Abstract
Purpose:
The presence of orientation-dependent susceptibility artifacts in magnetic resonance chemical shift thermometry (CST) can confound accurate temperature calculations. Here, we quantify the effect of white matter (WM) tract orientation on CST due to tissue-specific susceptibility.
Methods:
Twenty-nine healthy volunteers (27±4 years old) were scanned on a 3T MR scanner with a 32-channel head coil. Diffusion tensor imaging (DTI), T1-weighted imaging, and single voxel spectroscopy (SVS) for CST were acquired. Participants were then asked to rotate their head ~3 – 5° (yaw or roll) to alter the orientation of WM tracts relative to the external magnetic field. After head rotation, a second SVS scan and T1-weighted imaging were acquired. The WM-fraction-normalized DTI principal eigenvector (V1) images were used to calculate the length of the x-y component of V1, which was used as a surrogate for WM tracts perpendicular to B0. A linear regression model was used to determine the relationship between the perpendicular WM tracts and brain temperature.
Results:
Significant temperature differences between post- and pre-head rotation scans were observed for brain (−0.72 °C ± 1.36 °C, p = 0.01) but not body (0.012 °C ± 0.07 °C, p = 0.37) temperatures. The difference in brain temperature was positively associated with the corresponding change in perpendicular WM tracts after head rotation (R2 = 0.26, p = 0.005).
Conclusion:
Our results indicate WM tract orientation affects temperature calculations, suggesting artifacts from orientation-dependent susceptibility may be present in CST.
Keywords: brain thermometry, chemical shift thermometry, orientation-dependent susceptibility, magnetic susceptibility
1. Introduction
Brain temperature is a key physiological metric particularly after injury [1-4]. Magnetic resonance (MR) chemical shift thermometry (CST) has been demonstrated extensively as an approach for noninvasive temperature mapping using the chemical shift difference between water and a temperature-independent reference metabolite, e.g., N-acetylaspartate (NAA) [5-10]. Despite its promise as a non-invasive and absolute thermometry method, most CST has been limited to research applications due to a number of challenges.
One challenge with CST in the brain is the presence of tissue-specific magnetic susceptibility and microstructural differences that can confound absolute temperature calculations [11]. Lee et al. described the sensitivity of the proton resonance frequency (PRF) or chemical shift to orientation of tissue microstructure [12]. Maudsley et al. identified differences in the chemical shift difference between water and NAA, creatine, and choline which were attributed to tissue susceptibility differences in white matter (WM) and gray matter (GM) [13]. For NAA, these changes were attributed to WM tract distribution and orientation-dependent susceptibility as NAA is present intracellularly, whereas water is present both intra- and extracellularly. Depending on the WM fiber or tract orientation relative to the external magnetic field, this may result in changes in the NAA chemical shift due to susceptibility rather than temperature.
The magnitude of magnetic susceptibility-induced changes in chemical shift, independent of temperature, has been approximated to be on the order of 14-15 parts per billion (ppb), which corresponds to temperature errors on the order of ~1.4-1.5 °C [11, 14]. He and Yablonskiy postulated WM fibers perpendicular to the external magnetic field (B0) may experience an orientation-dependent susceptibility that can alter the resonance frequency, whereas fibers parallel to B0 are unaffected [14]. Many prior studies examining this effect, however, are either theoretical or focused largely on volume susceptibility, which would affect both water and NAA in a similar manner.
The goal of this study was to quantify the effect of WM fiber orientation on CST calculations in healthy volunteers. We evaluated the effects of head orientation and subsequent WM orientation changes on CST through the analysis of diffusion tensor images, T1-weighted images, and MR spectroscopy in 29 healthy human volunteers. To our knowledge, the presence of orientation-dependent susceptibility effects on CST using NAA as an internal reference has not been explicitly quantified or demonstrated, and we hypothesized this may be an important correction for in vivo CST.
2. Methods
2.1. Study design and MR acquisition
This prospective study was approved by the local Institutional Review Board and all subjects provided written informed consent. Inclusion criteria were subjects who were medically healthy, could undergo a 1-hour MRI scan, and were between 18-45 years of age. Subjects were excluded if they had a history of ischemia, neurodegenerative disease, epilepsy, central nervous system surgery, moderate to severe traumatic brain injury, or contraindications to MRI. A total of 29 healthy volunteers were included in the final analysis (14 female, 15 male; mean ± standard deviation (SD) age = 26 ± 4 years old; range: 20-36 years old). MR data were acquired on a 3T MR scanner (MAGNETOM PrismaFit, Siemens Healthcare, Erlangen, Germany) using a 32-channel phased-array receive head coil. Prior to data acquisition, subjects practiced rotating their head in the MR scanner until they could reliably turn ~3-5°, i.e., rotation around the superior-inferior axis (yaw) or around the posterior-anterior axis (roll), depending on the most comfortable position for the subject. This head rotation was designed to change the orientation of WM fibers relative to B0 and was confirmed using the post-head rotation T1-weighted image. Subjects were instructed to stay awake during the scan, provided a blanket upon request, and acclimated to the temperature in the scanner for >10 minutes prior to MR thermometry. Diffusion tensor imaging (DTI), T1-weighted structural imaging, and single voxel spectroscopy (SVS) used for CST were acquired prior to head rotation. Subjects were then asked to rotate their head, followed immediately by a second SVS scan and T1-weighted image. The time between SVS scans acquired pre- and post-head rotation was typically less than one minute to minimize physiological changes in brain temperature. T1-weighted images were acquired using a magnetization-prepared rapid gradient-echo (MPRAGE) sequence (TR/TI/TE = 2300/900/3.39 ms; flip angle = 9°; FOV = 256×256 mm2; matrix size = 192×192; 160 slices, slice thickness = 1 mm; generalized autocalibrating partial parallel acquisition (GRAPPA) factor = 2; scan time = 4 minutes and 8 s). DTI images were acquired with a gradient echo sequence (TR/TE = 3310/105 ms; flip angle=80°; FOV = 228 ×228 mm2, voxel size = 2×2×2 mm3; 69 slices; GRAPPA acceleration factor = 2; multiband acceleration factor = 3; b-values = 0, 1000, and 2000 s/mm2; phase encoding direction = anterior to posterior; single image acquired with reverse phase encoding of posterior to anterior; scan time = 9 minutes and 36 s [for both anterior-to-posterior and posterior-to-anterior directions]). For MR thermometry, SVS was acquired in a voxel centered on the corpus callosum genu and adjacent cingulate gyrus (Figure 1a, 1b) using the semi localized by adiabatic selective refocusing (sLASER) sequence (TR/TE = 2000/68 ms; flip angle = 90°; averages = 128; scan time = 4 minutes and 40 s; spectral bandwidth = 2000 Hz; complex data points = 2048; voxel size = 2x2x2 cm3; 120 Hz VAPOR water suppression bandwidth; non-water suppressed spectrum acquired with 8 averages). Given the proximity of the corpus callosum to the lateral ventricles, the voxel was positioned to contain only a part of the corpus callosum to avoid artifacts due to the ventricles after head rotation. MR acquisition is outlined in Figure 1c. Body temperature was measured continuously during the scan using an MR-compatible axillary probe, and mean values during each SVS acquisition are reported.
Figure 1.
Magnetic resonance spectroscopy (MRS) voxel position (yellow square) overlaid on the T1-weighted image for one subject (a) before and (b) after head rotation. (c) Schematic of MR scan protocol. DTI = diffusion tensor imaging.
2.2. MR image analysis
FSL v6.0.3 was used in conjunction with FMRIB’s Diffusion Toolbox [15] to analyze DTI data. TOPUP was used to correct susceptibility-induced currents using the reverse phase encoding image [16]. The B0 image was skull-stripped using the brain extraction tool (BET, v2.1) in FSL, followed by eddy current correction with eddy cuda (v9.1) and diffusion tensor analysis using DTIFIT [17]. The DTI principal eigenvector (V1) image was registered to the corresponding skull-stripped T1-weighted image from each subject using the spatial coregister function in SPM12 (https://www.fil.ion.ucl.ac.uk/spm/). MRS voxel masks were generated using SPM by registering the position of the MRS voxel onto the corresponding T1-weighted image.
2.3. Determination of orientation and susceptibility effects on brain temperature
To determine the orientation of WM in the MRS voxel, the DTI V1 images were normalized by multiplying the V1 value by the WM fraction voxel-wise. The mean x, y, and z components of the normalized V1 were then calculated for all DTI voxels contained within the larger MRS voxel. The Euclidean length of the x-y component of V1 was used as a surrogate for WM tracts perpendicular to B0, as the z component of V1 represents the superior-inferior direction which is parallel to B0. Temperature was calculated using the chemical shift difference between NAA and water as previously described [3] after fitting spectra in LCModel [18]. A paired t-test was used to determine differences in brain and body temperature before and after head rotation. Linear regression was used to determine the association between temperature differences (response) and change in perpendicular WM tracts (predictor) for both brain temperature and brain-body temperature differences (post-head rotation minus pre-head rotation). For all analyses, significance was determined at p≤0.05 and values are reported as the mean ± standard deviation (SD) unless otherwise noted.
3. Results and Discussion
Subject characteristics are shown in Table 1. On average, estimated brain temperature changed by −0.72 ± 1.36 °C (paired t-test: t = −2.84, p = 0.01), while body temperature was nearly constant (0.012 °C ± 0.07 °C, paired t-test: t = 0.91, p = 0.37) between scans (post-head rotation minus pre-head rotation) (Table 2). Given the short time frame between acquisitions (~one minute), physiological fluctuations, particularly in brain temperature, are expected to be minimal suggesting this change may be driven by other factors. This is further supported by our recent work using repeated measurements without head rotation, which resulted in no significant differences in brain temperature on the minutes time scale [19]. Body temperature changes were not significantly different between scans, further supporting minimal physiological or environmental changes. The difference in brain temperature was positively and significantly associated with the corresponding difference in perpendicular WM tracts after head rotation (R2 = 0.26, slope = 7.25, t = 3.08, p = 0.005), suggesting ~26% of the brain temperature change was explained by orientation-dependent changes in susceptibility (Figure 2). Similar results were observed for brain-body temperature differences (R2 = 0.25, slope = 7.06, t = 2.98, p = 0.006).
Table 1.
Subject characteristics including age (in years at the time of the scan), biological sex, and self-reported race. Subject 19 also identified as Hispanic ethnicity.
| Subject | Age | Sex | Race |
|---|---|---|---|
| 1 | 31 | M | Caucasian |
| 2 | 23 | F | African American |
| 3 | 25 | F | Caucasian |
| 4 | 24 | F | Caucasian |
| 5 | 31 | M | Asian |
| 6 | 26 | M | Caucasian |
| 7 | 26 | F | Asian |
| 8 | 28 | M | African American |
| 9 | 28 | M | Asian |
| 10 | 26 | M | Caucasian |
| 11 | 28 | M | Asian |
| 12 | 29 | M | Caucasian |
| 13 | 27 | M | African American |
| 14 | 27 | F | African American |
| 15 | 20 | F | Asian |
| 16 | 31 | F | Caucasian |
| 17 | 21 | M | Caucasian |
| 18 | 28 | M | Caucasian |
| 19 | 36 | F | Caucasian |
| 20 | 24 | M | Asian |
| 21 | 29 | M | Caucasian |
| 22 | 20 | M | African American |
| 23 | 24 | F | Caucasian |
| 24 | 29 | F | Caucasian |
| 25 | 25 | F | Other |
| 26 | 21 | M | Caucasian |
| 27 | 26 | M | Asian |
| 28 | 23 | F | Asian |
| 29 | 29 | F | Caucasian |
M = Male; F = Female
Table 2.
Brain and body temperatures and temperature differences for all subjects.
| Subject | Brain T Pre |
Brain T Post |
Brain T Diff |
Body T Pre |
Body T Post |
Body T Diff |
|---|---|---|---|---|---|---|
| 1 | 37.4 | 35.2 | −2.2 | 36.3 | 36.1 | −0.2 |
| 2 | 37.5 | 37.0 | −0.5 | 36.3 | 36.2 | −0.1 |
| 3 | 33.7 | 35.1 | 1.4 | 36.8 | 36.7 | −0.1 |
| 4 | 37.1 | 34.4 | −2.7 | 37.1 | 37.1 | −0.0 |
| 5 | 37.7 | 35.6 | −2.1 | 36.9 | 37.0 | 0.1 |
| 6 | 37.4 | 36.4 | −0.9 | 37.0 | 36.9 | −0.1 |
| 7 | 36.3 | 36.1 | −0.2 | 36.7 | 36.7 | −0.0 |
| 8 | 36.1 | 34.9 | −1.2 | 35.8 | 35.8 | −0.0 |
| 9 | 37.4 | 36.3 | −1.1 | 36.4 | 36.4 | −0.0 |
| 10 | 36.7 | 34.3 | −2.5 | 36.7 | 36.7 | −0.0 |
| 11 | 36.4 | 36.8 | 0.4 | 37.4 | 37.3 | −0.1 |
| 12 | 34.7 | 35.2 | 0.6 | 36.5 | 36.5 | −0.0 |
| 13 | 36.3 | 36.4 | 0.1 | 37.0 | 37.0 | −0.0 |
| 14 | 37.1 | 36.5 | −0.6 | 36.6 | 36.6 | −0.0 |
| 15 | 36.8 | 34.8 | −1.9 | 36.0 | 36.0 | −0.0 |
| 16 | 36.9 | 37.3 | 0.4 | 36.5 | 36.5 | −0.0 |
| 17 | 36.9 | 36.5 | −0.3 | 35.8 | 35.8 | −0.0 |
| 18 | 36.4 | 34.5 | −1.9 | 36.2 | 36.4 | 0.2 |
| 19 | 37.7 | 37.1 | −0.7 | 36.0 | 36.0 | −0.0 |
| 20 | 36.1 | 33.7 | −2.4 | 36.0 | 36.0 | −0.0 |
| 21 | 36.6 | 34.5 | −2.0 | 36.4 | 36.5 | 0.1 |
| 22 | 36.1 | 36.8 | 0.7 | 36.9 | 36.8 | −0.1 |
| 23 | 35.7 | 36.7 | 1.0 | 37.5 | 37.6 | 0.1 |
| 24 | 35.5 | 33.3 | −2.2 | 36.8 | 36.8 | −0.0 |
| 25 | 36.0 | 37.9 | 1.9 | 36.4 | 36.5 | 0.1 |
| 26 | 35.4 | 36.2 | 0.8 | 36.1 | 36.2 | 0.1 |
| 27 | 37.3 | 38.3 | 1.0 | 35.9 | 36.0 | 0.1 |
| 28 | 36.4 | 34.2 | −2.2 | 37.7 | 37.7 | −0.0 |
| 29 | 35.7 | 34.2 | −1.5 | 37.3 | 37.3 | −0.0 |
Values reported as 0.0 indicate those <0.05. T = temperature (°C); Pre = before head rotation; Post = after head rotation; Diff = Post-head rotation temperature minus pre-head rotation temperature
Figure 2.
Association between the change in brain temperature and the change in perpendicular white matter tracts after head rotation (post-head rotation minus pre-head rotation). Raw data (closed circles) and linear regression best fit line are displayed.
The remaining variation in temperature differences is attributed to variations in tissue composition and voxel placement after the head tilt. While the specific absorption rate (SAR) could result in tissue heating, the second temperature measurement was, on average, lower than the first measurement, suggesting temperature changes are likely not due to SAR effects. Given the close position of the corpus callosum fiber bundle to the ventricles, it was not possible to position the voxel to cover only the corpus callosum. We acknowledge the majority of temperature variation is likely due to voxel position; however, our results support orientation-dependent susceptibility effects as a significant factor in CST calculations consistent with prior studies [11-14]. Importantly, while the surrounding tissue varied, the predominant WM tract (i.e., corpus callosum) contained within the MRS voxel was kept largely constant to isolate the effect of WM orientation on calculated temperature.
These data suggest WM orientation, in addition to fractional tissue composition, may be an important factor in future brain CST studies. Particularly when fiber or tracts are damaged after injury or illness, these corrections may be more important. It also stands to reason some differences in temperature between studies [1, 11-13] are due to orientation artifacts which affected CST calculations. While our goal was to provide evidence for the presence of orientation-dependent susceptibility artifacts, we acknowledge the presence of other factors which contribute to observed and real (physiological) variation in brain temperature. Given known spatial gradients in brain temperature [3, 5, 8, 20], tissue composition and MRS voxel position likely contributed to temperature differences before and after head rotation [11-13, 21].
4. Conclusions
We observed WM fiber orientation can significantly affect CST calculations and may confound absolute temperature values. Considerations of orientation-dependent susceptibility artifacts may be an important consideration for CST, particularly in situations where WM fiber tracts are damaged or in reproducibility studies where voxel placement may vary.
Acknowledgments:
All imaging experiments were supported by the Emory Center for Systems Imaging Core.
Funding:
This work was supported by the National Institutes of Health (grant number R21EB029622) and a seed grant from the Emory University Department of Radiology and Imaging Sciences.
Abbreviations
- MR
Magnetic Resonance
- CST
chemical shift thermometry
- NAA
N-acetylaspartate
- PRF
proton resonance frequency
- WM
white matter
- GM
gray matter
- ppb
parts per billion
- SD
standard deviation
- DTI
diffusion tensor imaging
- SVS
single voxel spectroscopy
- MPRAGE
magnetization-prepared rapid gradient-echo
- FOV
field of view
- sLASER
semi localized by adiabatic selective refocusing
- VAPOR
Variable power radiofrequency pulses with optimized relaxation delays
Footnotes
Declarations of interest: none
References
- [1].Wang H, Wang B, Normoyle KP, Jackson K, Spitler K, Sharrock MF, et al. Brain temperature and its fundamental properties: a review for clinical neuroscientists. Front Neurosci 2014;8:307–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Busto R, Dietrich WD, Globus M, Ginsberg MD. The importance of brain temperature in cerebral ischemic injury. Stroke 1989;20(8):1113–4. [DOI] [PubMed] [Google Scholar]
- [3].Fleischer CC, Wu J, Qiu D, Park S-E, Nahab F, Dehkharghani S. The brain thermal response as a potential neuroimaging biomarker of cerebrovascular impairment. Am J Neuroradiol 2017;38(11):2044–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Soukup J, Zauner A, Doppenberg EM, Menzel M, Gilman C, Young HF, et al. The importance of brain temperature in patients after severe head injury: relationship to intracranial pressure, cerebral perfusion pressure, cerebral blood flow, and outcome. J Neurotrauma 2002;19(5):559–71. [DOI] [PubMed] [Google Scholar]
- [5].Dehkharghani S, Fleischer CC, Qiu D, Yepes M, Tong F. Cerebral temperature dysregulation: MR thermographic monitoring in a nonhuman primate study of acute ischemic stroke. Am J Neuroradiol 2017;38(4):712–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Kuroda K. Non-invasive MR thermography using the water proton chemical shift. Int J Hyperthermia 2005;21(6):547–60. [DOI] [PubMed] [Google Scholar]
- [7].Rieke V, Butts Pauly K. MR thermometry. J Magn Reson Imaging 2008;27(2):376–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Dehkharghani S, Mao H, Howell L, Zhang X, Pate K, Magrath P, et al. Proton resonance frequency chemical shift thermometry: experimental design and validation toward high-resolution noninvasive temperature monitoring and in vivo experience in a nonhuman primate model of acute ischemic stroke. Am J Neuroradiol 2015;36(6):1128–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Cady EB, D'Souza PC, Penrice J, Lorek A. The estimation of local brain temperature by in vivo 1H magnetic resonance spectroscopy. Magn Reson Med 1995;33(6):862–7. [DOI] [PubMed] [Google Scholar]
- [10].Corbett RJ, Laptook AR, Tollefsbol G, Kim B. Validation of a noninvasive method to measure brain temperature in vivo using 1H NMR spectroscopy. J Neurochem 1995;64(3):1224–30. [DOI] [PubMed] [Google Scholar]
- [11].Chadzynski GL, Bender B, Groeger A, Erb M, Klose U. Tissue specific resonance frequencies of water and metabolites within the human brain. J Magn Reson 2011;212(1):55–63. [DOI] [PubMed] [Google Scholar]
- [12].Lee J, Shmueli K, Fukunaga M, van Gelderen P, Merkle H, Silva AC, et al. Sensitivity of MRI resonance frequency to the orientation of brain tissue microstructure. Proceedings of the National Academy of Sciences 2010;107(11):5130–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Maudsley AA, Goryawala MZ, Sheriff S. Effects of tissue susceptibility on brain temperature mapping. Neuroimage 2017;146:1093–101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].He X, Yablonskiy DA. Biophysical mechanisms of phase contrast in gradient echo MRI. Proc Natl Acad Sci USA 2009;106(32):13558–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].Woolrich MW, Jbabdi S, Patenaude B, Chappell M, Makni S, Behrens T, et al. Bayesian analysis of neuroimaging data in FSL. Neuroimage 2009;45(1):S173–S86. [DOI] [PubMed] [Google Scholar]
- [16].Andersson JL, Skare S, Ashburner J. How to correct susceptibility distortions in spin-echo echo-planar images: application to diffusion tensor imaging. Neuroimage 2003;20(2):870–88. [DOI] [PubMed] [Google Scholar]
- [17].Jenkinson M, Beckmann CF, Behrens TE, Woolrich MW, Smith SM. Fsl. Neuroimage 2012;62(2):782–90. [DOI] [PubMed] [Google Scholar]
- [18].Provencher SW. Estimation of metabolite concentrations from localized in vivo proton NMR spectra. Magn Reson Med 1993;30(6):672–9. [DOI] [PubMed] [Google Scholar]
- [19].Sung D, Risk BB, Wang KJ, Allen JW, Fleischer CC. Resting-State Brain Temperature: Dynamic Fluctuations in Brain Temperature and the Brain–Body Temperature Gradient. J Magn Reson Imag 2022; 10.1002/jmri.28376(n/a). [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Sung D, Kottke PA, Risk BB, Allen JW, Nahab F, Fedorov AG, et al. Personalized predictions and non-invasive imaging of human brain temperature. Commun Phys 2021;4(1):1–10. [Google Scholar]
- [21].Yablonskii DA, Ackerman JJ, Raichle ME. Coupling between changes in human brain temperature and oxidative metabolism during prolonged visual stimulation. Proc Natl Acad Sci USA 2000;97(13):7603–8. [DOI] [PMC free article] [PubMed] [Google Scholar]