2169

Optimization of cardiac functional MR imaging in rats. Image quality versus scan tolerance in sick animals
El-Sayed H Ibrahim1, Matthew Budde1, Dhiraj Baruah1, Anne Frei1, Rachel Schlaak1, Michael Flister1, and Carmen Bergom1

1Medical College of Wisconsin, Milwaukee, WI, United States

Synopsis

Using MRI for studying cardiac function in rat models of radiation-therapy is important for better understanding of cardiotoxicity. In this study, we developed an optimized cardiac functional MRI protocol for rat imaging, and investigated the effects of changing imaging parameters on cardiac measurements. The results showed that blood-to-myocardium contrast-to-noise ratio in the cine images significantly affects ejection-fraction measurements, whereas reduced resolution has less effect on ejection-fraction. However, reduced resolution in the tagged images has significant effect on strain measurements, whereas tagline density has less effect, as long as sufficient resolution is maintained and more than one tagline intersects the myocardium.

INTRODUCTION

Small-animal, e.g. rats, MR imaging is essential for better understanding of cardiac dysfunction development in different cardiovascular diseases, as specific animal models can be studied under controlled conditions and the imaging parameters can be compared to histology results. Specifically, cardiac MR imaging of rat models of cancer, for better understanding of chemotherapy or radiation-therapy (RT) cardiotoxicity, is challenging as the animals are quite sick with deteriorated cardiac function and inconsistent breathing pattern. Further, due to small anatomy in rats, image averaging is typically conducted to maintain adequate signal-to-noise ratio (SNR).

In this study, we developed an optimized MRI protocol for evaluating both global and regional cardiac function in RT rat models, and investigated the effects of different imaging parameters on image quality and cardiac measurements.

METHODS

Adult female rats (n=6; both RT and control) were scanned on 9.4T Bruker Biospec MRI scanner with 30-cm bore diameter and equipped with 4-element surface coil. The RT rats received image-guided localized whole-heart RT to 24Gy using 3 equally-weighted fields. The MRI scan included both cine and tagging sequences. Both short-axis and long-axis images were acquired using fast low-angle shot (FLASH) pulse sequence with both cardiac and respiratory gating. Total scan time<1 hour.

The cine imaging parameters were as follows: TR=7ms, TE=2.1ms, flip angle=15⁰, matrix=176x176, FOV=40x40mm2, slice thickness=1mm, acquisition bandwidth=526Hz/pixel, #averages=2, #cardiac phases=20, and scan time/slice ~2:30 minutes. The effects of changing the following parameters on image quality and measured ejection-fraction (EF) were studied: #averages, imaging flip angle, imaging matrix, slice thickness, fat saturation, and TR.

The tagging imaging parameters were as follows: TR=7ms, TE=2.5ms, flip angle=15⁰, matrix=256x256, FOV=40x40 mm2, slice thickness=1mm, acquisition bandwidth=376Hz/pixel, #averages=3, #cardiac phases=20, and scan time/slice ~5 minutes. The effects of changing the following parameters on image quality and measured myocardial strain were studied: #averages, tagline width, and tag separation.

The cine images were processed by delineating the endocardial and epicardial contours, from which regional (in a certain slice) or global (whole heart) EF was measured. The tagged images were analyzed using the harmonic phase (HARP) technique to measure myocardial strain.

RESULTS

Optimal cine and tagging imaging parameters are listed in Table-1. Figure-1 shows an optimal cine image and the effects of changing SNR and blood-to-myocardium contrast-to-noise ratio (CNR) on the results. Reducing or increasing SNR by up-to 50% did not significantly affect image quality and the measured EF. Reducing #averages in half resulted in ≤5% change in EF, while increasing #averages from 2 to 3 resulted in ≤1% in EF. Increasing the flip angle resulted in improved blood-to-myocardium CNR at the cost of reduced SNR. Increasing the flip angle up-to 40⁰ resulted in ≤1% change in EF, while reducing the flip angle resulted in decreased CNR, and at 5⁰ flip angle, EF changed by >10%.

Figure-2 shows the effects of partial-volume averaging and image artifacts on image quality and FE. Reducing the imaging matrix in half (to 86x86) resulted in EF underestimation by 4-5%, while increasing the imaging matrix above 176x176 resulted in ≤1% change in EF. Increasing slice thickness resulted in partial-volume effect and EF underestimation by 3-4%. Adding fat saturation slightly affected image quality, but EF did not change. Increasing TR>10ms resulted in signal inhomogeneity artifacts and up-to 10% change in EF. Reducing FOV resulted in wrap-around artifacts that affected EF depending on the artifact severity.

Figure-3 shows an optimal tagging image, and the effects of changing SNR and resolution on strain measurements. Reducing #averages<3 resulted in 2-3% change in strain. Reducing image resolution in half (matrix=128x128) resulted in 7-10% change in strain, while further reduction in resolution resulted in unrealistic strain values. Figure-4 shows the effects of tagging profile on strain. Increasing tagline width to 50% of the tag separation resulted in ≤2% change in strain. Increasing the tagline density resulted in blurred tags and up-to 10% change in strain, whereas increasing tag separation resulted in significant change in measurements, such that unrealistic strain values were obtained when only one tagline intersected the myocardium.

DISCUSSION and CONCLUSION

Based on results from this study, blood-to-myocardium CNR and image artifacts in the cine images significantly affect EF measurements, whereas resolution has less effect on EF. On the other hand, reduced resolution in the tagged images has significant effect on strain measurements, whereas tagline density has less effect on strain, as long as sufficient image resolution is maintained and more than one tagline intersects the myocardium.

In conclusion, this study provides an optimized cardiac functional MRI protocol for imaging rats, which ensures accurate measurements while minimizing scan time to be tolerated by sick animals.

Acknowledgements

No acknowledgement found.

References

1. Filster et al, Breats Cancer Research Treat, 165:53-64

2. Osman et al, Magn Reson Med, 42:1048-1060

3. Ibrahim, Heart Mechanics MRI. CRC Press, 2017

4. Said et al, Ann Onc, 25:276-282

Figures

Figure 1. Effects of signal-to-noise ratio (SNR) and blood-to-myocardium contrast-to-noise ratio (CNR) on measured EF (red=significant change). (a) Standard cine image acquired with optimal imaging parameters (EF=61% in this example). (b) Reducing #averages from 2 to 1 results in 50% reduction in SNR, and EF changes to 66%. (c) Increasing #averages from 2 to 3 results in 50% increase in SNR; however, EF hardly changes. (d) Increasing flip angle from 15⁰ to 40⁰ increases blood-to-myocardium CNR at the cost of reduced SNR. Correct EF still obtained. (e) Reducing flip angle to 5⁰ results in shallow blood-to-myocardium CNR,which significantly affects EF.

Figure 2. Effects of partial-volume averaging and image artifacts on measured EF (red=significant change). (a) Reducing matrix from 176x176 to 86x86 results in partial-volume averaging, and EF decreases to 57% (correct measurement=61%). (b) Increasing matrix to 300x300 results in slight change in EF. (c) Increasing slice-thickness from 1mm to 3mm results in partial-volume averaging, and EF decreases to 59%. (d) Adding fat-saturation option slightly affects SNR, but correct EF is obtained. (e) Doubling TR from to 15ms results in signal inhomogeneity artifacts, which significantly affects EF (=53%). (f) Reducing field-of-view (FOV) results in wrap-around aliasing artifacts and incorrect EF (=68%).

Figure 3. Effects of signal-to-noise ratio (SNR) and resolution on strain imaging (red=significant change). (a) Standard tagging image acquired with optimal imaging parameters, and resulting strain curves. Average peak circumferential strain (Ecc) = -24%. (b) Reducing #averages results in SNR reduction, but has slight effect on strain measurement. (c) Reducing spatial resolution in half (matrix changes from 256x256 to 128x128) results in significant decreases in strain (=-17%). (d) Any further decrease in spatial resolution results in unrealistic results (strain=3%).

Figure 4. Effects of tagging profile on strain measurements (red=significant change). (a) Increasing the width of the taglines from 0.1mm to 0.5mm (tag separation = 1mm) has minimal effect on strain measurements (strain changes from -24% to -26%). (b) Doubling tagline density (tag separation = 0.5mm) results in blurry tags, which significantly affects strain measurement (Ecc changes from -24% to -15%). (c) Reducing tag density in half (tag separation =2mm) results in sparse taglines across the myocardium, such that strain analysis results in unrealistic strain value of -100%.

Table 1. Optimal imaging parameters for cine and tagging imaging sequences.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
2169