MRI-visible liquid crystal thermometer

Abstract

Purpose:

MRI parameters, such as T₁, T₂, and ADC, of tissue-mimicking materials in MRI phantoms can exhibit temperature dependence, and bore temperatures can vary over a 10°C range across different MRI systems. If this variation is not accurately corrected for, the quantitative nature of reference or phantom measurements is irrelevant. Available thermometers require opening the phantoms to probe the temperature, which can introduce contaminants that may affect the stability and accuracy of the phantom. An integrated, MRI-visible thermometer that can be read using typical imaging protocols is needed.

Theory and Methods:

An MRI-compatible thermometer was designed using liquid crystals (LCs) that exhibit rapid transitions between the LC cholesteric state and isotropic state in the room temperature range spanning 17°C to 23°C in 1.0°C increments. The LC thermometer was assessed visually and using superconducting quantum interference device magnetometry, NMR, and MRI techniques.

Results:

The signal generated from the LC thermometer was visible with spin-echo and gradient-echo MRI images. The LC state transition temperatures were visually referenced to a National Institute of Standards and Technology-traceable thermometer, and these LC state transitions were confirmed using superconducting quantum interference device magnetometry and NMR.

Conclusions:

The LC MR-visible thermometer had measurable changes in relative signal with temperature, which were invariant to a variety of imaging sequences used.

1 |. INTRODUCTION

Quantitative MRI allows for measurement of disease state or process using objectively measured parameters (biomarkers). For example, proton relaxation times (T₁, T₂) and the ADC have been correlated with tumor state^1,2 and used to monitor response to treatment.³ Differences in MRI hardware, software platforms, and user implementation produce variations in the measurement of these parameters, which can lead to misdiagnosis or inconclusive results. These variations can be quantified using a reference object or phantom, which contains solutions with exact and traceable values of, for example, T₁ and T₂ relaxation times and ADC. In most cases, the relevant properties of the phantom materials exhibit temperature dependence, and variations in bore temperatures can span 15°C to 25°C across different MRI systems depending on factors including scanner room temperature, time of day, and sequences used. These variations necessitate temperature-dependent corrections to phantom measurements.

There are invasive ways to measure the temperature of MRI phantoms using digital, alcohol, or fiberoptic thermometers. This requires either human interface with the phantom before and after the scan or setup and teardown of expensive fiberoptic thermometers. This method of temperature measurement increases the time spent performing the quality control procedures and relies on the temperature information to be stored separately from the image information, which could be a hurdle to information access at a later date. Invasive methods also create the potential for compromising the MRI phantom integrity by introducing unwanted microorganisms into the phantom. An integrated or in situ MR-visible thermometer that can be read using typical imaging protocols is preferred.

There has been significant effort over the last few decades to develop an integrated or in situ NMR thermometer. A well-known method to determine temperature is to image the chemical shift (the difference in proton spin resonant frequencies) observed between a hydrogen bonded to an oxygen atom (O-H) and a hydrogen atom bonded to a carbon atom (C-H) in a molecule.⁴ Typical chemicals used in such a thermometer include but are not limited to alcohols and glycols, which contain both types of hydrogen atoms. As the temperature changes, the O-H peak resonance frequency shows a pronounced shift, whereas the C-H peak resonance frequency remains constant. The separation in frequency between these peaks can be directly correlated to an absolute temperature. Whereas MRS can resolve and measure these peak separations, this protocol is not universally available and is not always simple to implement. Alternatively, one can use geometric distortion. When a tube of these chemicals is examined via MRI, the image reveals visible ghosting, with spatially separated copies of the tube formed by the protons in the two chemical environments. The spatial distance between the two tubes can be correlated to a temperature.⁵ Similar approaches used the shift of methyl proton resonances of the lanthanide chelates, such as Tm-DOTMA, which has a sensitivity of approximately 0.7 ppm/°C.⁶ Bulk magnetic susceptibility^7,8 of paramagnetic lanthanide complexes can also be used for measurement of absolute temperature.⁹ This was recently demonstrated in NMR with a coaxially arranged diamagnetic and paramagnetic solutions to have a temperature accuracy of ±0.15°C; the method has not yet been demonstrated on a clinical MRI system¹⁰ and is likely sensitive to the orientation of the tubes relative to the main magnetic field, B₀.

There are significant limitations to using the chemical shift-based methods. A chemical shift thermometer works well in an NMR environment where the main magnetic field is homogeneous, and other spatial dependencies can be ignored because the samples are small. For MRI, the sample sizes and FOV are large. Because the resonant frequency of the hydrogen atom is directly proportional to the strength of the magnetic field, variations in the main magnetic field (B₀) will introduce errors in the chemical shift-based thermometer. An even more compelling argument is that it is circular logic to use the MRI machine as an integral part of the temperature measurement during its own calibration process. If a variation in the machine exists, such as B₀ or B₁ inhomogeneity or gradient-induced spatial distortions, it could manifest itself in the temperature measurement and discredit the entire quality-control process that leads to the determination of the machine’s accuracy. Finally, chemical shift-based methods require non-trivial analysis, which may not be accessible to all users.

The desired MR thermometer would not require invasive actions by the user, would be less susceptible to machine variation, and would not require advanced processing techniques. Liquid crystals (LCs) have an optically visible transition with temperature,¹¹ appearing opaque in the crystalline phase and transparent in the isotropic phase,¹² which could make them a useful MR thermometer. Previous work incorporated LCs to measure the temperature in vivo during hyperthermia treatment of tumors for temperatures from 42°C to 45°C at 360 MHz¹³ and in a phantom of long, thin, cylindrical tubes for temperatures from 46.4°C to 50.0°C at 1.5 tesla (T).¹⁴ The demonstrated temperature range is valuable for hyperthermia studies; however, such a thermometer has not been demonstrated at MRI bore temperatures. This thermometer also required deuterated LCs to provide greater signal contrast against a water background.

We describe a LC MR-visible (LCMRV) thermometer that can be placed in an MR phantom and easily observed during routine scanning.

2 |. THEORY

LCs are highly anisotropic fluids that reversibly change between the crystalline (e.g., nematic, chiral nematic, smectic) and isotropic liquid phases. In this study, we focused on thermochromic LCs, specifically those with a cholesteric crystalline state. These molecules arrange in the crystalline state with the long molecular axes parallel to one another in-plane, with each layer incrementally displaced, resulting in helical stacking. Thermochromic LC state can be observed via a colorimetric change in the visible spectrum, shifting through various colors reflected by the crystalline mesophases until a clear liquid is observed during the isotropic phase. Thermochromic LCs with an adequate transition range are commercially available.

With typical MRI acquisitions, signal from solid materials such as crystalline plastics decays before data are acquired at the echo time. These materials appear black or without detectable signal. Our premise was that LCs, despite being liquids, would also lack a detectable signal in the crystalline state and that the transition between the cholesteric crystalline state and isotropic state could be detected using MRI as a dark and bright image, respectively.

Previous ¹H NMR studies of LCs showed distinct differences in the spectra between the various crystalline and isotropic phases.^15–17 One group detected the LC phase transition from nematic (crystalline) phase to isotropic phase. The NMR spectrum exhibited a single peak in the spectrum for the nematic phase and a multitude of peaks in the isotropic phase.¹³ Crystalline phases exhibit low-intensity broad spectral peaks, and whole isotropic phases exhibit high-intensity sharp peaks. Broad spectral peaks indicate ¹H-¹H dipolar coupling in the crystal lattice, which is responsible for the significant reduction in signal intensity. The changes in the various mesophases (e.g., smectic–nematic) of the LCs, which could provide greater temperature resolution, are not expected to be detectable by MRI.

The effect of magnetic fields on the temperature at which the cholesteric crystalline state to isotropic state transition occurs is important because these devices are to be used at varying magnetic fields (0.5 T to 7 T). Previous studies on LCs show negligible effects (less than 0.1°C) on the transition temperature for a small sampling of LC molecules.^18–20

3 |. METHODS

3.1 |. LCMRV thermometer

The LCs were placed in small polychlorotrifluoroethylene cylinders that interlock or stack to form a tower (Figure 1) and can be easily placed within an MRI reference object for imaging. Polychlorotrifluoroethylene was chosen for its low perfusion properties, chemical compatibility, and because it does not contribute to the MR signal. The volume and container shape were selected to ensure signal sufficient for visualization of a cross section of the LCs in one plane of the MRI system yet kept small to reduce any hysteretic effects due to heat-transfer equilibration times. The height of each cylinder is 7 mm, and the diameter is 8 mm for a total internal volume of 0.35 mL. Wall thickness was 0.7 mm to reduce the contribution of the plastics to the heat-transfer process.

FIGURE 1.

Experimental setup for the visual inspection (A): the LCMRV thermometer was placed within a water bath with PRT placed on either side of the LCMRV thermometer. Using the experimental setup in (A), we observe the visual transition of the target 17°C cell of the LCMRV thermometer from the cholesteric crystalline state (B), which appears iridescent, to the isotropic state (C), which appears clear or colorless. Finally, (D) shows the start of the transition of the target 18°C cell, with a slight color shift from bright white to bright green. For MRI experiments (E), the LCMRV thermometer was placed within an enclosed water bath surrounded by an outer shell containing Fluorinert FC-40 (TMC Industries, Waconia, MN). A fiberoptic thermometer was placed near the LCMRV thermometer. LC, liquid crystal; LCMRV, LC MR-visible; PRT, platinum resistance thermometers

The LCMRV thermometer estimates temperature of the phantom by signal changes in the MR image of the different cells. For example, at 19°C, the 3 LCMRV cells with transition temperatures lower than 19°C will appear bright (isotropic state), and 4 cells with higher transition temperatures will appear dark (cholesteric crystalline state). When the phantom is at 19°C, the temperature estimated by the LC thermometer will be between 18.40°C and 19.34°C.

A variety of LC combinations can be used to achieve crystalline-isotropic transitions corresponding to the temperatures typically found in the bore of an MRI scanner (16°C to 25°C).²¹ The LCs used here were cholesteric thermochromic (sterol-based) chemicals exhibiting a cholesteric crystalline state to isotropic state transition between 11°C (GB310, LCR Hallcrest, Glenview, IL) and 42.5°C (GB320, LCR Hallcrest). A series of 7 cholesteric thermochromic LC formulations was formulated by mixing weight by weight each of the LCs (Table 1) to obtain cholesteric crystalline state to isotropic state transitions between 17°C and 23°C.

TABLE 1.

Specifications of the LCMRV thermometer

Target Transition Temperature (°C)	% w/w 11.5°C LC	% w/w 42.5°C LC	Observed Start Temperature (SD)°C	Observed End Temperature (SD)°C
17	83.6	16.4	16.35 (0.067)	16.97 (0.168)
18	80.6	19.4	17.34 (0.246)	18.00 (0.080)
19	77.6	22.4	18.40 (0.195)	18.92 (0.014)
20	74.6	25.4	19.34 (0.068)	19.82 (0.044)
21	71.6	28.4	20.02 (0.038)	20.40 (0.012)
22	68.7	31.3	21.00 (0.010)	21.74 (0.451)
23	65.7	34.3	21.97 (0.310)	22.46 (0.377)

LC, liquid crystal; LCMRV, LC MR-visible; w/w, weight per weight.

Microencapsulating cholesteric thermochromic LCs protects them from oxidative attack on unsaturated bonds. However, it also broadens the temperature range of the transition from 0.1°C to 1.0°C or more. Greater resolution in the thermometer is desired for more accuracy in the measurement of the MRI parameters. We must protect the native chemicals from oxidative degradation. The oxygen in the native LC chemicals is removed by either vacuum degassing, sonication, and/or passing the chemical stream through a nonmetallic oxygen scavenger such as ascorbic acid, ascorbate salts, or catechol prior to filling and sealing a thermometer assembly.

3.2 |. Validating the LCMRV thermometer: visually, superconducting quantum interference device (SQUID), NMR, MRI

3.2.1 |. Visually

The transition temperature of each compartment in the thermometer was determined by placing the LCMRV in a water bath in Earth’s magnetic field conditions (Figure 1A). Two platinum resistance thermometers (PRT) were placed on either side of the LCMRV, and a video camera was placed above the bath to record the color changes. The water bath was set to 15°C, and every 4 min the temperature was increased by 0.5°C until the set point was 23.5°C. The video footage was reviewed by 3 viewers to record the start of the transition as the first change in hue from bright white to green and the end of the transition when the hue was clear (Figure 1B-D). The time stamp of the video at the occurrence of the transitions was matched to the PRT temperature measurements. The PRTs were calibrated by National Institute of Standards and Technology using the Digital Thermometers, Resistance Temperature Detectors, Thermistors and Other Types of Thermometers service (SKU 31110C) at 6 temperature points from 0°C to 50°C in 10°C increments. The deviation of either PRT measurement from the calibration bath temperature never exceeded 0.055°C; the reported PRT sensitivity was ±0.02°C; and the uncertainty was below 0.05°C. The error reported on these visual measurements was determined from the SD of the results from the 3 viewers.

3.2.2 |. SQUID

The diamagnetic magnetic moment versus temperature of the target 17°C and 18°C LC samples were assessed in a SQUID magnetometer. In addition, the magnetic moment of the target 17°C sample was measured at 1.5 T, 3 T, and 7 T; and the measured magnetic moment normalized by the field strength was reported. The temperatures reported for the SQUID magnetometer data were corrected based on fiberoptic probe (Opsens Medical, Québec, QC, Canada) temperature measurements because it was not possible to place a PRT into the SQUID magnetometer. The fiberoptic probe itself was calibrated in a water bath at 10°C increments from 0°C to 50°C using the two PRT. The sensitivity of the fiberoptic probe was ±0.1°C, and the accuracy was ±0.25°C.

3.2.3 |. NMR

NMR spectra were acquired at 1.5 T of the target 20°C sample at a range of temperatures (15°C-24°C) below and above its cholesteric crystalline state to isotropic state transition to understand the variation in MRI signal of the LC samples. The same fiberoptic probe for the SQUID measurements was used to record the temperature adjacent to the sample.

3.2.4 |. MRI

Several experiments were conducted in a unique test chamber in a small-bore MRI system at 3 T (Agilent, Santa Clara, CA) to examine the accuracy of the LCMRV thermometer interpretation via MR imaging. The LCMRV thermometer was placed within a temperature-controlled chamber (Figure 1E) that has an outer shell containing Fluorinert FC-40 (TMC Industries, Waconia, MN), which is connected to a programmable, temperature-controlled circulating bath. The chamber also contained a fiberoptic probe (Opsens Medical, Québec, QC, Canada) to record the temperature of the water within the chamber.

In the first experiment, the chamber was equilibrated at 23.1°C, which is above the target transition point for all LC cells. Then, the outer-shell temperature was driven to 10.0°C, which caused the water within the chamber to decrease in temperature via conductive heat transfer (loss). Over the 90-min experiment, the water bath temperature in the chamber, measured by fiberoptic probe, dropped to below 16°C. At 5-min increments over the entire 90-min experiment, gradient-echo and spin-echo images of the LCMRV thermometer were obtained (sequence parameters given in Table 2).

TABLE 2.

MRI sequence parameters

Sequence	Gradient Echo	Spin Echo	Spoiled Gradient-Echo Sequence
Flip angle	70°	90°	70°
TR	200 ms	20.10 ms	100 ms
Echo time	2.5–5 ms	13.82 ms	2.47 ms
Resolution	0.55–0.625 mm2	0.55 mm2	0.625 mm2
Acquisition time	51.2 s	41.2 s	51.9 s

In the second experiment, the temperature was initially stabilized at 15.7°C, below the target transition of all LC cells, and the outer-shell temperature was increased to 27.0°C. Over the 70-min experiment, the water temperature in the chamber, measured by fiberoptic probe, rose to 23.1°C. At 5-min increments over the entire 70-min experiment, gradient-echo and spin-echo images were obtained of the LCMRV thermometer (sequence parameters given in Table 2).

In addition, two experiments were completed at constant temperature such that some of the LC cells were in the cholesteric crystalline state and some were in the isotropic state. The first experiment was designed to look for any susceptibility or magnetization transfer-induced artifacts with changes in resolution and slice thickness. Gradient-echo (TR 200 ms, TE 6.6 ms) and spin-echo (TR 31.5 ms, TE 21.9 ms) images were collected with three different in-plane sampling rates (128 × 128, 256 × 256, and 512 × 512), two slice thicknesses (3 mm and 1.5 mm), and variable signal averaging to maintain signal to noise across all acquisitions. The second experiment examined one cell undergoing the transition between the cholesteric crystalline state and isotropic state using high-resolution images parallel and perpendicular to the orientation of the main magnetic field.

Finally, the LCMRV thermometer was placed within an MRI phantom and imaged using a 2D spoiled gradient-echo sequence at the start and end of an imaging session (sequence parameters in Table 2). For comparison, a National Institute of Standards and Technology-traceable digital thermometer (Traceable Ultra Long-Stem Thermometer, Control Company, Friendswood, TX) was used to measure the temperature before and after scanning; both times, the measurement was made outside the scanner room.

4 |. RESULTS

The LCMRV thermometer was placed in an imaging phantom, and the signal generated from the LCMRV thermometer cells in the isotropic state was clearly visible on the clinical MRI (Figure 2). The temperature of the phantom was measured using a digital thermometer prior to setup in the scanner room, 18.6°C ± 0.1°C, and at the end of the 1 h imaging session, 19.6°C ± 0.1°C. At initial scan (Figure 2A), three cells were bright (isotropic state), indicating the temperature was between 18.40°C and 19.34°C. At the end of the imaging session (Figure 2B), four cells were bright, indicating the temperature was between 19.34°C and 20.20°C. These LCMRV measurements are consistent with the digital thermometer, especially given the time for setup of the phantom within the scanner and completion of the initial 2D spoiled gradient-echo sequence acquisition, during which time the temperature in the phantom may have increased slightly.

FIGURE 2.

Demonstration of the LCMRV thermometer in an MRI phantom imaged at 3 T on a clinical system before and after routine measurements. The temperature at initial scan (A) was in the range of 18.4°C to 19.34°C (indicated by 3 bright and 4 dark cells). At the end of the 1 h session (B), the temperature increased to a range of 19.34°C to 20.2°C (indicated by 4 bright and 3 dark cells). The cell that changes from dark to bright is indicated by orange arrows in both images. T, Tesla

The measured experimental transition temperatures were determined from the visual experiment with the LCMRV thermometer in a water bath at Earth’s magnetic field conditions (Table 1). The phase change as a function of temperature was easily detected visually (Figure 1B-D) as a change from iridescent and opaque to colorless and translucent.

In the temperature-controlled MRI experiments (Figures 3 and 4), the temperature measured with the LCMRV generally agreed with the fiberoptic probe measurement. In the cooling experiment (Figure 3), the LCMRV thermometer measurement was consistent with the fiberoptic probe measurement at 12 of 14 time points. Specifically, when the fiberoptic probe measured 19.17°C to 19.25°C, the LCMRV measured temperature was between 19.34°C and 19.82°C. Again, when the fiberoptic probe measured 19.73°C to 19.85°C, the LCMRV measured temperatures between 20.02°C to 20.40°C. In the heating experiment (Figure 4), the LCMRV thermometer measurements were consistent with the fiberoptic probe measurements.

FIGURE 3.

The images and fiberoptic thermometer measurements from the ramp-down experiment (A). Temperature ranges in black were inferred from the MR images of the LC thermometer. Temperatures in gray (along the y-axis) were measured by the fiberoptic probe near the LCMRV thermometer at the start and end of the acquired image (scan time duration 51.2 s). The * indicates that the fiberoptic temperature was within a SD of the lower bound of the LCMRV temperature (22.083°C, 21.66°C, or 16.802°C). In (B), the LCMRV-measured temperature is plotted against the fiberoptic probe-measured temperature. The bar on the LCMRV temperature represents the range of values for the LCMRV thermometer (Table 1) for each MRI acquisition, and the bar on the fiberoptic temperature is the range of fiberoptic probe values from the start of the MRI acquisition until the end of the MRI acquisition. Note that the LCMRV minimum temperature is 16.35°C, and the maximum temperature is 22.46°C. Each of the minimum and maxiumum LCMRV temperatures was measured 3 times at the end and start of the experiment, respectively (indicated in blue on the plot). LC, liquid crystal

FIGURE 4.

Similar to Figure 3; however, in this case the temperature was ramped up. In (A), the * indicates that the fiberoptic temperature was within a SD of the lower or upper bound of the LCMRV temperature (21.66°C, 18.595°C, or 16.283°C). In (B), the LCMRV measured temperature is plotted against the fiberoptic probe-measured temperature. The bar on the LCMRV temperature represents the range of values for the LCMRV thermometer (Table 1) for each MRI acquisition, and the bar on the fiberoptic temperature is the range of fiberoptic probe values from the start of the MRI acquisition until the end of the MRI acquisition. Note that the LCMRV minimum temperature is 16.35°C, and maximum temperature is 22.46°C. Each of the minimum or maximum LCMRV temperatures was were measured 1 time at the start and 2 times at the end of the experiment, respectively (indicated in plot).

The LCMRV thermometer measurement was consistent across the gradient-echo and spin-echo sequences (Figure 5A-D vs. E-H) and the tested scan parameters. Short, fast scans provided adequate signal to distinguish whether LCMRV cells were bright or dark (Figure 5A,B), and the result did not change with changes in voxel size (Figure 5C,D compared to B; Figure 5G,H compared to F) or slice thickness (Figure 5A compared to 5B; Figure 5E compared to 5F). In addition, the signal variation across LCMRV cells was easily detected using a line scan averaged across 2 pixels.

FIGURE 5.

The images and line scans (averaged signal across 2 pixels) confirmed that the MR signal from the LCMRV thermometer was independent of the pixel size and slice thickness. A-D are gradient-echo images using TR 200 ms; TE 6.6 ms; slice thickness 1.5 mm (A) or 3 mm (B-D); and pixel size 1.09 × 1.09 mm² (A-B), 0.55 × 0.55 mm² (C), or 0.27 × 0.27 mm² (D). E-F are spin-echo images using similar characteristics as described for the gradient-echo case, although for the spin-echo images the number of signal averages varies to obtain consistent signal-to-noise ratio.

We applied an external magnetic field at clinical field strengths up to 7 T using SQUID magnetometry, and the phase change between isotropic state and cholesteric crystalline state was not affected by this external field (Figure 6A). Examining the normalized magnetic moment, the target 17°C sample transitioned from cholesteric crystalline state to isotropic state over the temperature range 16.3°C to 16.7°C at 1.5 T, 3 T, and 7 T (Figure 6A). Small hysteresis was observed in the diamagnetic magnetic moment curves for the 2 samples tested: 0.12°C for the target 17°C sample and 0.14°C for the target 18°C sample (Figure 6B).

FIGURE 6.

Using superconducting quantum interference device magnetometry, we observed a small change in the magnetic moment of the LC samples from the cholesteric crystalline state to isotropic state. We reported the magnetic moment normalized by the field strength of the measurement. (A) The target 17°C sample was measured at 1.5 T, 3 T, and 7 T. The observed transition temperature and sensitivity was field-independent. (B) Small hysteresis, 0.12°C and 0.14°C, respectively, was observed in the magnetic moment curves for the target 17°C sample (black) and target 18°C sample (pink).

Using NMR (Figure 7), several peaks were visible on either side of 0 Hz (location of water peak) when the spectra of the target 20°C sample was measured in its isotropic state (temperatures > 19°C). When the ambient temperature was lowered to below 19°C, the sample transitioned from the isotropic state to the cholesteric crystalline state and minimal signal was measured.

FIGURE 7.

NMR spectra of an LC sample (target 20°C sample) measured at multiple temperatures. The isotropic phase has several visible peaks (temperatures > 19°C), whereas the cholesteric crystalline state has minimal signal (temperatures ≤ 19°C).

The LCMRV thermometer remained in the bore of the MRI scanner for approximately 12 h to thermally equilibrate at 20.2°C (Figure 8). The transition range for the target 21°C cell occurs between 20.02 (0.038)°C and 20.40 (0.012)°C; this cell appeared to be equilibrated in the transition between the isotropic state and cholesteric crystalline state, including an alignment of the crystals in the cholesteric crystalline state (Figure 8B). The alignment of the crystals was congruent with the B₀ field of the MRI magnet.

FIGURE 8.

We used MR images to examine the transition between the cholesteric crystalline state and isotropic state by arranging the surrounding bath temperature at a constant 20.2°C, which is in the middle of the measured transition temperature range (20.03°−20.40°C) for the target 21°C sample. Interestingly, there appears to be an alignment of the cholesteric state crystals with the main magnetic field, B₀.

5 |. DISCUSSION

The purpose of this effort was to develop a method of determining temperature within an MRI phantom without the use of external thermometers, special imaging sequences, or techniques sensitive to MRI machine performance. The LCMRV thermometer developed in this study provides an integrated or in situ measurement of temperature within an MRI phantom in less than 1 min using the signal difference produced by the LC transition between a cholesteric crystalline state and isotropic state. This LCMRV thermometer can be independently calibrated outside the MRI against a National Institute of Standards and Technology-traceable thermometer and is robust to variations in MR acquisition protocol and magnetic field variations. Based on the experiments here, the LCMRV is accurate to ±0.5°C.

Previous NMR literature on alternative LCs shows a change in the spectra between the liquid state and crystalline state.¹⁶ The LCs used in the LCMRV thermometer exhibit similar transitions, supporting the ability to differentiate the states by clinical MRI. Three different sequences on two scanners, one clinical and one preclinical, were easily able to detect the bright and dark cells and the resulting LCMRV-measured temperature. Additionally, we evaluated the effect of voxel size and slice thickness to test for any susceptibility or magnetization transfer-induced changes in the MR image. Based on those results, the method described in this paper is independent of several sequence parameters, and the fastest-possible sequence can be implemented. In particular, the LCMRV-measured temperature can be obtained during a standard initial scout sequence.

In the cooling experiment in the MRI system, there were two occasions when there was a discrepancy between the observed temperature in the LCMRV and the measured temperature using a fiberoptic probe. In both cases, the discrepancy between the two measurement techniques was less than 0.5°C. In each of the cooling and heating experiments, there were three situations in which the fiberoptic probe measurement and the LCMRV measured-temperature were consistent only when accounting for the SD of the LCMRV measurement. The SDs all were below 0.5°C. These minor discrepancies could be due to the physical distance between the location of the LCMRV and the location of the fiberoptic probe.

It is important to assess whether the B₀ field affects the cholesteric crystalline state to isotropic state characteristic transition temperature. Our experiments show no change in characteristic transition temperature between zero applied field and the magnetic field strengths used in clinical applications. Additionally, the SQUID hysteresis curve width is approximately the same as the 0.1°C target transition range (although SQUID hysteresis curves are dependent on the speed of cycling through the transition).

For this study, determination of the transition temperatures was performed by human eye, which is a limitation of the results. A more accurate assessment of the transition temperatures would implement a photodetector to measure the cholesteric crystalline state to isotropic state transition, and a faster way to determine transition temperature would be to employ computer vision algorithms to examine the image and determine the point between bright and dark cells as referenced to background signal. The measured temperature would be determined by the location of the bright–dark transition between the LCMRV cells. The same algorithm could be applied to the MRI data to measure temperature. Alternatively, a line scan on the MR image through the center of the LCMRV thermometer can be plotted as a function of position. The temperature will be indicated by the point at which the function goes from high to low signal. One limitation of the design of this thermometer is that it requires individual compartments for each temperature change. As a result, if 0.25°C resolution is desired over the range 17°C to 23°C, 28 compartments are then required.

The temperature range of the observed phase transitions in these experiments were approximately 0.3°C to 0.7°C, greater than the theoretical limit of 0.1°C. We note that the accuracy of the LCMRV, ±0.5°C, is smaller than the reported errors of the proton resonance frequency shift method, ±1°C to 3°C.²² The broader-than-expected transition range could be due to imperfect preparation of the LC solutions. For example, mixing must be done in the isotropic state, which was achieved by heating the solutions to a temperature above 42.5°C. When the mixture was transferred to the thermometer cell containers, which were at room temperature, the rapid cooling by contact with the containers may have caused some phase change and potential demixing of the solution. Ideally, the LC solutions would be mixed with equipment that maintains the temperature of all cell containers above the isotropic transition temperature. Also, the native LCs are sensitive to oxidative degradation. These solutions were prepared in air, and each cell contained an air bubble after filling. Ideally, the thermometers would be assembled in an inert environment to prevent oxidative degradation. These imperfections may account for the broader-than-expected transition temperature range and also the discrepancies between the targeted and measured cholesteric crystalline state to isotropic state transition temperatures. Another reason for the broader transition range could be the characteristics of heat transfer through the LCMRV thermometer. The measured transition temperature range is a convolution of the characteristic transition range of the LCs, heat-transfer characteristics for the assembly (wall thickness, shape, volume, material), and equilibration time. Each thermometer should be calibrated in its final design to report both transition temperature and transition range.

6 |. CONCLUSION

The LCMRV thermometer can be used to rapidly measure temperature and allows for this value to be encoded with the images. The integrated LCMRV thermometer reduces the potential for introducing organisms that could grow in the phantom environment and alter the target MRI parameters. The output of the LCMRV can be used to correct the MRI parameters of interest for temperature and to better establish the uncertainty of MRI-based measurements in the clinic.