INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE BRAIN IS ONE OF THE MOST active of all organs in the mammalian body in terms of energy metabolism (40, 41). Increased neuronal activity in the cerebral cortex leads to an increase in metabolic activity and blood flow (3). A net change in brain temperature (T) depends on two dominant processes: heat production during metabolic activity and heat removal by blood flow (38). The ability to measure brain T and relate changes to metabolic activity without altering tissue structure and integrity is important, given the high degree of microvascular network and organization in the brain (54).

The classic way to measure absolute T in living organisms is by thermocouple wires (55) directly inserted into the region of interest (ROI). This approach has been extensively applied for studying brain T regulation and dynamics in animal subjects (14, 29, 45, 47) and human patients (8, 27, 42, 56). Because chronic implantation of thermocouple wires can lead to complications (e.g., infections; see Ref. 13), the method is best suited for acute measurements. Although the spatial resolution of the method can be extended to the micrometer range, the temporal resolution cannot be improved more than a few tenths of a second (e.g., see Refs. 22, 23). Thin thermocouple wires are ideally suited for combining with other devices (e.g., oxygen partial pressure or electrophysiology) for multimodal measurements of brain physiology (22). Although the tissue damage caused by the insertion of the thermocouple wire into the ROI can be reduced by miniaturization of the probe (23), alternative measurement sites (e.g., axilla, rectum, urinary bladder, pulmonary artery, nasopharynx, temporal muscle, tympanic membrane, or esophagus) are used in routine clinical practice to reflect brain T (30, 33). Because measurements from these sites are well correlated with brain T at steady state, this alternative approach works well for monitoring cerebral stability over long time periods noninvasively with respect to the ROI. However, discrepancies of this correlation are known to occur over transient changes (e.g., therapeutic induction of hypothermia) in body or brain T (43).

Relative changes in brain surface T can be measured noninvasively by infrared spectroscopy (6, 36), but information from deeper layers of the brain cannot be easily detected with this method. In contrast to thermocouples or infrared thermometry, NMR techniques can provide information on T in different regions of the brain noninvasively (i.e., within and outside the ROI). The earliest NMR methods used changes in the longitudinal relaxation time (T₁) and the apparent diffusion coefficient of tissue water to map T differences in an NMR image (11). More recently, another NMR imaging method that is based on assigning T-induced changes in the phase of the water signal has been developed (25, 35). Although these NMR approaches have relatively high spatial and temporal resolutions (due to the high water concentration in biological tissues), the methods are best suited for steady-state measurements in acute settings.

However, the sensitivities of these NMR methods are low, and calibrations are problematic (e.g., dependence on tissue perfusion and tissue types). An alternative approach is T-induced variations in NMR chemical shifts of a range of molecules. Although the chemical shift of tissue water detected by ¹H-NMR, which has a T sensitivity of 0.01 parts per million (ppm)/°C (31), has been the most popular endogenous thermometric sensor because of the abundance of the molecule in biological tissues, the accuracy of the method is affected by tissue fat content and macroscopic susceptibility differences. Thus exogenous molecules with higher T sensitivity and reliability, e.g., paramagnetic transition metal compounds, are being tested as alternative NMR thermometric sensors. However, the T sensitivity of paramagnetic transition metal compounds varies from one complex to another. For example, the praseodymium complex of 10-(2-methoxyethyl)-1,4,7,10-tetraazacyclododecane-1,4,7-triacetate (9) and the ytterbium chelate with tetra-methyl-1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (1) with ¹H-NMR have T sensitivities of 0.1 and 0.4 ppm/°C, respectively. In contrast, the complex between the thulium ion and the macrocyclic chelate 1,4,7, 10-tetraazacyclododecane-N,N',N",N'"-tetra (methylene phosphonate) (Na[TmDOTP⁵]) has a sensitivity of 1.2 ppm/°C (57) and allows the possibility of simultaneous detection of changes in pH (44) with ¹H-NMR. In contrast to thermocouples, infrared thermometry, or water-based NMR methods mentioned above, the temporal and spatial resolutions of the thermometric sensor-based NMR method are much lower due to the low concentration of the infused sensor.

Although this thulium-based NMR thermometric sensor has been successfully used to measure T changes in several organs (44, 57), it has not shown to be of much use in the brain (4, 24). Here we report the use of the thulium complex for T measurements in rat brain in vivo by ¹H-NMR. The previously reported lack of ¹H-NMR visibility of this substance in the brain (4, 24) was probably due to poor sensitivity as a result of its high rate of systemic clearance and low cell membrane permeability. By combination of in vivo and ex vivo studies, we demonstrate that the localized ¹H-NMR signals of Na[TmDOTP⁵] in the brain are primarily due to its presence in the extracellular space. We validated the NMR method by direct comparisons with thermocouple-based T measurements from brain tissue. Because the resonances of the sensor show pH sensitivity as well, we describe an approach to simultaneously measure both absolute T and pH. Other methodological concerns and future research directions with Na[TmDOTP⁵] are discussed.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

In vitro. Aqueous solutions of Na[TmDOTP⁵] (25 mM; Macrocyclics, Dallas, TX) were prepared in 95% deuterated water (Cambridge Isotope Laboratories, Andover, MA) and 5% deionized water. Different samples with varying pH (7.0-7.8) and additional cationic presence ([Ca²⁺]: 0-10 mM; [K⁺]: 0-10 mM; [Na⁺] 140-160 mM, where brackets denote concentration) were prepared in 5-mm NMR sample tubes, where 3-trimethylsilyl[2,2,3,3,-D]propionate (TSP; Sigma Chemical, St. Louis, MO), serving as an internal reference (¹H chemical shift of 0.0 ppm), was added in each sample (20 mM). The in vitro NMR data were acquired on an 11.7-T (500 MHz for ¹H) Bruker vertical-bore spectrometer (Billerica, MA) with a variable T controller in the radio-frequency (RF) probe. Nonlocalized, water-suppressed ¹H spectra were acquired with the following parameters: flip angle = 75°; recycle time (TR) = 500 ms; number of experiments = 128-1,024; dummy scans = 32. The T fluctuations in the sample throughout an entire experimental run (as measured by the T sensor in the RF probe) were negligible (±0.01°C), despite the fact that square-shaped RF pulses with short duration, high amplitude, and short TR were used. Figure 1 shows the chemical structure of the complex along with portions of its ¹H spectrum (fully relaxed). The four visible peaks emanate from four (out of six) ¹H's marked in the molecule. Note that, while typical in vitro and in vivo ¹H-NMR experiments operate around a bandwidth of ±5 ppm centered around 4.7 ppm (¹H chemical shift of water), the detection of the complex requires a much wider bandwidth (±200 ppm). Due to the high efficiency of the in vitro RF coil, both the downfield (_H2 and _H3) and the upfield (_H6 and _H1) peaks () were observed simultaneously with a wide spectral bandwidth (±200 ppm).

View larger version (21K): [in this window] [in a new window]   Fig. 1.   1H NMR spectrum of Na[TmDOTP5] in vitro under fully relaxed conditions (at 500 MHz), and a structural depiction of the molecule. The observed upfield (H6 and H1; magnitude) or downfield (H2 and H3; phased) peaks were detected with a wide spectral bandwidth [±200 parts/million (ppm)]. The other 2 1H peaks could not be detected under the present experimental settings because their chemical shifts were beyond the current spectral bandwidth. Tm3+, thulium ion.

(1A)

(1B)

(1C)

(2A)

(2B)

Animal preparation. Male Sprague-Dawley rats (150-300 g) were tracheotomized and artificially ventilated (70% N₂O-30% O₂). Intraperitoneal lines (Becton-Dickinson, Parsippany, NJ) were used for injections of -chloralose (40 mg · kg¹ · h¹; Sigma-Aldrich, St. Louis, MO) and d-tubocurarine chloride (0.05 mg · kg¹ · h¹; Sigma-Aldrich). Femoral arterial and venous lines (Becton-Dickinson) were used to measure systemic physiology (blood pressure, pH, gases) and infuse Na[TmDOTP⁵]. Both renal arteries were ligated through an abdominal midline incision to minimize the renal excretion of Na[TmDOTP⁵] (57). Two (out of 12) experiments were abandoned due to hemodynamic instability (i.e., systolic blood pressure of <80 mmHg) caused by Na[TmDOTP⁵] infusion. In the experimental rats (n = 10), ~1.5 mmol/kg of an 100 mM aqueous solution of Na[TmDOTP⁵], with a total volume of 2-4 ml, was infused at a rate adjusted to stable physiology (as measured by blood pressure, blood pH, blood gases), according to an infusion protocol reported by Winter and Bansal (49) to keep the animal within the autoregulatory range of cerebral perfusion (15).

Ex vivo (n = 8). After an infusion period of ~3 h, 5- to 10-µl samples of clear cerebral spinal fluid (CSF) were collected through a suboccipital puncture with a small 30G needle (51). Periodic blood samples (50 µl) were removed for analysis of gases and pH, and portions were saved for determining the concentration of the thermometric sensor in the blood plasma. After dilution of the CSF and blood plasma samples by 1:10 in deuterated water, ¹H spectra were acquired (see In vitro above for NMR acquisition parameters) with a known concentration of TSP. The concentration of Na[TmDOTP⁵] in CSF or blood plasma and in a fresh in vitro sample (see In vitro above for sample preparation), both with known concentrations of TSP in each sample, were measured at 500 MHz by comparison of peak integrals. The integral bandwidths of the _H2 and the _TSP peaks were chosen to be ±4.0 and ±0.4 ppm, respectively, due to the wide and narrow bases of each peak attributed mainly to their vastly differing transverse relaxation times (58). A factor K (Eq. 3) was calculated from the in vitro sample because the two peak integrals are proportional to concentrations of Na[TmDOTP⁵] and TSP

(3)

The ex vivo concentrations of Na[TmDOTP⁵] (in CSF or blood plasma) were calculated by using K determined above (Eq. 3) and by comparison of the ex vivo peak integrals (Eq. 4)

(4)

Tests of Na[TmDOTP⁵] concentration prediction for in vitro samples (see In vitro above) using Eq. 4 showed excellent accuracy (±5%) and worked equivalently well for the other peaks, although the values of K in Eqs. 3 and 4 were different and the largest uncertainty (±15%) was with the H1 peak. The value of K incorporates differences in T₁ relaxation for ¹H's in TSP and Na[TmDOTP⁵]. If the same pulse sequence parameters are used for acquisition of in vitro and ex vivo data, the combined use of Eqs. 3 and 4 for quantification of Na[TmDOTP⁵] concentration is justified because the T₁ saturation effects are normalized.

In vivo (n = 2). The rat's scalp was retracted to expose the skull, and either a 100-µm (Oxford Optronix, Oxford, UK) or a 300-µm (Physiotemp, New Jersey, NJ) thermocouple probe was placed through a craniotomy window and fixed with bone cement on the right hemisphere (1 mm anterior and 4 mm lateral from bregma; depth of 1 mm at layer 4). The in vivo NMR data were acquired on a 7.0-T (300 MHz for ¹H) Bruker horizontal-bore spectrometer (Billerica, MA) with a ¹H resonator/surface-coil RF probe (18). High-resolution anatomic images were obtained to guide the localized NMR spectroscopic measurements (see Ref. 19 for details). Outer volume suppression (37) was used to select a 4 × 4 × 4-mm³ voxel in the cortex (1 mm posterior and 0 mm lateral from bregma; depth of 2 mm from brain surface, strictly avoiding the sagittal sinus). Thus the 64-µl volume was devoid of NMR artifacts induced by the placement of the thermocouple. Localized, water-suppressed ¹H spectra were acquired with the following parameters: flip angle = 75°; TR = 500 ms; number of experiments = 128-512; dummy scans = 32. Due to the low efficiency of the in vivo RF coil (in comparison to the in vitro RF coil), either the downfield (_H2 and _H3) or the upfield (_H6 and _H1) peaks were observed (at any one experimental run) with a narrow spectral bandwidth (±25 ppm). Because square-shaped RF pulses with short duration, high amplitude, and short TR were used, we measured the possibility of RF-induced heating of the sample. The same pulse sequence parameters (on the 7.0-T system) used on an in vitro sample showed that the T fluctuations in the sample throughout an entire experimental run (as measured by a thermocouple wire attached to the sample) could be neglected (±0.03°C). An additional baseline in vivo spectrum of metabolites was collected (see Ref. 17 for details) to assign N-acetyl aspartate as the internal reference (¹H chemical shift of 2.01 ppm) for the in vivo spectra of Na[TmDOTP⁵]. Small volumes of blood samples (50 µl) were removed periodically to monitor systemic physiology. A water-heating blanket was used to alter the rat's body T. Chemical shifts of the downfield (_H2 and _H3) and/or the upfield (_H6 and _H1) peaks were analyzed and corrected for blood pH effects (wherever necessary).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

In vitro measurements. The T and pH sensitivities of the four observed peaks of Na[TmDOTP⁵] determined by linear regression analysis (see Eq. 1) are summarized in Table 1. The summary of the T and pH sensitivities of the _H2 peak is graphically presented in Fig. 2. Although the _H1 peak showed the largest changes in chemical shift with T alterations (1.17 ± 0.06 ppm/°C), its low-peak intensity (see Fig. 1) provides a poor signal-to-noise ratio (SNR) compared with the _H6 peak, which also has a high-T sensitivity (0.93 ± 0.05 ppm/°C). The _H2 and _H3 peaks showed slightly lower T sensitivities (0.55 ± 0.03 and 0.43 ± 0.02 ppm/°C, respectively). The pH sensitivities of all four peaks were approximately the same (~3 ppm/pH). There were no measurable changes in chemical shifts of any of the peaks with varying concentrations of Ca²⁺ (0-7.5 mM), Na⁺ (140-160 mM), and K⁺ (0-10 mM). Spectral data of Na[TmDOTP⁵] without and with the cations (at these concentrations) showed no statistically significant differences. In fact, the T and pH sensitivities of the four peaks (Table 1) were not statistically different from sensitivities determined in the presence of the cations (data not shown). However, some precipitation was observed with Ca²⁺ concentrations higher than 7.5 mM. Because the influence of cations on our in vitro calibration of each peak can be neglected, the effects of T and pH on the chemical shift, as shown in Fig. 2 (for the _H2 peak), can be readily applied to the in vivo data (see below), because the chemical shift dependencies of the thulium-based sensor are field independent (44, 57).

View this table: [in this window] [in a new window]   Table 1.   In vitro T and pH sensitivities of Na[TmDOTP5] (Eq. 1)

View larger version (12K): [in this window] [in a new window]   Fig. 2.   Dependence of changes in the chemical shift of the H2 peak at different brain temperature (T) and pH values. The in vitro measurements were performed using (at 500 MHz). See Tables 1 and 2 as well as RESULTS for other details.

Ex vivo and in vivo measurements. The concentration of Na[TmDOTP⁵] in paired blood plasma and CSF samples was 5.9 ± 1.4 and 0.7 ± 0.5 mM, respectively, as determined from Eqs. 3 and 4. Figure 3 shows ex vivo spectra (_H2 and _H3 "downfield" peaks) for typical blood plasma and CSF samples removed from a rat being infused with Na[TmDOTP⁵] for 181 ± 65 min (n = 10). A typical in vivo spectrum (_H2 and _H3 downfield peaks) from a 64-µl voxel in rat brain after an infusion time of slightly longer than 3 h is shown at the bottom of Fig. 3. Although similar spectra were obtained for the _H6 and _H1 "upfield" peaks (data not shown), the SNR of the _H1 peak was significantly compromised in both the in vivo and ex vivo data. The comparison of the in vitro, ex vivo, and in vivo spectra of Na[TmDOTP⁵] in Fig. 3 shows good agreement for the chemical shift difference between the _H2 and _H3 downfield peaks.

View larger version (40K): [in this window] [in a new window]   Fig. 3.   Comparison of the downfield (H2 and H3; phased) peaks under similar experimental conditions with respect to T and pH. The in vitro, ex vivo (blood), and ex vivo [cerebral spinal fluid (CSF)] spectra were obtained at 500 MHz, whereas the in vivo spectrum was acquired at 300 MHz. See RESULTS for other details.

5

5

View larger version (9K): [in this window] [in a new window]   Fig. 4.   In vivo calibration of the NMR-based T data from rat brain (300 MHz; 64-µl voxel) with direct T measured with a thermocouple wire (Oxylite or Physiotemp). A: calibration of the H2 peak by using a linear regression analysis (Eq. 1), where the brain pH was assumed to be the same as plasma pH. B: calibration of the H6 and H2 peaks used simultaneously by using the empirical models (Eq. 2), where the brain pH was not assumed for the prediction of temperature. Comparison of the predicted pH and the plasma pH are shown in the inset. See RESULTS for other details.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The observation of concomitant ¹H-NMR signals of Na[TmDOTP⁵] in blood plasma and CSF from ex vivo samples strongly suggests that the observed in vivo signals (Fig. 3) emanate from the thermometric sensor in the brain tissue. Our NMR visibility of Na[TmDOTP⁵] in the brain stands in contrast to previous results (4, 24). This discrepancy may be partly attributed to lack of NMR sensitivity at lower magnetic field strengths of the prior studies in contrast to the present studies, which were conducted at a much higher field strength. Because the prior studies (4, 24) used Ca²⁺ to counteract hemodynamic side effects during Na[TmDOTP⁵] infusion, it is possible that precipitation of the compound (see RESULTS and Ref. 2) may have confounded their experimental protocol as well as results. In this study, no extra Ca²⁺ infusion was necessary because we used a slower Na[TmDOTP⁵] infusion protocol tapered toward a constant systemic blood pressure throughout the experiment within the autoregulatory range (15). Furthermore, the rats in the previous studies (4, 24) were not nephrectomized, so that loss of Na[TmDOTP⁵] through renal clearance, the main route of its excretion from the body, may have attributed to low sensitivity by compromised presence of the sensor in the blood reaching the brain.

The other reasons for the low sensitivity for NMR visibility of Na[TmDOTP⁵] in the brain may be attributed to issues concerning the sensor's distribution volume throughout the body and the permeability across the blood-brain barrier. This sensor has been used in a variety of organs (as well as tumor models) with high NMR visibility (5, 34, 50, 53), except the normal brain tissue. Because the present infusion protocol results in a steady-state Na[TmDOTP⁵] concentration of ~6 mM in the plasma with an infusion dose of ~2 mM/kg, the apparent volume of distribution (48) throughout the entire body would be ~0.3 l/kg. Because the body's intravascular and extracellular compartments together comprise a smaller total volume, it can be suggested that the large distribution volume of Na[TmDOTP⁵] is due to the accumulation of the substance in specific organs. The present data, therefore, suggest that Na[TmDOTP⁵] has a large distribution volume throughout the rest of the body besides the brain. Furthermore, Na[TmDOTP⁵] is also used as an NMR shift reagent, which affects the chemical shift of target molecules in the extracellular space, because the reagent does not readily cross the cellular membrane. Therefore, we suggest that the low NMR visibility of Na[TmDOTP⁵] may be partly due to its high-distribution volume throughout the body and low permeability across the cell membrane.

Our results suggest that Na[TmDOTP⁵] can cross the blood-brain barrier into the interstitial space and remains at a concentration slightly higher than 10% of the concomitant blood plasma level (Fig. 3). To achieve the high blood plasma levels of Na[TmDOTP⁵], we and others (44, 49, 57) chose to perform a functional nephrectomy. Further studies are needed to evaluate the renal disposal mechanism of Na[TmDOTP⁵], because it would be advantageous to circumvent the excretion by limiting the renal clearance (e.g., pharmacologically). If specific transporters are involved in the elimination of Na[TmDOTP⁵], these potentially could be blocked by using a pharmacological approach.

Although a negatively charged molecule such as Na[TmDOTP⁵] is not an ideal compound to cross the blood-brain barrier under physiological conditions, the observation of the substance in the CSF (Fig. 3) suggests that the blood-brain barrier "is not an absolute barrier" (7), which may also explain why a range of NMR agents (e.g., gadolinium) can cross the intact blood-brain barrier into the CSF (32). When the integrity of the blood-brain barrier is compromised under pathological situations (7), then the thulium-based sensor may enter the extracellular space more readily to enhance the SNR of this NMR method for T quantification. However, it should be emphasized that the calibration of the thulium-based sensor is not concentration dependent (57, 58). If we assume that all of the NMR-observable signal of Na[TmDOTP⁵] is due to its presence in the extracellular milieu, then the concentration of the sensor after ~3 h of infusion is ~1 mM (see RESULTS).

Therefore, Na[TmDOTP⁵] can serve as an extracellular thermal probe and be employed to detect absolute T changes in the rat brain. Although the comparison of in vitro, ex vivo, and in vivo spectra of Na[TmDOTP⁵] (Fig. 3) shows very good agreement for the chemical shift differences between peaks, there is a notable difference in the peak widths observed in vivo (~3 ppm) and in vitro (~1.5 ppm). Because the static line-broadening effect induced by large magnetic field inhomogeneities in the 64-µl voxel (see ref. 21) would be negligible given the natural line broadening of the peaks due to the extremely short transverse relaxation time values (see Ref. 58), we suspect that the broadness of the in vivo peaks may be partly due to a distribution of T as well as pH values in the 64-µl voxel. Another possibility for this broadening effect of the in vivo peaks, however, may be due to contributions of Na[TmDOTP⁵] present in blood. Although the concentration of Na[TmDOTP⁵] in the intravascular space is approximately six times higher than in the extracellular space (see RESULTS), the intravascular space is ~20 times smaller than the extracellular space. Thus the largest contribution from blood would be ~20% of the in vivo observed signal. This fraction could be reduced even more if paramagnetic effects of hemoglobin in blood are included. Therefore, it is more likely that the broadening effect of the in vivo peaks is due to T and pH distributions across different brain regions.

Future NMR studies of chemical shift imaging (CSI) with multiple smaller voxels representing the entire 64-µl voxel (see Ref. 19) may help address the issue of T and pH distributions across different brain regions. Although the Na[TmDOTP⁵] concentration is lower than previously studied molecules with CSI (e.g., glutamate in Ref. 19), the extremely short T₁ values of the thulium-based sensor (58) favor significantly more signal averaging for typical sampling periods needed for CSI experiments. Thus ¹H CSI experiments of Na[TmDOTP⁵] with <4 µl voxels are very realistic, with modest SNR enhancements via RF modifications (26, 28, 52). Because the in vivo spectra of Na[TmDOTP⁵] reported here were acquired in <5 min of signal averaging with a SNR of 15-20 (for _H3 and _H2 peaks), some SNR can be compromised to improve the temporal resolution even more.

The NMR visibility of Na[TmDOTP⁵] in vivo allowed the calibration of the thermometric sensor to direct and simultaneous measurements of brain T by using thermocouple wires (Fig. 4). Because either the upfield or downfield peaks could be detected in a single in vivo experiment due to technical limitations (see MATERIALS AND METHODS), we applied the linear regression analysis results from our in vitro data (Table 1; Eq. 1) to most of our in vivo data (Fig. 4A). The use of the in vitro results toward the in vivo data was justified, given the fact that Ca²⁺, Na⁺, and K⁺ had negligible effects on the chemical shift of Na[TmDOTP⁵] as well as T sensitivity (see RESULTS). Although the T calibration accuracy with this analytical approach was acceptable, it was necessary to assume the brain pH from measured plasma pH. This assumption was removed by use of the empirical models (Table 2; Eq. 2), where both the upfield and downfield peaks were detected in sequential NMR experiments (conducted on the same rat), and the results (Fig. 4B) suggest that T (and pH) could be predicted with good accuracy. Future NMR studies with concurrent experiments (see Ref. 20) of both the upfield and downfield peaks may remove the concerns of T changes during sequential NMR experiments. Because the best _pH (±0.1) and _T (±0.3°C) are obtained from simultaneous detection of the _H6 (upfield) and _H3 (downfield) peaks with ~5 min of temporal resolution (see RESULTS; Table 2), with moderate signal averaging it would be possible to detect changes in brain pH (12, 46) and T (6, 23) during functional activation (16, 39).

Conclusion. We have shown that Na[TmDOTP⁵] can be used to measure T in rat brain in vivo. Using ex vivo and in vivo ¹H-NMR experiments, we showed that this substance crosses the blood-brain barrier into the interstitial space. Our calibration of the NMR signals from Na[TmDOTP⁵] in the extracellular space with thermocouple measurements of T suggests that this substance can be used as a noninvasive thermometric sensor in vivo with good accuracy. Future applications of this method in neurobiological models may provide insight into mechanisms of T as well as pH distributions throughout the entire brain, changes induced by alterations in neuronal activity, and regulation in response to therapy.