Introduction

Quantitative Magnetic Resonance Imaging (qMRI) via T1 and T2 mapping is increasingly being used in cardiac imaging to guide clinical care [1]. Currently T1 maps are typically acquired using a Modified Look Locker sequence (MOLLI) [2]. T2 maps are commonly acquired using a Gradient-Spin-Echo (GraSE) sequence [3]. Both methods would benefit from scan acceleration to reduce the number of breath holds (BHs) required, as currently a single BH only permits acquisition of a single slice. T1 and T2 mapping methods can be accelerated using Sensitivity encoding (SENSE) which benefits from a high density receive array. Currently SENSE=2 is applied in our clinic in combination with the vendor-supplied 16-channel anterior receiver array. In recent years we have developed a 72 channel receive array which would allow for higher acceleration factors and thus potentially faster T1 and T2 mapping acquisition [4]. The goal of this study was to accelerate native cardiac T1 and T2 mapping using a dedicated 72-channel receive array. SENSE factors up to 4 and 6 were explored to facilitate native T1 and T2 mapping of 3 slices in 1 and 2 BHs, respectively.

Methods

20 healthy subjects (10 female), aged 34 ± 7 years, body weight 76 ± 9 kg, heart rate 57 ± 10 beats/min (bpm) where included. MR examinations were performed with a 3T MR scanner (Ingenia, Philips Healthcare, Best, The Netherlands ). A novel 72-channel cardiac array coil was used which has 2 flexible arrays of 36 coils which were placed on top of the chest of the subject, as shown in Figure 1. For MOLLI, a shared inversion pulse interleaved approach was designed to acquire 3 slices after the same inversion (Figure 2). Thereby, only 1 BH was needed to acquire the 3 slices. We took into account that because all slices share the same inversion pulse, the inversion time for the 3 slices was slightly different. For T2 mapping we used a multi-shot GraSE acquisition. With higher acquisition factors less shots, and thus heartbeats, are needed to acquired one image. Therefore multiple slices can be acquired within one BH. For both T1 and T2 mapping acquisition we acquired 3 axial slices covering the basal, mid-ventricular and apical myocardium. Images with different inversion times (T1 mapping) and for different echo-times (T2 mapping) were aligned (Elastix) using a b-spline registration and pixel-wise fitted to provide T1 and T2 maps. Average T1 and T2 values were determined in 16 segments of the 17-segment AHA model (excluding the apical segment) and compared per slice and per segment between SENSE=2, 4, and 6 accelerated acquisitions. To quantify the differences, average T1 and T2 values were statistically compared on a slice and segment level. Normality of data was checked and a two-way factor ANOVA analysis was applied with acceleration factor and segments as factors. All statistical analyses were performed using R (v. 4.0.3) and Rstudio (2021.09.0.351). P-values of < 0.05 were considered significant.

Results

Figure 3a shows a representative example of native T1 maps acquired with SENSE=2 (3BHs x 11 sec at 60 bpm), 4 (1BH x 11 sec at 60 bpm) and 6 (1BH x 11 sec at 60 bpm). Similarly, Figure 3b shows typical T2 maps acquired with SENSE=2 (3BHs x 13 sec at 60 bpm), 4 (3BHs x 9 sec at 60 bpm) and 6 (2BHs x 10 sec at 60 bpm). Figure 4c shows the descriptive statistics of the native T1 and T2 map acquisitions using SENSE=2, 4 and 6. Mean T1 and T2 values per segment are shown in figures 5a-b. The two-factor ANOVA revealed no difference in mean and segmental T1 values per slice: no significant differences were found in any of the 16 segments between SENSE=2 and 4 and SENSE=2 and 6 and no interaction was present (figure 5a). The two-factor ANOVA revealed statistical differences per slice in mean T2 values due to PI factor and segments, and an interaction effect was present between both conditions. For T2 on segmental level, significant differences were seen between SENSE=2 and 4 in segments 1, 3 (p<0.05) and 2, 8 (p<0.01) . For SENSE=6 this was in segments 1, 2, 5 (p<0.001), 3, 6, 7, 8, 12 (p<0.01) and 13, 14 (p<0.05) (figure 5b).

Discussion

Generally, both T1 and T2 maps have a noisier appearance with increasing acceleration, particularly for the basal slices. However, 16-segmental multislice native myocardial T1 mapping acquired in a single BH using an accelerated acquisition with PI factors up to 6 were similar to reference values acquired in 3 BHs. The accelerated cardiac T2 mapping values significantly deviated from reference values particularly for the basal slices which we believe could be due to motion sensitivity of the T2 mapping sequence. Currently a second 72 channel coil has been build doubling the total of receive elements to 144.

Conclusion

Native myocardial T1 mapping quantification of 3 slices in just 1 BH is possible using a modified MOLLI sequence in combination with a new 72-channel receive array and SENSE=4 or 6. The myocardial T2 mapping protocol can be accelerated to achieve fewer BHs. However, this goes at the expense of decreased accuracy of the T2 values in specific segments.

Acknowledgements

No acknowledgement found.