An Acetone Based Phantom for Quantitative Diffusion Magnetic Resonance Imaging

Abstract

Purpose

The purpose of this study was to propose and evaluate an acetone-D₂O phantom which has extended range of ADC for quantitative diffusion MRI, as well as to compare its properties to previously described water-based phantoms.

Materials and Methods

The proposed acetone-D₂O, and previously described sucrose water solution and PVP water solution phantoms were constructed in a number of concentrations between 0% and 50%. At 1.5T field strength, diffusion-weighted MR spectroscopy (DW-MRS) based on a point resolved spectroscopy (PRESS) acquisition, non-diffusion-weighted stimulated echo acquisition mode (STEAM)-MRS and diffusion-weighted echo-planar imaging (DW-EPI) were used to evaluate each phantom. The MR spectra, diffusion-weighted signal decay pattern, tunability of ADC, and ADC range of each phantom were all evaluated.

Results

When placed in an ice-water bath, all phantoms provided desirable signal properties, including single-peak signal with Gaussian diffusion and tunable ADC. At 0°C, however, water-based phantoms had ADC limited to less than 1.1·10⁻³ mm²·s⁻¹ (0.2–1.1·10⁻³ mm²·s⁻¹) while the proposed acetone-based phantom had ADC values spanning a wider range (0.6–3.5 ·10⁻³ mm²·s⁻¹).

Conclusion

The proposed acetone-D₂O phantom provided desirable signal properties over a wide range of ADC with temperature controlled using an ice-water bath.

INTRODUCTION

Quantitative diffusion MRI has been the subject of intensive research developments that seek to improve data acquisition as well as diffusion signal modeling and reconstruction(1,2). There is an emerging body of evidence demonstrating the potential of quantitative diffusion MRI techniques for diagnosis, staging and treatment monitoring of cancer in multiple organs(3–7). Unfortunately, widespread dissemination and application of quantitative diffusion MRI has been limited. Large variations in measured diffusion parameters have been observed between studies and research sites in both pathological and normal tissues(8). The potential sources of these variations include physiological variations such as motion and the presence of fat(9). Other sources of variability include hardware imperfections such as image distortions caused by susceptibility(10) and eddy currents(11), as well as b-value error(12) from imperfect gradient amplitude calibration(13). In order to characterize the effect of these technical confounding factors, it is highly desirable to develop a phantom that provides accurately known reference apparent diffusion coefficient (ADC) values that are unconfounded by the presence of physiological motion, the presence of fat, or diffusion modeling mismatches.

Early phantoms(14) for diffusion MRI were constructed using different pure substances, including water, acetone and various oils(15–18). These phantoms are easy to construct and reproducible, but provide a very limited number of ADC values. Alternatively, solution phantoms have been proposed. In these phantoms, water serves as the solvent and provides the MRI signal. Its diffusion behavior is tuned by dissolving a solute that reduces the ADC of water in a concentration dependent manner. Two important examples of solution phantoms include designs based on sucrose(19,20) and polyvinylpyrrolidone (PVP) (21,22), each dissolved in water. A major challenge with these phantoms is that solutes such as sucrose and PVP generate MR signal with multiple spectral peaks(23,24). Although preliminary studies have examined the properties of these phantoms(19–22,25), comprehensive validation is still required to rule out possible confounding effects from the solute signal on the measured ADC.

An essential component of any diffusion phantom is temperature control because the diffusion of liquids such as water is highly dependent on temperature(26). For this reason, the use of ice-water baths has become a well-accepted means to maintain the temperature of a diffusion phantom(12) at a highly reproducible temperature (0°C). Unfortunately, water has limited ADC (ADC<1.1 ·10⁻³ mm²·s⁻¹)(26) at 0°C, while ADC in tissue may be up to 2.6 ·10⁻³ mm²·s⁻¹ at body temperature(27). Although scanning water-based phantoms at higher temperatures (e.g., 37.5°C) is possible(28), accurate control at such temperature is very challenging compared with temperature control at 0°C.

In this work, we propose a diffusion phantom design using acetone as the signal source. Compared to water, acetone has very high ADC (e.g., ADC>3.0 ·10⁻³ mm²·s⁻¹ at 0°C)(29). Further, it has been shown that in a mixture of water and acetone, the diffusion coefficient of acetone can be lowered due to hydrogen bonding between water and acetone(30). Unfortunately, acetone-water mixtures produce MR signals from both acetone and water (i.e., two spectral peaks with different diffusion properties), which will confound quantitative diffusion MRI measures. To avoid water signals, we propose to substitute water with deuterium oxide (D₂O) as the solute. In the proposed phantom, D₂O alters the ADC of acetone the same way as H₂O without producing any MR-visible signal at proton resonance frequencies. Because of high ADC of acetone at 0°C, the proposed phantom may reach the entire physiological ADC range (0.6 ·10⁻³ mm²·s⁻¹ –2.6 ·10⁻³ mm²·s⁻¹) under ice-water bath temperature control.

Therefore, the purpose of this study was to propose and evaluate an acetone-D₂O phantom for quantitative diffusion MRI, as well as to compare its properties to previously described water-based phantoms.

MATERIALS AND METHODS

Phantom Construction

Previously proposed water-based solution phantoms, as well as the acetone-based phantom design proposed in this work were constructed without any doping agents, as follows:

Sucrose phantom: a sucrose phantom was constructed using an agar gel matrix following the recipe described by Lavdas et al(19). In the five vials constructed, the concentrations of sucrose (Sigma-Aldrich, St. Louis, MO) were 0%, 10%, 20%, 30%, 40% weight/volume (w/v), dissolved in deionized water.

PVP phantom: six vials were built with PVP (Sigma-Aldrich, St. Louis, MO) at concentrations of 0%, 10%, 20%, 30%, 40%, 50% w/v, dissolved in deionized water. Similar to the work by Pierpaoli et al.(21), other components included sodium chloride (9g/L) which modifies the phantom’s dielectric properties, and sodium benzoate as a preservative (3mM). These ingredients were added to conform the phantom to the original recipe.

Acetone-D₂O phantom: the proposed acetone-D₂O phantom was built by mixing pure acetone and D₂O (Sigma-Aldrich, St. Louis, MO) with the following concentrations of D₂O: 0%, 5%, 10%, 20%, 40% v/v, mixed in acetone. Additionally, an acetone-H₂O phantom was built with the same concentrations of H₂O for comparison.

All phantoms were stored in glass vials (Sigma-Aldrich, St. Louis, MO) 9.5 cm in height and 2.75 cm in diameter.

Study of Phantom Properties of Interest

Imaging and spectroscopic data were acquired to examine the following properties of each diffusion phantom:

1. Single-peak MR spectrum. Phantoms with multiple MR spectral peaks will result in severe chemical shift artifacts(31) in single shot echo-planar imaging (EPI)-based diffusion MRI, therefore a single-peak MR spectrum is highly desirable. Single-voxel multi-echo stimulated echo acquisition mode (STEAM)-MRS and diffusion-weighted (DW)-MRS(32) were performed to study the MR spectrum of phantoms as well as the diffusion of each chemical species.

2. Reproducible diffusion behavior. Due to the sensitivity of diffusion to changes in temperature, temperature control is required for reproducible diffusion behavior. An ice-water bath was used in order to attain reproducible diffusion behavior in DW-MRS and DW-EPI experiments. Temperature was monitored using a fiber optic thermometer as described below.

3. Isotropic Gaussian diffusion. Validation in the setting of Gaussian diffusion is a first and necessary step in the validation of all diffusion MRI techniques. Isotropic Gaussian diffusion is generally expected when diffusion is unhindered by any spatial restrictions (e.g., cell boundaries). However, in solutions where hydrogen bonding exists, a distribution of diffusion rates may arise (i.e., spins involved in hydrogen bonds may diffuse at a different rate compared to spins that are not involved in hydrogen bonds)(33). Given the presence of hydrogen bonding in the solution phantoms analyzed in this study, it is unknown whether this effect could lead to non-Gaussian diffusion behavior. To examine whether the assumption of Gaussian diffusion holds, DW-EPI was performed. The presence of monoexponential signal decay with increasing b-values was tested as a surrogate of Gaussian diffusion behavior (When spins undergo Gaussian diffusion, signal acquired with increasing diffusion encoding, i.e., b-values, will experience a true monoexponential decay) as follows. In this study, to validate monoexponential diffusion decay, one ADC of each phantom was measured from DW-EPI images with two small b-values, while another from two large b-values respectively. Should two ADC measurements agree, this would support the presence of Gaussian diffusion. Additionally, DW-EPI data were acquired with multiple diffusion gradient durations. The agreement between ADC calculated using these different diffusion gradient durations was evaluated for additional validation of Gaussian diffusion. Although we expected the phantom to exhibit isotropic diffusion, all acquisitions were repeated with X, Y, Z diffusion directions, to test the reproducibility of ADC measurements with respect to diffusion direction.

4. Tunable ADC values. ADC measured from DW-EPI data was also used to test the feasibility of tuning ADC values and investigate the range of ADC tuning capacity of the solute. To determine the feasibility of ADC tuning by changing concentration of the solute (sucrose, PVP and D₂O, respectively) in the phantoms, ADC measured in multiple vials with increasing solute concentrations (as described above) were compared for each type of phantom.

5. Wide range of ADC. The attainable ADC values should cover the entire clinically relevant range (0.6 ·10⁻³ - 2.6 ·10⁻³ mm²·s⁻¹) (27). The range of ADC values measured from DW-MRS and DW-EPI data of each phantom was evaluated.

The temperature control setup (ice-water bath), acquisition parameters and data processing for imaging and spectroscopic experiments are described in detail in the ensuing paragraphs.

Ice-water Bath

MR experiments were conducted in an ice-water bath for all phantoms. Additionally, the same experiments were repeated at room temperature for the sucrose and PVP phantoms in order to extend their ADC range.

The ice-water bath was conducted in a plastic container of dimensions 20cm×14cm×10cm. A layer consisting of 125ml of ice was formed in the bottom of the container, with approximately 250ml of cold water mixed with the ice, in order to immerse the vials completely. Vials with high concentrations of MnCl2 (approximately 10 mM), and therefore with no visible MR signal, were used to hold the vials of interest in place. Two separate fiber optic thermometer probes were securely placed within the ice-water bath (between the phantom vials), and the average of their measurements was used to monitor temperature changes during the scan.

Signal arising from the ice-water bath was eliminated by adding manganese chloride (2mM) to shorten the T2 to less than 5ms(34). This was necessary because a significant chemical shift can cause overlap between acetone and surrounding water in diffusion weighted-echo planar imaging (DW-EPI).

Data Acquisition

MRS and MRI data were acquired to examine the properties of each diffusion phantom. All experiments were conducted in a clinical 1.5T MRI system (HDxt, GE Healthcare, Waukesha, WI) magnet using a standard 8-channel cardiac phased array coil.

1. DW-MRS: DW-MRS based on a point resolved spectroscopy (PRESS) acquisition(32) was performed in each vial within each phantom, using b-values 0, 100, 250, 500, 750, 1000, 1500 s/mm², including flow compensation for diffusion encoding gradients. Other parameters include voxel size= 13mm×13mm×28mm, TE=146ms, TR= 2000ms, NEX=8, diffusion gradients applied in the S/I direction.

2. Multi-TE STEAM: To sample the MR spectrum without heavy T2 or diffusion weighting, STEAM-MRS(35) was acquired with multiple short echo times. STEAM-MRS parameters included multiple TEs=8.6, 13.6, 18.6, 23.6, 28.6ms, voxel size=13mm× 13mm× 27.8mm, TR= 2000ms, mixing time (TM)=5ms. Multi-TE STEAM was acquired once in each vial within each phantom.

3. DW-EPI: DW-EPI was performed on all the phantoms using a dual spin-echo single-shot EPI sequence. Acquisition parameters included TR=6000ms, TE=100ms, FOV=34cm×17cm, matrix size=128×64, slice thickness=6 mm, number of slices=4, slice in axial plane, no parallel imaging acceleration. b=0, 100, 300, 500, 750, 1000, 1250s/mm² (the same as those used for DW-MRS). The diffusion gradient duration was 25.3ms and diffusion time was 31.3ms. For the purpose of validating reproducibility of ADC against the changes in diffusion gradient duration, DW-EPI acquisitions with b = 0, 100, 300, 500 were performed with diffusion gradient durations of 18.6ms, 25.3ms and 31.1ms and diffusion times of 24.6ms, 31.3ms and 37.1ms, respectively. Separate acquisitions were also performed using the same b-value combinations, but with diffusion gradients applied in the X, Y, and Z directions, respectively.

Data Processing and Analysis

1. DW-MRS: At each b-value, solute and solvent signal in each phantom were estimated individually when both signals could be detected with sufficient amplitude (i.e., when the amplitude of the smaller signal peak was no less than 10% of the larger peak). This was performed for the purpose of measuring individual ADC of each chemical species. Signal estimation was performed by fitting the spectrum to a linear combination of Lorentzian spectral shapes(36). Individual ADC values were measured for the solvent and the solute signal separately by fitting their signal decay over b-values to a monoexponential curve. When the solute signal was too weak or absent, only the solvent ADC was measured.

2. Multi-TE STEAM: Single-voxel spectroscopy data using a multi-TE STEAM pulse sequence were displayed to visualize the presence of MR spectral peaks from all chemical species in each phantom.

3. DW-EPI:

   1. A circular region-of-interest (ROI) of size 0.85cm² located on the central slice was used to measure the average signal acquired at each b-value for sucrose (40%), PVP (50%), acetone-D₂O (40%) phantoms. Two separate ADC values were measured by fitting the ROI signals from two small b-values (b=0, 500 s/mm²) and two large b-values (b=750, 1250 s/mm²), respectively, to a monoexponential decay signal model. The two ADC measurements were compared. Using two b-values is not the optimal way to estimate ADC accurately, but here two combinations of b-values were used to confirm monoexponential signal decay (i.e., Gaussian diffusion).
   2. In order to optimally estimate ADC from each vial, using each diffusion acquisition protocol (diffusion direction, diffusion gradient duration and diffusion time), DW-EPI based ADC measurements were also performed from all b-values (using monoexponential least-squares fitting) on a voxel by voxel basis. A single ADC value was calculated for each vial in each acquisition protocol by averaging ADC values inside a ROI of size 0.85cm² on the central slice.
   3. The feasibility of ADC tuning and the range of ADC were determined using ADC values obtained from DW-EPI data acquired with all b-values and diffusion gradient duration of 25.3ms, diffusion time of 31.3ms.
   4. The DW-EPI based ADC measurements in sucrose (40%), PVP(50%) and acetone-D₂O(40%) phantoms were compared across different diffusion gradient durations as well as diffusion direction for additional validation for Gaussian diffusion.

Relaxometry of the acetone-D₂O phantom

In order to assess the relaxation properties of the proposed acetone-D₂O phantom, T1 and T2 were measured for each vial. To measure the T1 of the acetone-D₂O phantom in ice-water bath, 2D fast spin echo-inversion recovery (FSE-IR) was performed in the axial plane. Acquisition parameters included inversion times (TI) of 400, 800, 1200, 1600, 2200, 3000ms, TE=400ms (a long TE was used in order to avoid signals from the surrounding doped ice-water bath), FOV=24cm×24cm, slice thickness=10mm, TR=10,000ms. A T1 map was calculated by fitting inversion recovery signal model to the signals (37) on a pixel-by-pixel basis. For each vial, a single T1 was measured by the by averaging T1 measurements in a circular ROI of 73.8cm² chosen in the center of each vial.

To measure T2 of the acetone signal in the acetone-D₂O phantom at 0°C, 2D spin echo (SE) was performed in the axial plane. Acquisition parameters included TE=14, 500, 1250, 2000ms, FOV=24mm × 24mm, slice thickness=10mm, TR=7000ms. For each voxel, a monoexponential signal model was fit on a pixel-by-pixel basis to estimate T2 maps. For each vial, individual T2 estimates were averaged over a circular ROI of 73.8cm².

The effect of manganese chloride on temperature of ice-water bath

Adding a doping agent (MnCl₂) to the ice-water bath may lead to undesired deviation of temperature from 0°C. To study this potential effect, a container of ice-water was doped with several MnCl₂ concentrations (0mM, 0.5mM, 1mM, 1.5mM, 2mM). For each concentration, 80 ml of ice-water as well as the corresponding weight of MnCl₂ were mixed in a beaker roughly 7cm in height and 5cm in diameter. The temperature of the resulting ice-water bath was measured using a fiber optical thermometer at the ice-water interface, and this measurement was repeated 10 times.

Statistical Analysis

Linear regression of the measured temperature and known concentration of MnCl₂ was performed to characterize ice-water temperature with respect to MnCl₂ concentrations used in this experiment.

RESULTS

Study of Phantom Properties of Interest

1) Single-peak MR spectrum

Representative DW-MRS and STEAM-MRS spectra of the sucrose phantom (40% sucrose) at 0°C and at room temperature are shown in Figure 1. The STEAM-MRS spectra, acquired with short echo times (TE between 8.6ms–28.6ms), show both water and sucrose signal at both temperatures. At 0°C, sucrose signal was not observed in the DW-MRS spectrum, which was acquired at long echo time (TE=146ms). At room temperature, two sucrose peaks were observed in DW-MRS, one on each side of the main water peak. These peaks demonstrated slower decay than the water peak with increasing b-values, due to the slower diffusion of sucrose compared with water, i.e., the diffusion signal from this phantom has multiple components, each with different ADC values.

Figure 1.

Using DW-MRS, sucrose signal was observed in a sucrose phantom at room temperature but not at 0°C. Shown are DW spectra (TE=146ms) as well as short-TE non-DW STEAM spectra of a sucrose phantom (40% sucrose in water solution), both in an ice-water bath (0°C) and at room temperature. At room temperature, sucrose signal was found in STEAM-MRS and DW-MRS. In DW-MRS the high signal is likely due to long sucrose T2. This high sucrose signal complicates the use of room temperature sucrose phantoms for quantitative diffusion MRI. However, no apparent sucrose signal was observed at 0°C in DW-MRS despite the sucrose peak shown in STEAM-MRS, hence this phantom may be considered single-peak in ice-water bath when a long echo time is utilized.

Representative DW-MRS and STEAM-MRS spectra of the PVP phantom (50% PVP) at 0°C and at room temperature are shown in Figure 2. Using STEAM-MRS, at room temperature two PVP peaks were observed between 2 and 4 ppm(24). for the TE=8.6ms acquisition. These peaks decay very quickly with increasing TE (i.e., PVP has short T2), and demonstrate near complete decay at TE=28.6ms. A single signal peak was observed in the spectra of 50% PVP phantom using DW-MRS at both temperatures.

Figure 2.

PVP phantom shows single peak spectrum in ice-water bath and at room temperature. The plots show DW-MRS (TE=146ms) and STEAM-MRS with no diffusion weighting acquired in the PVP phantom (50% PVP) in an ice-water bath and at room temperature. Nearly single peak spectra were observed in both DW-MRS and STEAM-MRS at both temperatures. After zooming in on STEAM-MRS at room temperature, a fast decaying PVP signal was found at 2–3ppm. This suggests the single-peak spectrum results from the low intensity and rapid decay of the PVP signal at either ice-water or room temperature.

In the acetone-H₂O (20% H₂O) phantom, both acetone and H₂O generate a single MR spectral peak separated by approximately 2ppm (Figure 3). However, in the acetone-D₂O (20% D₂O) phantom, only a single spectral peak was observed, due to lack of MR signal from deuterium. The acetone signal decayed at a similar rate (ADC of acetone was measured as 1.21 ·10⁻³ mm²·s⁻¹ in acetone-D₂O, 1.43 ·10⁻³ mm²·s⁻¹ in acetone-H₂O using DW-MRS), demonstrating that D₂O and H₂O have similar impact on the diffusion of acetone molecules. Note in Figure 3 that the H₂O signal in acetone-H₂O creates a ghost image on DW-EPI images, which overlaps with the acetone signal.

Figure 3.

Acetone signal showed similar diffusion decay in acetone-D₂O and acetone-H₂O phantom of the same concentration (20% H₂O or D₂O, respectively). The plots show DW-MRS (TE=146ms) and STEAM-MRS of acetone-D₂O and acetone-H₂O phantoms in an ice-water bath. Importantly, H₂O gives rise to a large peak, whereas D₂O produces no NMR signal. Acetone-D₂O and acetone-H₂O phantoms provide similar acetone diffusion signal behavior. However, H₂O produces signal which appears as a ghost in the DW-EPI images, whereas D₂O produces no MR signals.

2) Reproducible diffusion behavior

Temperature was measured between 0.4 °C and 1.75 °C during the scans in ice-water bath.

3) Isotropic Gaussian diffusion

The logarithms of signal decay curves measured by DW-EPI of sucrose (40%), PVP (50%), acetone-D₂O (40%) phantoms in ice-water bath and sucrose, PVP phantoms at room temperature are shown in Figure 4. Deviations from a straight line indicate non-Gaussian diffusion behavior. Among the sucrose, PVP and acetone-D₂O phantoms, only the sucrose phantom at room temperature demonstrated substantial deviation (r² = 0.986) from monoexponential decay (r² > 0.997 in other cases). The presence of sucrose signal at room temperature results in clear non-monoexponential signal decay. Further the ADC values measured for each phantom using b=0, 500s/mm² and b=750, 1250s/mm² respectively in DW-EPI are listed in Table 1. Sucrose, PVP, and acetone-D₂O, with the exception of the sucrose phantom at room temperature (ΔADC = 0.28·10⁻³mm²·s⁻¹), showed differences smaller than 0.07·10⁻³mm²·s⁻¹ between the ADC values measured using the two sets of b-values (i.e., demonstrating monoexponential diffusion signal decay).

Figure 4.

PVP phantom and acetone-D₂O phantom showed monoexponential diffusion signal decay. Color coded lines show the logarithm of relative signal intensity at each b-value for PVP phantom (50%), sucrose phantom (40%), acetone-D₂O phantom (40%). Sucrose phantom’s diffusion decay pattern deviates from a monoexponential model, especially at room temperature (see arrow). Signals were averaged in an ROI (0.85cm²) inside each vial on DW-EPI images.

Table 1.

ADC measured in PVP and acetone-D₂O phantoms, were robust to estimation using different groups of b-values. Specifically, ADC was estimated using a subset of small b-values (0,500· mm⁻²·s¹) and a subset of large b-values (750,1250· mm⁻²·s¹). The difference between these two ADC values in the sucrose phantom indicates signals with multi-exponential decay from multiple signal sources.

Phantom	ADCa/(10−3mm2·s−1) Measured using b=0,500· mm−2·s1	ADC/(10−3mm2·s−1) Measured using b=750,1250· mm−2·s1	(ADCb=0,500- ADCb=750,1250)/ ADCb=750,1250×100%
PVPb (Room Temperature) 50% PVP	0.43	0.47	−8.5
PVP (0 °C) 50% PVP	0.20	0.23	−10.7
Sucrose (Room Temperature) 40% Sucrose	0.68	0.4	42
Sucrose (0 °C) 40% Sucrose	0.42	0.36	16.5
Acetone-D2O 40%D2O (0 °C)	0.54	0.56	−4.5

^aApparent diffusion coefficient

^bPolyvinylpyrrolidone

ADC values measured using different diffusion gradient durations (Table 2) were within a 0.12·10⁻³mm²·s⁻¹ range from each other for all phantoms, and within 0.03 ·10⁻³mm²·s⁻¹ range for acetone-D₂O and PVP phantoms. The fact that ADC measurements were independent of diffusion gradient duration (and therefore diffusion time) is consistent with Gaussian diffusion behavior.

Table 2.

ADC measured from acetone-D₂O phantom, with different diffusion gradient durations (different diffusion times). No monotonic changes in ADC were observed in sucrose phantom with increasing diffusion time. Closely agreeing ADC (ΔADC≤0.03·10⁻³ mm²·s⁻¹) was measured using different diffusion gradient duration and diffusion time for PVP and acetone-D₂O phantoms.

Phantom	solute concentration %	ADCa (diffusion gradient duration =18.6ms)	ADC(diffusion gradient duration =25.3ms)	ADC(diffusion gradient duration =31.1ms)
Sucrose (room temperature)	40%	0.61	0.65	0.53
Sucrose(0 °C)	40%	0.37	0.44	0.35
PVPb (room temperature)	50%	0.46	0.45	0.43
PVP (0 °C)	50%	0.23	0.23	0.23
Acetone-D2O	40%	0.55	0.55	0.56

^aApparent diffusion coefficient

^bPolyvinylpyrrolidone

DW-EPI based ADC measured with all diffusion direction was compared for 40% sucrose phantom, 50% PVP phantom and 40% acetone-D₂O phantom. The measurements were within a 0.10·10⁻³mm²·s⁻¹ range from each other for sucrose phantom, and within a 0.04·10⁻³ mm²·s⁻¹ range for PVP and acetone-D₂O phantoms.

4) Tunable ADC values and 5) Wide range of ADC

The ADC measurements from DW-MRS and DW-EPI utilizing all b-values from all phantoms are summarized in Figure 5. In all phantoms, the ADC of the solvent is tunable by modifying the solute concentration. In the water-based phantoms (sucrose and PVP), the range of achievable ADC is limited to less than approximately 1.1·10⁻³ mm²·s⁻¹ at ice-water temperature. In contrast, substantially higher ADC (3.4·10⁻³ mm²·s⁻¹) is achievable in the proposed acetone-D₂O phantom.

Figure 5.

The proposed acetone-D₂O phantom covers the entire physiological ADC range at ice-water temperature. In all phantoms, the ADC of the solvent is modulated by the solute concentration. ADC measurements from DW-EPI and DW-MRS are shown for PVP and sucrose phantoms at both room temperature and 0°C, and for acetone-D₂O and acetone-H₂O phantoms at 0°C. Two ADC values were measured by DW-MRS for the solvent and solute signals when the solute signal intensity was high enough. Sucrose and PVP phantoms were limited to low ADC values, particularly when scanned at 0°C, whereas the proposed acetone-D₂O phantom attained a wide range of ADC at 0°C, covering the entire physiological ADC range. In sucrose phantoms at room temperature, although solvent ADC was modulated by solute, the ADC measured by DW-EPI is confounded by the presence of solute signal.

DW-MRS measures two very different ADC for sucrose (0.13·10⁻³ mm²·s⁻¹ with 40% sucrose) and water (0.72·10⁻³ mm²·s⁻¹ with 40% sucrose) in the sucrose phantom at room temperature. The concentration of sucrose as a solute successfully modulated the ADC of H₂O signal measured by DW-MRS at this temperature. However, the ADC measured using DW-EPI (0.40·10⁻³ mm²·s⁻¹ with 40% sucrose) falls between that of sucrose and water measured using DW-MRS, demonstrating explicitly the confounding effect of the solute signal from sucrose, which leads to multi-exponential diffusion decay behavior in DW-EPI.

In the cases without detectable solute signal, ADC measured by DW-MRS and DW-EPI had differences smaller than 0.13·10⁻³ mm²·s⁻¹.

T1 and T2 of Acetone-D₂O Phantom

The T1 of acetone was measured using FSE-IR images as 2.41s, 2.59s, 2.60s, 2.65s, 2.61s for ace-tone-D₂O phantom with D₂O concentrations of 0%, 5%, 10%, 20%, 40%, respectively. In the same phantom, the T2 of acetone was measured using 2D SE images as 2.30s, 2.42s, 2.42s, 2.53s, 2.39s for D₂O concentrations of 0%, 5%, 10%, 20%, 40%, respectively.

The Effect of Manganese Chloride on Temperature of Ice-water Bath

The temperature measurements obtained in pure ice-water, as well as in doped ice-water with different MnCl₂ concentrations are shown in Figure 6. The linear regression slope between MnCl₂ concentration and temperature was −0.036 with confidence interval (95%) of [−0.08, 0.00].

Figure 6.

No significant linear relationship was observed between MnCl₂ concentration and temperature at ice-water interface(P=0.08). Temperature measured at ice-water interface in ice-water doped with MnCl₂ with various concentrations. Linear regression was performed with regressand being temperature and regressor being the concentration of MnCl₂. A t-test was used to determine whether a linear dependence between temperature and the concentration of MnCl₂.

DISCUSSION

In this study, we have proposed and evaluated a new phantom based on a solution of D₂O dissolved in acetone, and compared its characteristics to previously described phantoms based on water-based solutions of PVP and sucrose, respectively. The proposed acetone-D₂O diffusion phantom overcomes the limited ADC range of water-based phantoms at 0°C. A physiological range of ADC was achieved using acetone as a signal source (solvent) and D₂O as an MR invisible solute that can be used to modulate the ADC of acetone. In addition to the expanded ADC range, the proposed phantom also has a single-peak MR spectrum, isotropic Gaussian diffusion, and easily tunable ADC. As a result, the proposed design effectively provides a wide range of ADC while minimizing the influence of factors such as temperature, chemical shift artifacts in EPI as well as model mismatch caused by multiple signal sources. Therefore, the proposed phantom may prove useful in the development and quality assurance of quantitative diffusion MRI techniques.

Our observations of monoexponential signal decay with increasing b-value and ADC measurements independent of diffusion time support our hypothesis that hydrogen bonding between the solvent and the solute maintains Gaussian diffusion in the acetone-H₂O phantom, acetone-D₂O phantom as well as sucrose and PVP phantoms. The non-monoexponential decay observed in sucrose phantom at room temperature can be explained by the confounding effect of sucrose signal contributing to multi-exponential signal decay.

At 0°C, sucrose and PVP phantoms showed a single-peak spectrum, Gaussian diffusion, and easily tunable ADC. However, at 0°C the span of ADC values was limited to equal or less than the ADC of pure water at 0°C, ~1.1·10⁻³ mm²·s⁻¹(26). One way to extend ADC range of water-based phantom is to image them at higher temperatures. At room temperature, the ADC range of the water-based sucrose and PVP phantoms was higher than at ice-water temperature. However, at room temperature, signal from sucrose was observed. This leads to multi-exponential diffusion signal decay, which confounds DW-EPI of sucrose phantoms at room temperature. In contrast, no significant signal from PVP was observed at echo times used for DW-EPI and DW-MRS, even at room temperature. Although at shorter echo times, PVP signal was observed at room temperature, the rapid decay indicates very short T2 of PVP compared with water. This likely led to the lack of PVP signal in DW-EPI and DW-MRS that are typically acquired at much longer echo times. For this reason, the PVP phantom demonstrated monoexponential signal decay with increasing b-values, in good agreement with previous studies (21,22). Nevertheless, the main limitation for water-based phantoms at higher temperatures is the need for temperature control more sophisticated than ice-water bath, in order to attain reproducible ADC measurements (28). This requirement introduces significant complexity into the phantom setup, and may limit the widespread applicability of water-based phantoms at higher temperatures. Therefore, the proposed acetone-D₂O phantom may provide an effective approach to obtain a wide ADC range with simple ice-water temperature control.

This study had several limitations. First, the use of acetone poses some challenges. It is often desirable to tune the T1 and T2 of phantoms, in order to better mimic tissue properties and to optimize SNR. Certain salts such as copper sulfate and nickel chloride are commonly used to shorten the T1 and T2 of water. However, neither of these are soluble in acetone. Alternative agents that alter the relaxivity of acetone may address this limitation, and further investigation is needed to optimize the relaxation parameters of acetone. Another limitation of the proposed acetone-D₂O phantom is the need to eliminate the signal from the surrounding ice-water bath. In this study, MnCl₂ was added to the ice-water bath for this purpose. Importantly, this process does not result in substantial changes in the ice-water temperature.

Further, the potential for proton and deuteron exchange between D₂O and acetone may limit the shelf-life of the proposed phantom by generating unwanted water signal. In preliminary results, an H₂O peak appears at the fourth month after the phantom construction if stored at room-temperature. However when acetone-D₂O was stored in a freezer, no H₂O peak was detected a year after the construction of the phantom. However, systematic evaluation of the shelf life of the proposed acetone-D₂O phantom needs to be performed in future studies.

To demonstrate the utility of the proposed phantom for the validation of diffusion MRI techniques, multi-center studies must also be performed(22,38). Reproducibility across sites, field strengths and platforms is critical for the establishment of quantitative diffusion MRI techniques as quantitative imaging biomarkers. Diffusion MRI phantoms used in multi-site reproducibility studies need to show reproducible diffusion behavior over time and across sites.

In conclusion, this study has proposed and characterized the performance of an acetone-D₂O diffusion phantom. This phantom provides single-peak MR spectrum, Gaussian diffusion behavior and a wide range of tunable ADC, covering the entire physiological range of ADC values at 0°C. This phantom may have utility for the technical development of new diffusion MRI methods and for protocol harmonization and quality assurance in multi-center studies using quantitative diffusion MRI.