Accuracy evaluation of pulse sequences using inversion recovery pulse for T1 measurement.
Noriyuki Tawara1,2, Kanokvalee Ponkanist2, Shuichi Shiratori2, Anchali Krisanachinda2, and Toru Yamamoto1

1Faculty of Health Sciences, Hokkaido University, Sapporo, Japan, 2Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand

Synopsis

It has been reported that T1 in skeletal muscles is proportional to the percentage of typeⅠfibers which would be a determinant for athletic performance. However, the precision of the measured T1 values is unclear due to various pulse sequences used for the measurement. In this study, we evaluated the reliability of T1 measured by pulse sequences with inversion recovery technique that is believed to be most accurate in T1 measurement. A value of goodness of fitting was employed for the quantitative evaluation of the obtained data. The influence of SNR on the T1 measurement was also examined.

Purpose

Typical human skeletal muscle is composed of typeⅠand typeⅡmuscle fibers. The relative percentage of typeⅠfibers would be a determinant for athletic performance in either endurance or sprint events. Since it has been reported that T1 is proportional to the percentage of typeⅠfibers in human skeletal muscles [1], the T1 measurement in skeletal muscles has been tried. However, the precision of the measured T1 values is unclear due to various pulse sequences for the measurement. In this study, we evaluated the reliability of T1 measurements of pulse sequences with inversion recovery technique that is believed to be most accurate in T1 measurement. A theoretical fitting parameter was employed for the quantitative evaluation of the obtained data. The influence of SNR on the T1 measurement was also examined.

Materials and Methods

Polyvinyl alcohol (PVA) gel (T1 = 850 ms, NIKKO FINES INDUSTRIES Co.,Ltd., Tokyo, Japan) was examined on 1.5T whole-body MRI units (Magnetom Symphony or Aera, Siemens AG, Erlangen, Germany) using an extremity coil (Ex) or a built-in body coil (Bo). Temperature was 27℃. An inversion recovery pulse sequence (IR) used Spin-Echo sequence (SE; IR-SE), TurboFLASH (TF; IR-TF), and Spin-Echo Echo-Planar-Imaging (EPI; IR-EPI). The echo time (TE) of this acquisition scheme was 10 ms (SE and TF) and 30 ms (EPI), respectively. The other protocols of SE measurements were performed with TR 5000 ms, TE 10 ms, voxel size 0.93×0.93×10 mm, matrix 256×256, field of view (FOV) 240 mm×240 mm, and NEX 1. Thirteen inversion times (TI) were used, varying from 50 to 4500 ms (50, 100, 125, 170, 200, 250, 500, 1000, 1500, 2000, 3000, 4000, 4500). Total acquisition time was 3 h 24 min for 13 slices. The condition of TF were with TE 1.29 ms, matrix 128×128, and thirteen inversion times (TI) were used, varying from 50 to 4500 ms (110, 125, 150, 170, 200, 250, 500, 1000, 1500, 2000, 3000, 4000, 4500). Total acquisition time was 1 m 30 s for all slices. TR and FOV was the same as SE. The condition of EPI was with TE 30 ms, and a total acquisition time of 26 s. The other condition was same as TF. However, EPI-images were only acquired by the extremity coil.

Assuming full relaxation of the spin system before each excitation, the absolute MR signal (M) as a function of inversion time TI is given by

M(TI) = MTI=0×ABS[1 – 2NRF×exp(-TI/T1)]

where the factor NRF took into account the effect of an incomplete inversion. NRF is related to the flip angle θ of the preparation pulse by: NRF=(1-cosθ)/2.

Five ROI were placed in physical object from each image. Imperfections of the inversion pulses are accounted by factor according to Blüml S et al [2]. T1 relaxation time was calculated by Levenberg-Marquardt method for non-linear regression analysis. All analysis was performed in Matlab R2014b programming environment (The Mathworks, Natick, MA). The estimation of fitting calculating results was used the parameter Q that provides a stringent measure of goodness of fit using chi-square [3]. If the results of fitting were good, the values of chi-square close to zero, and the parameter Q close to one.

Evaluation of significant difference was as follows. Highest accuracy T1 calculated by each pulse sequence was set as the standard value. Differences between the standard T1 and the others value exceeding 10 percent of the T1 value were assumed to be significant. About the difference in the parameter Q about NRF, the differences of over 5 percent of parameter Q were assumed to be significant in each sequence by unpaired t-test.

Results and Discussion

Both fitting evaluation values Q and NRF for the data obtained by using IR-TF were the best and showing the value of T1 is around 880 ms as the most reliable value in this experiment (Figure 1). In general, when TR is greater than 5T1, the calculated T1 is insensitive to the effect of RF inhomogeneity and giving rise to yield a reliable T1 [4]. Although TR (5000 ms) is greater enough in our experiment, the obtained T1 depends on the used pulse sequence. There should exist unideal flip angle portion especially in the vicinity of the edge plane of the slice depending on the slice profile. The dispersion of the longitudinal magnetization after read out of the MR signal affects the T1 measurement. When the longitudinal magnetization after inversion time is Mz(TI), the longitudinal magnetization after read out pulse are expressed as Mz(TI)cosαat the position of flip angle α for gradient echo pulse sequence. For the SE pulse sequence, that is expressed as Mz(TI)cosαcos2α. The sinusoidal part at the edge of the slice where α is less than 90° becomes minus for the SE pulse sequence. This may deteriorate the accuracy of T1 measurement. Although the reason of the worst T1 accuracy for IR-EPI (Figure 1) is unknown, this may be caused by broad slice profile of soft RF pulses to reduce SAR. The parameter Q affected by NRF was significant, especially the value of SE had weight changes by using RF coil.

The results of different T1 versus each sequence condition of IR sequences studies summarized in Figure 1. The parameter Q of all sequence conditions using Ex except EPI was higher than the value using Bo. These results indicate that T1 measurement is strongly susceptible to SNR. It is suggested that the image noise has an influence on underestimate in calculating T1. Figure 2 shows typical evaluated and fitted T1 relaxation curve from pixels in PVA-gel on IR-TF and IR-EPI images. The trend of SE agrees with the result of TF.

Regarding rapid imaging for T1 measurement, the weak points of EPI are in harmony with these results [5]. Likewise, these results about useful TF are consistent with the previous studies. Therefore, there is every possibility that IR-TF could be applied for rapid T1 measurement.

Conclusions

Among pulse sequences for the T1 measurement using IR technique, IR-TF is the best. Moreover, IR-TF can be applied for rapid T1 measurement.

Acknowledgements

This work was supported by a Grant-in-Aid for Scientific Research (B) from the Ministry of Education, Culture, Sports, Science, and Technology (25282170). Furthermore, we thank Japan Institute of Sports Sciences (JISS) for only assistance with data collection.

References

[1] Houmard JA. et al. Relationship between MRI relaxation time and muscle fiber composition. J Appl Physiol. 1995;78(3):807-809.

[2] Blüml S et al. Spin-Lattice relaxation time measurement by means of a TurboFLASH technique. Magn Reson Med. 1993;30(3):289-295.

[3] Gambarota G et al. Osmotic effects on the T2 relaxation decay of in vivo muscle. Magn Reson Med. 2001;46(3):592-599.

[4] Wang J et al. T1 measurements incorporating flip angle calibration and correction in vivo. J Magn Reson. 2006;182(2):283-292.

[5] Nekolla S et al. T1 maps by K-space reduced Snapshot-FLASH MRI. J Comput Assist Tomogr. 1992;16(2):327-332.

Figures

The value Q of goodness of fitting, fitting parameter NRF and obtained T1 value in each sequence condition. #The mean difference value more than 10 percent versus standard T1.

Data plot of T1 relaxation measurement of PVA-gel phantom using (a) IR-TF with an extremity coil, (b) IR-TF with a body coil, and (c) IR-EPI with an extremity coil. The solid lines represent the best fitting and the values of the fitting parameter NRF are shown in each panel. The dashed lines represent the result of the fitting with NRF = 1.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
4481