Improving the reproducibility of proton magnetic resonance spectroscopy brain thermometry: Theoretical and empirical approaches

In proton magnetic resonance spectroscopy (¹ H MRS)-based thermometry of brain, averaging temperatures measured from more than one reference peak offers several advantages, including improving the reproducibility (i.e., precision) of the measurement. This paper proposes theoretically and empirically optimal weighting factors to improve the weighted average of temperatures measured from three references. We first proposed concepts of equivalent noise and equivalent signal-to-noise ratio in terms of frequency measurement and a concept of relative frequency that allows the combination of different peaks in a spectrum for improving the precision of frequency measurement. Based on these, we then derived a theoretically optimal weighting factor and proposed an empirical weighting factor, both involving equivalent noise levels, for a weighted average of temperatures measured from three references (i.e., the singlets of NAA, Cr, and Ch in the ¹ H MR spectrum). We assessed these two weighting factors by comparing their errors in measurement of temperatures with the errors of temperatures measured from individual references; we also compared these two new weighting factors with two previously proposed weighting factors. These errors were defined as the standard deviations in repeated measurements or in Monte Carlo studies. Both the proposed theoretical and empirical weighting factors outperformed the two previously proposed weighting factors as well as the three individual references in all phantom and in vivo experiments. In phantom experiments with 4- or 10-Hz line broadening, the theoretical weighting factor outperformed the empirical one, but the latter was superior in all other repeated and Monte Carlo tests performed on phantom and in vivo data. The proposed weighting factors are superior to the two previously proposed weighting factors and can improve the reproducibility of temperature measurement using ¹ H MRS-based thermometry.

Does cerebrospinal fluid pulsation affect diffusion-weighted imaging thermometry? A study in healthy volunteers

Diffusion-weighted imaging (DWI)-based thermometry offers potential as a noninvasive method for measuring temperatures deep inside the human brain. However, DWI might be influenced by the pulsatile flow of cerebrospinal fluid (CSF). This study aimed to investigate the influence of such pulsations on DWI thermometry in healthy individuals. A total of 104 participants (50 men, 54 women; mean [± standard deviation] age, 44.2 ± 14.3 years; range 21-69 years) were investigated. DWI-based brain temperature (T_DWI ) was acquired at three speeds (maximum and minimum speeds of ascending flow and random timing at the cerebral aqueduct) of CSF pulsation using a 3-T magnetic resonance imaging scanner. Magnetic resonance spectroscopy (MRS)-based temperature (T_MRS ) at the thalamus was also obtained as a reference standard for brain temperature. The three different CSF pulsatile flows were monitored by heart rate during the scan. The difference between reference temperature and brain temperature (ΔT = T_DWI - T_MRS ) along with the three CSF speeds were statistically compared using Student's matched pair t-test. No significant difference in ΔT was evident among CSF speeds (p > 0.05). No significant linear correlation between ΔT and CSF flow speed at the cerebral aqueduct was observed. Using DWI thermometry with clinical acquisition settings, which utilizes mean values within thresholds, no effect of CSF pulsation speed was observed in the estimation of ΔT.

Proton magnetic resonance spectroscopy thermometry: Impact of separately acquired full water or partially suppressed water data on quantification and measurement error

In proton magnetic resonance spectroscopy (¹ H MRS) thermometry, separately acquired full water and partially suppressed water are commonly used for measuring temperature. This paper compares these two approaches. Single-voxel ¹ H MRS data were collected on a 3-T GE scanner from 26 human subjects. Every subject underwent five continuous MRS sessions, each separated by a 2-min phase. Each MRS session lasted 13 min and consisted of two free induction decays (FIDs) without water suppression (with full water [FW or w]) and 64 FIDs with partial water suppression (with partially suppressed water [PW or w']). Frequency differences between the two FWs, the first two PWs, the second FW and the first PW (FW₂ , PW₁ ), or between averaged water ( wav' ) and N-acetylaspartate (NAA), were measured. Intrasubject and intersubject variations of the frequency differences were used as a metric for the error in temperature measurement. The intrasubject variations of frequency differences between FW₂ and PW₁fw2-fw1' , calculated from the five MRS sessions for each subject, were larger than those between the two FWs or between the first two PWs (p = 1.54 x 10^-4 and p = 1.72 x 10^-4 , respectively). The mean values of intrasubject variations of fw2-fw1' for all subjects were 4.7 and 4.5 times those of fw2-fw1 and fw2'-fw1' , respectively. The intrasubject variations of the temperatures based on frequency differences, fw2-fNAA or ( fw1'-fNAA ), were about 2.5 times greater than those based on averaged water and NAA frequencies (fwav'-fNAA ). The mean temperature measured from (fwav'-fNAA ) (n = 26) was 0.29°C lower than that measured from fw2-fNAA and was 0.83°C higher than that from ( fw1'-fNAA ). It was concluded that the use of separately acquired unsuppressed or partially suppressed water signals may result in large errors in frequency and, consequently, temperature measurement.

Effects of Tissue Temperature and Injury on ADC during Therapeutic Hypothermia in Newborn Hypoxic-Ischemic Encephalopathy

BACKGROUND AND PURPOSE

ADC changes are useful in detecting ischemic brain injury, but mechanisms other than tissue pathology may affect the kinetic movement and diffusion of water molecules. We aimed to determine the effects of brain temperature on the corresponding ADC in infants undergoing therapeutic hypothermia.

MATERIALS AND METHODS

Brain temperature and ADC values in the basal ganglia, thalamus, cortical GM, and WM were analyzed during and after therapeutic hypothermia. The study cohort was categorized as having no-injury or injury. Among infants without injury, the correlation between ADC values and temperature was analyzed using the Pearson correlation. Intrasubject comparison of ADC changes during and after therapeutic hypothermia were analyzed, excluding patients who had an MR image interval of >5 days to minimize the effects of injury evolution.

RESULTS

Thirty-nine infants with hypoxic-ischemic encephalopathy were enrolled (23 no-injury; 16 injury). The median ADC was significantly lower during therapeutic hypothermia (837; interquartile range, 771-928, versus 906; interquartile range, 844-1032 ×10^-6mm²/s; P < .001). There was no difference in the ADC between the no-injury and injury groups during therapeutic hypothermia (823; interquartile range, 782-868, versus 842; interquartile range, 770-1008 ×10^-6mm²/s; P  = .4). In the no-injury group, in which ADC is presumed least affected by the evolution of injury, the median ADC was significantly lower during therapeutic hypothermia (826; interquartile range, 771-866, versus 897; interquartile range, 846-936 ×10^-6mm²/s; P  < .001). There was a moderate correlation between temperature and ADC in the no-injury group (during therapeutic hypothermia: Spearman ρ, 0.48; P  < .001; after therapeutic hypothermia: ρ, 0.4; P  < .001).

CONCLUSIONS

Aside from brain injury, reduced tissue temperature may also contribute to diffusion restriction on MR imaging in infants undergoing therapeutic hypothermia.

Brain temperature monitoring in newborn infants: Current methodologies and prospects

Brain tissue temperature is a dynamic balance between heat generation from metabolism, passive loss of energy to the environment, and thermoregulatory processes such as perfusion. Perinatal brain injuries, particularly neonatal encephalopathy, and seizures, have a significant impact on the metabolic and haemodynamic state of the developing brain, and thereby likely induce changes in brain temperature. In healthy newborn brains, brain temperature is higher than the core temperature. Magnetic resonance spectroscopy (MRS) has been used as a viable, non-invasive tool to measure temperature in the newborn brain with a reported accuracy of up to 0.2 degrees Celcius and a precision of 0.3 degrees Celcius. This measurement is based on the separation of chemical shifts between the temperature-sensitive water peaks and temperature-insensitive singlet metabolite peaks. MRS thermometry requires transport to an MRI scanner and a lengthy single-point measurement. Optical monitoring, using near infrared spectroscopy (NIRS), offers an alternative which overcomes this limitation in its ability to monitor newborn brain tissue temperature continuously at the cot side in real-time. Near infrared spectroscopy uses linear temperature-dependent changes in water absorption spectra in the near infrared range to estimate the tissue temperature. This review focuses on the currently available methodologies and their viability for accurate measurement, the potential benefits of monitoring newborn brain temperature in the neonatal intensive care unit, and the important challenges that still need to be addressed.

Brain Temperature Measured by Magnetic Resonance Spectroscopy to Predict Clinical Outcome in Patients with Infarction

Acute ischemic stroke is characterized by dynamic changes in metabolism and hemodynamics, which can affect brain temperature. We used proton magnetic resonance (MR) spectroscopy under everyday clinical settings to measure brain temperature in seven patients with internal carotid artery occlusion to explore the relationship between lesion temperature and clinical course. Regions of interest were selected in the infarct area and the corresponding contralateral region. Single-voxel MR spectroscopy was performed using the following parameters: 2000-ms repetition time, 144-ms echo time, and 128 excitations. Brain temperature was calculated from the chemical shift between water and N-acetyl aspartate, choline-containing compounds, or creatine phosphate. Within 48 h of onset, compared with the contralateral region temperature, brain temperature in the ischemic lesion was lower in five patients and higher in two patients. Severe brain swelling occurred subsequently in three of the five patients with lower lesion temperatures, but in neither of the two patients with higher lesion temperatures. The use of proton MR spectroscopy to measure brain temperature in patients with internal carotid artery occlusion may predict brain swelling and subsequent motor deficits, allowing for more effective early surgical intervention. Moreover, our methodology allows for MR spectroscopy to be used in everyday clinical settings.

MR Thermometry in Cerebrovascular Disease: Physiologic Basis, Hemodynamic Dependence, and a New Frontier in Stroke Imaging

The remarkable temperature sensitivity of the brain is widely recognized and has been studied for its role in the potentiation of ischemic and other neurologic injuries. Pyrexia frequently complicates large-vessel acute ischemic stroke and develops commonly in critically ill neurologic patients; the profound sensitivity of the brain even to minor intraischemic temperature changes, together with the discovery of brain-to-systemic as well as intracerebral temperature gradients, has thus compelled the exploration of cerebral thermoregulation and uncovered its immutable dependence on cerebral blood flow. A lack of pragmatic and noninvasive tools for spatially and temporally resolved brain thermometry has historically restricted empiric study of cerebral temperature homeostasis; however, MR thermometry (MRT) leveraging temperature-sensitive nuclear magnetic resonance phenomena is well-suited to bridging this long-standing gap. This review aims to introduce the reader to the following: 1) fundamental aspects of cerebral thermoregulation, 2) the physical basis of noninvasive MRT, and 3) the physiologic interdependence of cerebral temperature, perfusion, metabolism, and viability.

The Brain Thermal Response as a Potential Neuroimaging Biomarker of Cerebrovascular Impairment

BACKGROUND AND PURPOSE

Brain temperature is critical for homeostasis, relating intimately to cerebral perfusion and metabolism. Cerebral thermometry is historically challenged by the cost and invasiveness of clinical and laboratory methodologies. We propose the use of noninvasive MR thermometry in patients with cerebrovascular disease, hypothesizing the presence of a measurable brain thermal response reflecting the tissue hemodynamic state.

MATERIALS AND METHODS

Contemporaneous imaging and MR thermometry were performed in 10 patients (32-68 years of age) undergoing acetazolamide challenge for chronic, anterior circulation steno-occlusive disease. Cerebrovascular reactivity was calculated with blood oxygen level-dependent imaging and arterial spin-labeling methods. Brain temperature was calculated pre- and post-acetazolamide using previously established chemical shift thermometry. Mixed-effects models of the voxelwise relationships between the brain thermal response and cerebrovascular reactivity were computed, and the significance of model coefficients was determined with an F test (P < .05).

RESULTS

We observed significant, voxelwise quadratic relationships between cerebrovascular reactivity from blood oxygen level-dependent imaging and the brain thermal response (x coefficient = 0.052, P < .001; x²coefficient = 0.0068, P < .001) and baseline brain temperatures (x coefficient = 0.59, P = .008; x² coefficient = -0.13, P < .001). A significant linear relationship was observed for the brain thermal response with cerebrovascular reactivity from arterial spin-labeling (P = .001).

CONCLUSIONS

The findings support the presence of a brain thermal response exhibiting complex but significant interactions with tissue hemodynamics, which we posit to reflect a relative balance of heat-producing versus heat-dissipating tissue states. The brain thermal response is a potential noninvasive biomarker for cerebrovascular impairment.

Prediction of brain tissue temperature using near-infrared spectroscopy

Broadband near-infrared spectroscopy (NIRS) can provide an endogenous indicator of tissue temperature based on the temperature dependence of the water absorption spectrum. We describe a first evaluation of the calibration and prediction of brain tissue temperature obtained during hypothermia in newborn piglets (animal dataset) and rewarming in newborn infants (human dataset) based on measured body (rectal) temperature. The calibration using partial least squares regression proved to be a reliable method to predict brain tissue temperature with respect to core body temperature in the wavelength interval of 720 to 880 nm with a strong mean predictive power of [Formula: see text] (animal dataset) and [Formula: see text] (human dataset). In addition, we applied regression receiver operating characteristic curves for the first time to evaluate the temperature prediction, which provided an overall mean error bias between NIRS predicted brain temperature and body temperature of [Formula: see text] (animal dataset) and [Formula: see text] (human dataset). We discuss main methodological aspects, particularly the well-known aspect of over- versus underestimation between brain and body temperature, which is relevant for potential clinical applications.

Human Brown Adipose Tissue Temperature and Fat Fraction Are Related to Its Metabolic Activity

BACKGROUND AND AIM

The metabolic activity of human brown adipose tissue (BAT) has been previously examined using positron emission tomography (PET). The aim of this study was to use proton magnetic resonance spectroscopy (1H MRS) to investigate whether the temperature and the fat fraction (FF) of BAT and white adipose tissue (WAT) are associated with BAT metabolic activity determined by deoxy-2-18F-fluoro-d-glucose (18F-FDG)-PET.

MATERIALS AND METHODS

Ten healthy subjects (four women, six men; 25 to 45 years of age) were studied using PET-magnetic resonance imaging during acute cold exposure and at ambient room temperature. BAT and subcutaneous WAT 1H MRS were measured. The tissue temperature and the FF were derived from the spectra. Tissue metabolic activity was studied through glucose uptake using dynamic FDG PET scanning during cold exposure. A 2-hour hyperinsulinemic euglycemic clamp was performed on eight subjects.

RESULTS

The metabolic activity of BAT associated directly with the heat production capacity and inversely with the FF of the tissue. In addition, the lipid-burning capacity of BAT associated with whole-body insulin sensitivity. During cold exposure, the FF of BAT was lower than at room temperature, and cold-induced FF of BAT associated inversely with high-density lipoprotein and directly with low-density lipoprotein cholesterol.

CONCLUSION

Both 1H MRS-derived temperature and FF are promising methods to study BAT activity noninvasively. The association between the lipid-burning capacity of BAT and whole-body insulin sensitivity emphasizes the role of BAT in glucose handling. Furthermore, the relation of FF to high-density lipoprotein and low-density lipoprotein cholesterol suggests that BAT has a role in lipid clearance, thus protecting tissues from excess lipid load.

Brain temperature measured by 1H-magnetic resonance spectroscopy in acute and subacute carbon monoxide poisoning

INTRODUCTION

Brain temperature (BT) is associated with the balance between cerebral blood flow and metabolism according to the "heat-removal" theory. The present study investigated whether BT is abnormally altered in acute and subacute CO-poisoned patients by using (1)H-magnetic resonance spectroscopy (MRS).

METHODS

Eight adult CO-poisoned patients underwent 3-T magnetic resonance imaging in the acute and subacute phases after CO exposure. MRS was performed on deep cerebral white matter in the centrum semiovale, and MRS-based BT was estimated by the chemical shift difference between water and the N-acetyl aspartate signal. We defined the mean BT + 1.96 standard deviations of the BT in 15 healthy controls as the cutoff value for abnormal BT increases (p < 0.05) in CO-poisoned patients.

RESULTS

BT of CO-poisoned patients in both the acute and subacute phases was significantly higher than that of the healthy control group. However, BT in the subacute phase was significantly lower than in the acute phase. On the other hand, no significant difference in body temperature was observed between acute and subacute CO-poisoned patients. BT weakly correlated with body temperature, but this correlation was not statistically significant (rho = 0.304, p = 0.2909).

CONCLUSIONS

The present results suggest that BT in CO-poisoned patients is abnormally high in the acute phase and remains abnormal in the subacute phase. BT alteration in these patients may be associated with brain perfusion and metabolism rather than other factors such as systemic inflammation and body temperature.

Hyperperfusion syndrome after stent/coiling of a ruptured carotid bifurcation aneurysm

The authors report a syndrome of regional, symptomatic cerebral hyperperfusion, and edema mimicking infarction in a 54-year-old woman following coiling of a ruptured right carotid bifurcation aneurysm and stenting of the right middle cerebral artery. The patient presented with a Hunt and Hess grade III subarachnoid hemorrhage 7 days after developing thunderclap headache. She underwent successful coiling under general anesthesia of the 1.6 × 1.5 × 1.6 cm aneurysm, but immediately after the coil was placed occlusion of the proximal M1 segment was developed. This occlusion was stented after ~5-min delay, and flow restored without angiographic evidence of distal emboli. Following the procedure, she was extubated and noted to have left hemiparesis, neglect, and mutism without a CT correlate. Cerebral infarction was suspected, but urgent repeat angiography demonstrated patent cerebral vasculature. On the following day, symptoms persisted, and non-contrast head CT now showed cerebral edema localized to the right middle cerebral artery territory mimicking subacute infarction. CT perfusion imaging and angiography showed a widely patent MCA circulation, and suggested a regional hyperperfusion syndrome. The blood pressure was incrementally lowered, with rapid and sustained neurological improvement. Hyperperfusion events following aneurysm repair and related circumstances are reviewed.

A computational model study of the influence of the anatomy of the circle of willis on cerebral hyperperfusion following carotid artery surgery

BACKGROUND

Cerebral hyperperfusion syndrome develops in a small subset of patients following carotid artery surgery (CAS) performed to treat severe carotid artery stenosis. This syndrome has been found to have a close correlation with cerebral hyperperfusion occurring after CAS. The purpose of this study is to investigate whether and how the anatomy of the Circle of Willis (CoW) of the cerebral circulation influences post-CAS cerebral hyperperfusion.

METHODS

A computational model of the cerebral circulation coupled with the global cardiovascular system has been developed to investigate hemodynamic events associated with CAS. Nine topological structures of the CoW were investigated in combination with various distribution patterns of stenosis in the feeding arteries of the cerebral circulation.

RESULTS

The occurrence of post-CAS cerebral hyperperfusion was predicted for the CoW structures that have poor collateral pathways between the stenosed cerebral feeding arteries and the remaining normal feeding arteries. The risk and the localization of post-CAS hyperperfusion were determined jointly by the anatomy of the CoW and the distribution pattern of stenosis in the cerebral feeding arteries. The presence of basilar artery stenosis or contralateral ICA stenosis increased the risk of post-CAS hyperperfusion and enlarged the cerebral region affected by hyperperfusion. For a certain CoW structure, the diameters of the cerebral communicating arteries and the severity of carotid artery stenosis both had a significant influence on the computed post-CAS cerebral hyperperfusion rates. Moreover, post-CAS cerebral hyperperfusion was predicted to be accompanied with an excessively high capillary transmural pressure.

CONCLUSIONS

This study demonstrated the importance of considering the anatomy of the CoW in assessing the risk of post-CAS cerebral hyperperfusion. Particularly, since the anatomy of the CoW and the distribution pattern of stenosis in the cerebral feeding arteries jointly determine the risk and localization of post-CAS cerebral hyperperfusion, a patient-specific hemodynamic analysis aimed to help physicians identify patients at high risk of cerebral hyperperfusion should account for the combined effect of the anatomy of cerebral arteries and cerebral feeding artery stenoses on cerebral hemodynamics.