients were able to hold their breath successfully during the breath-hold MGRE exam and both techniques produced T2* maps of good image quality

Fig. 4

Examples of the T2*-weighted source images and their corresponding T2* maps in two patients (top and bottom) who failed to hold their breath during the breath-hold MGRE T2* scan: Severe ghosting and image blurring artifacts caused by respiratory motion during image acquisition are evident in the T2*-weighted images acquired with segmented MGRE; the corresponding T2* maps were also corrupted by respiratory motion (a and c), while the single-shot MOCO GRE-EPI acquisition effectively froze respiratory motion during the image acquisition and produced artifact-free T2* maps during free-breathing (b and d)

Fig. 5

One example of T2*-weighted source images of the first three echoes and their resulting T2* maps in a patient with normal heart and severe hepatic iron overload. Images were acquired using breath-hold MGRE (a) and free-breathing GRE-EPI approaches (b) Both techniques failed to provide T2* quantification in the liver. (c) Curves of mean signal intensity within the ROI in the liver (red circle in a and b) vs. echo times, showing that the signal is below the noise level even in the image from the earliest echo

Fig. 6

Bland-Altman plots showing the agreement for the T2* in the interventricular septum (a) and the liver (b) from free-breathing MOCO GRE-EPI and breath-hold MGRE measurements