Quantitative Measurements of Proton Spin-Lattice (T 1 ) and Spin Spin (T 2 ) Relaxation Times in the Mouse Brain at 7.0 T

Transcription

1 Magnetic Resonance in Medicine 49: (2003) Quantitative Measurements of Proton Spin-Lattice (T 1 ) and Spin Spin (T 2 ) Relaxation Times in the Mouse Brain at 7.0 T David N. Guilfoyle, 1 * Victor V. Dyakin, 1 Jacqueline O Shea, 1 Gaby S. Pell, 1 and Joseph A. Helpern 1 4 The goal of this work is to provide regional T 1 and T 2 values at a field strength of 7 T for the normal mouse brain at 6 weeks and 1 year old. A novel segmented snapshot FLASH sequence was used to measure T 1 in the hippocampus, corpus callosum, and the retrosplenial granular (RSG) cortex; T 2 measurements were made in the same regions using a single spin echo sequence repeated at six separate echo times. Both T 1 and T 2 measurements were validated with phantom measurements. Magn Reson Med 49: , Wiley-Liss, Inc. Key words: T 1 ; T 2 ; k-space; snapshot FLASH; hippocampus; corpus callosum; retrosplenial granular (RSG) cortex Historically, rats have been used routinely as disease models, but with the advent of transgenesis, and the economic advantages of creating transgenic mouse models of diseases, there is an increasing body of literature citing the use of mice in MRI. Much of the work is purely anatomical imaging. While values for T 1 and T 2 are published for normal and ischemic rat brain at 4.7 T (1,2), there is no equivalent study for the mouse brain, although there is a report of relaxation measurements in the hippocampus at 2 T (3) and a report of T 2 in the frontal cortex at 7 T (4). The ultimate aim of the work presented in this note is to characterize the T 1 and T 2 relaxation times in the mouse brain in order to fill a void in the literature for quantitative baseline values in an animal model of increasing importance. In addition, this study was designed to assess if a fast T 1 acquisition method could provide T 1 data comparable with the standard T 1 inversion recovery method. The mice used in the study are from the same background strain (B6/SW) that is (generally) used to generate transgenic mice with Alzheimer s pathology. Thus, this work is intended to provide baseline measurements of T 1 and T 2 in nontransgenic controls. Transgenic mice with Alzheimer s disease overexpress amyloid precursor protein. These an- 1 Nathan S. Kline Institute, Center for Advanced Brain Imaging, Orangeburg, New York. 2 Department of Radiology, New York University School of Medicine, New York, New York. 3 Department of Psychiatry, New York University School of Medicine, New York, New York. 4 Department of Physiology & Neuroscience, New York University School of Medicine, New York, New York. Grant sponsor: NIA; Grant number: P01 AG ; Grant sponsor: Wyeth Ayerst Pharmaceuticals. *Correspondence to: Dr. David N. Guilfoyle, Division of Medical Physics, The Nathan Kline Institute, 140 Old Orangeburg Road, Orangeburg, NY Received 1 May 2002; revised 30 September 2002; accepted 1 October DOI /mrm Published online in Wiley InterScience ( Wiley-Liss, Inc. 576 imals are not easily maintained in the magnet under anesthesia for long periods, especially as they age. Fast image acquisitions, therefore, were an important criterion in this study but not at the expense of accuracy. Both T 1 and T 2 measurements were validated with phantom studies. Spin echo and inversion recovery methods are considered the workhorses of MRI for T 2 and T 1 measurement, respectively, despite their long acquisition times, especially in the latter case. Faster methods are necessary, particularly for T 1 measurement in vivo at 7 T, where the T 1 of brain tissue is on the order of several seconds. The TR necessary for standard inversion recovery measurements results in unacceptably long scan times. In order to achieve full relaxation, a TR time of approximately 5 T 1 is necessary, for even a low resolution (64 64) standard inversion recovery with only a few time points along the recovery curve would require a total acquisition time of several hours. Echo planar Imaging can be used for high-speed and accurate T 1 mapping (5). However, the susceptibility distortion in a mouse brain at 7 T prohibits the use of this technique. A considerable increase in the speed of T 1 mapping can be achieved by implementing the Look- Locker method (6). This method uses many low flip angle acquisitions to inspect a single recovery curve following an inversion pulse. The most common variant of the Look- Locker approach is the snapshot FLASH (fast low angle shot) experiment first implemented by Haase et al. (7). A snapshot FLASH image sequence is repeated many times along the inversion recovery curve. In this way, a series of images at linear time points along the recovery curve are acquired and a T 1 map can be generated from these data. The restriction in this case is that the flip angle must be very small so that the perturbation of the recovering longitudinal magnetization is negligible. Typically, the flip angle must be less than 5. This restriction makes the signal-to-noise ratio (SNR) too low for our targeted voxel resolution of mm 3. Another variant of the Look-Locker technique is to use interleaved reciprocal lattice or k-space sampling. The advantage of this approach is the use of lower bandwidth acquisitions and lower bandwidth RF pulses. The former increases the SNR and the latter allows the use of narrower slice thickness compared to a snapshot FLASH approach. One such technique is PURR (progressively unsaturated relaxation during perturbed recovery from inversion) (8). In this approach the recovery curve is logarithmically sampled with 4 6 lines of k-space at each time point after the inversion. In evaluating the PURR technique, we found that using more than 4 lines of k-space for each time

2 Relaxation Times in the Mouse Brain at 7.0 T 577 interval yielded inaccurate T 1 measurements because the increase in the number of lines caused the longitudinal magnetization to be perturbed. Linear sampling of the recovery curve allows more lines of k-space to be used for each time point without perturbing the longitudinal magnetization. In the limit, all lines of k-space are sampled at each time point, as in snapshot FLASH. However, high acquisition bandwidths are required to minimize acquisition time. This constraint can be relaxed with the use of interleaved FLASH with linear sampling along the recovery curve. MATERIALS AND METHODS All experiments were performed with 7 T 40-cm horizontal bore magnet (Magnex Scientific, Abingdon, UK) interfaced to a Marconi (formerly SMIS) console (Farnham, UK). This system is equipped with a gradient set capable of 100 gauss/cm with 200 s rise times. An eight rod, 3 cm ID quadrature birdcage transmit/receive RF coil (Morris Instruments, Canada) was used for these studies. For mouse brain imaging, the goal was to achieve mm 3 voxel resolution. FIG. 1. Amplitude modulation in the phase-encode direction for three types of acquisition ordering for a segmented snapshot FLASH. a: Linear coverage of k-space. The order of phase encode steps are 128 to 121 for the first pass, 120 to 113 for the second. b: Interleaved center-out coverage with acquisition order: 0, 1, 1, 2, 2, 3, 3, 4 for the first pass and 4, 5, 5, 6, 6, 7, 7, 8 for the second. c: Interleaved segmented center-out coverage with acquisition order: 0, 16, 32, 48, 64, 80, 96, 112 for the first pass and 1, 17, 33, 49, 65, 81, 97, 113 for the second. T 1 Measurements The sequence used for T 1 measurements was an interleaved FLASH method. This would allow 8 k-space lines per time interval, linearly dispersed over the recovery curve. Eight k-space lines are sampled at each time point along the recovery curve to give n time points in total. This is then repeated with a second inversion pulse followed by the next 8 lines of k-space also linearly sampled n times along the recovery curve. This is repeated with subsequent inversion pulses until all the lines of k-space have been sampled. In our experiments the matrix size used was ; therefore, 16 separate inversions were required with 64 time points along the inversion curve. The order in which the k-space lines are acquired is very important to minimize artifacts caused by amplitude modulation of the k-space data in the phase encode direction. For an interleaved linear coverage of k-space the acquisition order of k-space lines is line number 128 to 121 inclusive for the first pass, 120 to 113 for the second pass, etc. The acquisition order for an interleaved center-out coverage would be line number 0, 1, 1, 2, 2, 3, 3, 4 for the first pass, and 4, 5, 5, 6, 6, 7, 7, 8 for the second. For an interleaved segmented center-out k-space coverage the acquisition order is line number 0, 16, 32, 48, 64, 80, 96, 112 for the first pass, 1, 17, 33, 49, 65, 81, 97, 113 for the second pass, and 1, 17, 33, 49, 65, 81, 97, 113 for the third pass, etc. The amplitude modulation in k-space for these three cases is shown in Fig. 1. The interleaved segmented center-out acquisition scheme (Fig, 1c) clearly shows the least discontinuity errors. The in-plane resolution used was 200 m with the same slice thickness. The field-of-view was 25.6 mm with a matrix size. The echo time (TE) of each line of k-space was 6 ms with a repetition time of 11 ms; 64 time points were taken along the recovery curve with an additional 10-sec delay before subsequent inversion. An adiabatic hyperbolic secant RF pulse was used for the inversion with a 6 ms duration and a 2.6 khz bandwidth. The flip angle used was 3 with a 5 lobe sinc pulse of 1500 Hz bandwidth and 4 ms duration. The acquisition bandwidth used was 33 khz. A 4-ms crusher gradient was applied after the inversion pulse. The inversion thickness was 2 mm. The total acquisition time for four averages was 20 min for a single slice. The images were imported into IDL (Floating Point Systems, Boulder, CO) where maps of relaxation times were generated using a nonlinear curvefitting procedure. The T 1 data was fitted to the standard inversion recovery curve M(t) M 0 (1 2 exp( t/t 1 )) voxel by voxel. A three-parameter fit provided the values of the equilibrium tissue magnetization M 0, the degree of the inversion, and the T 1. T 2 Measurements It is well established that RF pulse imperfections give stimulated echoes in multiple spin-echo sequences leading to inaccurate T 2 measurements (9 11). The use of time-varying spoiler gradients is the most effective method to eliminate the unwanted echoes (11). However, this approach is restricted to single-slice acquisition because it

3 578 Guilfoyle et al. Table 1 Relaxation Measurements of the Phantom Used for Calibration T 1 (ms) T 2 (ms) Stand inv. recovery Segmented flash Multislice T 2 Single-slice T 2 Cylinder Cylinder Cylinder Cylinder 4* *Distilled water. makes use of nonselective refocusing pulses. For our T 2 measurements we used a simple multislice (48 slices) single echo acquisition repeated at six echo times. This is not the most time-efficient method for spatial mapping of T 2 but it is the most accurate. The possible sources of contamination are stimulated echoes from the multislice loop and also during animal experiments nonlinear temporal field shifts could be produced due to respiration and/or heartbeat and potentially affect evolving spin coherences due to multiple RF excitations resulting in spurious echoes (12 14). This effect could occur if TR is on the same order as the T 2 as in a multislice loop, where the repetition time between slices is very short. To address these concerns the T 2 measurements were performed in multislice mode as well as single slice mode in both phantoms and in vivo. A distilled water phantom was used in the phantom measurements. This has a very long T 2 and allows for checking of contamination in measurements from unwanted spin coherences. The in-plane resolution used was 200 m with the same slice thickness. An FOV of 25.6 mm with a matrix size was used. The acquisition bandwidth was 50 khz. The slice selection pulse used was a 1500 Hz 3 lobe sinc of 2.6 ms duration. The six echo times measured were 15, 20, 25, 35, 55, and 75 ms. The repetition time was 4 sec with a single average. The total measurement time for the T 2 acquisitions was therefore 38.4 min ( sec). The T 2 maps were obtained on a voxel-by-voxel basis using nonlinear least-squares fit from the six images taken at each echo time. Animal Preparation Animals were anesthetized with isoflurane (2%) in NO 2 (75%) and O 2 (22%). For maintenance of anesthesia, isoflurane was reduced to 1% with slight correction for body weight. After anesthesia, animals were positioned in a head holder custom-designed to fit inside the imaging coil, consisting of a plastic bite bar on which the animals front teeth were secured, thus minimizing head movement during imaging. The chest and abdomen of the animal outside the RF coil were covered with a circulating water pad made from silicon tubing connected to a heating water bath to monitor and maintain rectal temperature at 37 C 0.5 C throughout imaging. phantom was imaged with a standard inversion recovery sequence repeated 10 times at the following inversion times: 100, 300, 400, 600, 800, 1000, 1500, 2000, 3000, and 5000 ms. The matrix size used was with an FOV of 25.6 mm and a 0.2 mm slice thickness. The repetition time used was 20 sec, which gave a total acquisition time of 2.6 hr. To validate the T 2 measurements the T 2 measurement was repeated in single-slice mode. The results of the comparisons are shown in the Table 1. The mean and SD are from three data sets repeated at different times but using the same ROI. Figure 2a is a segmented snapshot FLASH image of the midsection of a mouse brain taken at the first time point along the inversion recovery curve, 44 ms after the inversion pulse. Figure 2b is the T 1 map generated from the 64 time points along the recovery curve taken in increments of 88 ms. Full white intensity represents 3 sec. Figure 3a shows a spin echo image taken at the same slice TE 15 ms. Figure 3b is the generated T 2 map using six images taken at the following echo times: 15, 20, 25, 35, 55, and 75 ms. Full white intensity represents 100 ms. The relaxation measurements were made in a group of normal 6-week-old mice and in a group of normal 1-yearold mice (six mice in each group). The regional variation in relaxation parameters was analyzed in the corpus callosum, hippocampus, and retrosplenial granular (RSG) cortex. The results of this analysis are shown in Table 2 reporting the mean T 1 and T 2 of all six mice in each group along with their SDs from each brain region. The regions of interest (ROIs) selected in a single slice at position bregma 1.46 mm (15) are shown in Fig. 4a imposed on a spin echo image. Red represents the hippocampus ROI, green the RSG cortex, and blue the corpus callosum. Figure 4b is a bar chart representation of the mean T 1 from each region RESULTS The phantom used for T 1 and the T 2 measurements consisted of four 4-mm cylinders, three containing various concentrations of manganese chloride and one containing distilled water. To validate the T 1 calculations the same FIG. 2. a: FLASH image of the midsection of a mouse brain taken at 44 ms after the inversion. b: Generated T 1 map from 64 images taken in increments of 88 ms along the inversion curve. Full white represents 3 sec. The in-plane resolution was 0.2 mm with the same slice thickness.

4 Relaxation Times in the Mouse Brain at 7.0 T 579 FIG. 3. a: Spin echo image taken at the midsection of a mouse brain withateof15ms.b: Generated T 2 map from six images taken at a TE of 15, 20, 25, 35, 55, and 75 ms. Full white represents 100 ms. The in-plane resolution was 0.2 mm with the same slice thickness. and group. A routine (two-tailed) t-test was applied to these data to determine significant differences between the two age groups. The only statistically significant age dependent difference was seen in the T 2 of the corpus callosum, with a P value of DISCUSSION The T 2 and T 1 acquisition methods used in the mouse protocols have been validated with phantom measurements. There is very close agreement between the T 1 measured with standard inversion recovery and the segmented FLASH sequence used in these experiments. This indicated that the 10-sec recovery delay used allows full relaxation for all T 1 values measured in the mouse brain. Furthermore, there is no evidence of spurious echoes contaminating the T 2 measurements, as the T 2 values measured in single-slice mode are identical to the values measured in multislice mode. This is the case for both phantom and in vivo measurements. The relaxation times reported are of relevance to those interested in optimizing contrast in MR images of the mouse brain at 7 T and a standard for comparison of relaxation measurements made with genetic mouse models such as the transgenic PS-APP amyloid overexpressing mouse. We have reported T 1 values ranging 1503 ms, 1632 ms, and 1767 ms in the corpus callosum, RSG cortex, and hippocampus, respectively, in 6-week-old mice with small but insignificant differences in 1-year-old mice. A T 1 of 1040 ms has been reported in mouse cortex at 2 T (3). T 2 values in this study range from 37 ms, 38 ms, and 41 ms in the corpus callosum, RSG cortex, and hippocampus, respectively, in 6-week-old mice with a significant decrease (P 0.039) to 34.7 ms in the corpus callosum of Table 2 Regional Values of T 1 and T 2 for Young and Old Mice Regional T1 (ms) Hippocampus RSG cortex Corp. callosum Young Old Regional T2 (ms) Young * Old * *P FIG. 4. The ROIs selected in a single slice at position bregma 1.46 mm are shown imposed on a spin echo image (a). Red represents the hippocampus ROI, green the RSG cortex, and blue the corpus callosum. b: Bar chart representation of the mean T 1 from each region and group (dark gray represents young and light gray old). c: Bar chart representation of the mean T 2 from each region and group (dark gray represents young and light gray old). 1-year-old mice. A comparable value of T 2 (38.5 ms) in the frontal cortex of 8 12-month-old mice has been reported elsewhere at 7 T (4). The corpus callosum is predominantly white matter and the myelin sheath could be a relaxation sink by increasing the proportion of bound pro-

5 580 Guilfoyle et al. tons, thereby decreasing T 2. It is possible that younger mice have incompletely myelinated fibers and thus a higher T 2. Another possible reason for age-dependent T 2 decreases in the corpus callosum may be due to shrinkage of cells in this region due to neurodegeneration or apoptosis, a phenomenon with an associated cell shrinkage (16). A reduction in T 2 during cell shrinkage has also been demonstrated in hyperosmolar conditions in the turtle cerebellum (17). In conclusion, T 1 and T 2 times were measured in the mouse brain at 7 T with a degree of high spatial resolution. The only significant difference between infant and mature mice was found to be in T 2 of the corpus callosum. Knowledge of these relaxation times are necessary for contrast optimization, efficient experiment planning, and baseline reference for relaxation times in an increasingly important area of murine imaging. REFERENCES 1. Hoehn-Berlage M, Eis M, Back T, Kohno K, Yamashita K. Changes of relaxation times (T 1,T 2 ) and apparent diffusion coefficient after permanent middle cerebral artery occlusion in the rat: temporal evolution, regional extent and comparison with histology. Magn Reson Med 1995; 34: Calamante F, Lythgoe MF, Pell GS, Thomas DL, King MD, Busza AL, Sotak CH, Williams SR, Ordidge RJ, Gadian DG. Early changes in water diffusion, perfusion, T 1 and T 2 during focal cerebral ischeamia in the rat studied at 8.5T Magn Reson Med 1999;41: Zaharchuk G, Hara H, Huang PL, Fishman MC, Moskowitz MA, Jenkins BG, Rosen BR. Neuronal nitric oxide synthase mutant mice show smaller infarcts and attenuated apparent diffusion coefficient changes in the peri-infarct zone during focal cerebral ischemia. Magn Reson Med 1997;37: Dunn JF, Zaim-Wadghiri Y. Quantitative magnetic resonance imaging of the mdx mouse model of Duchenne muscular dystrophy. Muscle Nerve 1999;22: Clare S, Jezzard P. Rapid T 1 mapping using multislice echo planar imaging. Magn Reson Med : Look DC, Locker DR. Time saving in measurement of NMR and EPR relaxation times. Rev Sci Instrum 1970;41: Haase A, Frahm J, Matthaei D, Hanicke W, Merboldt KD. FLASH imaging. Rapid NMR imaging using low flip-angle pulses. J Magn Reson 1986;67: Labadie C, Lee JH, Vetek G, Springer CS. Relaxographic imaging. J Magn Reson 1994;B105: Kroeker RM, Henkelman RM. Analyis of biological NMR relaxation data with continuous distributions of relaxation times. J Magn Reson 1986;69: Crawley AP, Henkelman RM. Errors in T 2 estimation using multiple echo imaging. Magn Reson Med 1987;4: Poon CS, Henkelman RM. Practical T 2 quantitation for clinical applications. J Magn Reson Imag 1992;2: Zur Y, Stokar S, Bendal P. An analysis of fast imaging sequences with steady-state transverse magnetization refocusing. Magn Reson Med 1988;6: Zur Y, Wood ML, Neuringer LJ. Spoiling of transverse magnetization in steady state sequences. Magn Reson Med 1991;21: Zhao X, Bodurka J, Jesmanowicz A, Li SJ. B 0 fluctuation-induced temporal variation in epi image series due to disturbance of steady-state free precession. Magn Reson Med 2000;44: Franklin KBJ, Paxinos G. The mouse brain in stereotaxic coordinates. New York: Academic Press; Bortner CD, Cidlowski JA. A necessary role for cell shrinkage in apoptosis. Biochem Pharmacol 1998;56: O Shea J, Williams SR, van Bruggen N, Gardner-Medwin AR. Apparent diffusion coefficient and MR relaxation during osmotic manipulation in isolated turtle cerebellum. Magn Reson Med 2000;44: