5439

PRF thermometry accuracy at 7.0T for thermal MR applications with and without the presence of fat

Michal Pham¹, Yiyi Ji¹, Hendrik Paysen¹, Eva Oberacker¹, Till Huelnhagen¹, Thoralf Niendorf^1,2,3, and Lukas Winter¹

¹Berlin Ultrahigh Field Facility (B.U.F.F.), Max Delbrück Center for Molecular Medicine in the Helmholtz Association, Berlin, Germany, ²Experimental and Clinical Research Center (ECRC), a joint cooperation between the Charité Medical Faculty and the Max Delbrück Center for Molecular Medicine, Berlin, Germany, ³MRI.TOOLS GmbH, Berlin, Germany

Synopsis

Proton resonance frequency (PRF) shift temperature mapping is influenced by B₀ field drift, temperature dependent electromagnetic property changes that alter phase velocity, alpha calibration and temperature dependent magnetic susceptibility changes. This work details PRF thermometry at 7.0T of a thermal MR setup. In a homogeneous agarose phantom PRF reading accuracy is within the accuracy of fiber optic sensor readings (0.29±0.21)°C. Temperature dependent magnetic susceptibility changes in pure fat lead to significant temperature reading errors. However, for more realistic thermal MR applications in the human brain (ΔT<10°C, volume fat fractions <10-20%), temperature reading accuracy is comparable to water-based samples.

Purpose

Proton resonance frequency (PRF) shift temperature mapping is influenced by B₀ field drift, temperature dependent electromagnetic property changes that alter phase velocity, alpha calibration and temperature dependent magnetic susceptibility changes^1-2. The findings on temperature reading accuracy are mostly restricted to focused ultrasound ablation with high temperature changes of ΔT>10°C in a short time (<1min), small hotspot locations (<1cm) and low field (B₀=1.5T) MR thermometry readings, while validation studies using fiber optic sensor readings are mostly missing³. This work details PRF thermometry at 7.0T of a thermal MR setup (f=300MHz) that has the SNR advantage of an integrated system using dipole antennas at UHF^4-5 and targets temperature changes of ΔT≤10°C in several minutes (>5min) in the whole field of view. PRF reading accuracy has been validated with fiber optic sensor readings taking into account temperature dependent magnetic susceptibility changes of fatty tissue.

Methods

The influence of B₀ drift, temperature dependent changes of electromagnetic properties and magnetic susceptibility on a potential temperature change for ΔT=10°C was investigated. For this purpose two agarose phantoms (dimensions: (15x20x8)cm³ and (18x26x9)cm³; σ=1.03±0.01S/m, ε=71.9±2.8; T₁=1560±84ms) were constructed (Fig.1a-b) based on electromagnetic field simulations (Sim4Life, Zurich Med Tech, Zurich, Switzerland). B₀ drift correction was performed in five oil reference probes positioned outside the phantom. A constant drift was determined as an average over all oil reference probes.

Thermal MR (RF heating and MR imaging) was applied at B₀=7.0T (f=300MHz) with a bow tie dipole antenna⁴ driven over a high peak and average power T/R switch⁶ (P_{peak_max}≈4kW, P_avg=70W) for a duration of 3x3min. A dual gradient echo imaging technique was applied for PRF thermometry (spatial resolution=(1.5x1.5x4)mm³, TR=61ms, TE₁=2.26ms and TE₂=6.34ms, receiver bandwidth=900Hz/pixel). The proportionality constant alpha was calibrated in the homogenous phantom (Fig.1a), using four fiber optic sensors at different locations and four different measurement time points. The effects of temperature dependent magnetic susceptibilities of fat were investigated using four different samples a) agarose b) 10% c) 20% volume concentration of sunflower oil in in an agarose (2g/100ml) deionized water mixture and d) pure sunflower oil. Three fiber optic sensors (Neoptix, Quebec, Canada) were used for independent temperature validation.

Results

In the homogenous phantom the PRF temperature proportionality constant α was -0.0106ppm/°C. This value correlates well with the values found in the literature⁷ of -0.0103ppm/°C. B₀ field drift was 0.004ppm after 8min of measurement, which accounted to an error of (2.2±0.5)°C in comparison to fiber optic readings (Fig.2a). Using the oil reference probes the error could be corrected to (0.35±0.49)°C using a single echo (TE=2.26ms) (Fig. 2b) and (0.29±0.21)°C using a double echo (TE₁=2.26ms, TE₂=6.34ms) (Fig.2c). Applying all corrections, PRF temperature reading accuracy approaches fiber optic sensor reading accuracy of ±0.2°C for the homogenous phantom. A single echo readout has only slightly higher inaccuracy, however temperature dependent phase velocity changes are expected to be larger for larger heating volumes, suggesting that a double echo PRF method should be applied.

Temperature simulations show (Fig. 3) that a dipolar heating pattern around the central compartment is already expected from the E-field distribution alone and is stronger pronounced for 100% oil with lower electrical conductivity and permittivity values altering the wavelength. On top of these changes temperature dependent magnetic susceptibility affects temperature reading accuracy. PRF temperature maps for all probes (agarose, 10%, 20% and 100% fat concentration) and fiber optic locations are displayed in Fig.4. Temperature evolution of the fiber optic readings is displayed in Fig.5 For pure sunflower oil, the temperature accuracy of the PRF method was (3.95±1.24)°C showing a strong effect of temperature dependent magnetic susceptibility changes⁸. For more realistic fat fractions found in vivo⁹ the accuracies were (0.93±0.95)°C for a 10% and (0.88±0.72)°C for a 20% fat fraction, which is similar to the accuracy found for agarose (0.81±0.52)°C (Fig.5).

Conclusion

The accuracy of the PRF shift method of (0.29±0.21)°C is within the accuracy of fiber optic sensor readings in a homogeneous phantom. Temperature dependent magnetic susceptibility changes in pure fat lead to significant temperature reading errors⁸. However, for thermal MR applications in the human brain within a temperature change of ΔT<10°C and small volume fat fractions <10-20%, temperature reading accuracy is comparable to the agarose samples. These initial results are encouraging and an important step towards accurate thermal dose determination (CEM43)^10-11 for safety evaluations and thermal therapies¹².

Acknowledgements

This work was supported in part (T.N., L.W.) by the German Federal Ministry of Education and Research, “KMU-innovativ”: Medizintechnik 13GW0102.

References

¹ Rieke V and Pauly KB, "MR Thermometry" JMRI, 2008; 27(2):376-390

² Winter L, et al., "Magnetic resonance thermometry: Methodology, pitfalls and practical solutions", Int J Hyperthermia, 2015; 6736:1-13

³ McDannold N, "Quantitative MRI-based temperature mapping based on the proton resonant frequency shift: review of validation studies", Int J Hyperthermia, 2005;21(6):533-546

⁴ Winter L, et al., "Design and Evaluation of a Hybrid Radiofrequency Applicator for Magnetic Resonance Imaging and RF Induced Hyperthermia: Electromagnetic Field Simulations up to 14.0 Tesla and Proof-of-Concept at 7.0 Tesla", PLOS ONE, 2013; 8(4):e61661

⁵ Winter L, et al., "Thermal magnetic resonance: physics considerations and electromagnetic field simulations up to 23.5 Tesla (1GHz)", Radiat Oncol, 2015; 10:201

⁶ Ji Y, et al., "High Power Transmit/Receive Switch for Thermal MR at 7.0 Tesla", ISMRM Workshop on UHF MRI, 2016; #19

⁷ Tarasek MR, et al., "Validation of MR thermometry: Method for temperature probe sensor registration accuracy in head and neck phantoms", Int J Hyperthermia, 2014; 30(2):142-149

⁸ Sprinkhuizen SM, et al., "Temperature-induced tissue susceptibility changes lead to significant temperature errors in PRFS-based MR thermometry during thermal interventions", MRM, 2010; 64(5):1360-1372

⁹ Woodard HQ and White DR, "The Composition of Body Tissues", Brit J Radiol, 1986; 59(708):1209-1218

¹⁰ Van Rhoon GC, et al., "CEM43° C thermal dose thresholds: a potential guide for magnetic resonance radiofrequency exposure levels?", Eur Radiol, 2013; 23(8):2215-2257

¹¹ Murbach M, et al., "Thermal Tissue Damage Model Analyzed for Different Whole-Body SAR and Scan Durations for Standard MR Body Coils", MRM, 2013; 71:421-431

¹² Sneed PK, "Survival benefit of hyperthermia in a prospective randomized trial of brachytherapy boost+/-hyperthermia for glioblastoma multiforme", Int J Radiat Oncol Biol Phys, 1998; 40(2):287-295

Figures

Figure 1 – Schematic and photograph of (a-b) the homogeneous phantom and (c-d) the inhomogeneous phantom used in the RF heating experiments at B₀=7.0T. The inhomogeneous phantom (c) contains several compartments for inclusion of various material probes. Both phantoms have a 4x4 matrix of PET tubes for fiber optic sensor positioning.

Figure 2 – Results of the PRF temperature mapping accuracy based on a gradient echo. (a) Results obtained applying a double echo without B₀-drift correction. The slope of -0.074 translates to α=-0.0097ppm/°C and a temperature accuracy of 2.2±0.5°C (maximal deviation 2.9°C) versus fiber optic readings. b) Applying a B₀ drift correction and a single echo (TE₁=2.26ms), results in a slope of -0.047, α=-0.0111ppm/°C and a reading accuracy of 0.35±0.49°C with a maximal deviation of 1.74°C. c) A B₀-drift correction and double echo improves these results to-0.081 , α=-0.0106ppm/°C, a reading accuracy of 0.29±0.21°C and a maximal deviation of 0.76°C.

Figure 3 – Thermal simulations based on electromagnetic field simulations of the inhomogenous phantom with a) agarose b) 10% c) 20% and d) 100% volume fraction of sunflower oil in the central compartment after 3min, 6min and 9min of RF heating. Please note that the dipolar heating pattern around the central compartment is an effect of the RF field of the antenna. This effect is stronger for the 100% oil probe (d) than for agarose (a) due to lower permittivity and conductivity of oil versus agarose, altering the wavelength in the material and the effective E-field distribution.

Figure 4 – PRF temperature maps acquired in the compartment phantom after 3min, 6min and 9min for agarose, 10%, 20% and 100% oil volume fraction in the central compartment. The black squares display the locations of the fiber optic sensors for temperature monitoring, which were positioned 5mm below the surface of the phantom.

Figure 5 – Fiber optic sensor readings versus PRF thermometry for (a) agarose (b) 10% (c) 20% and (d) 100% volume fraction of sunflower oil at three distinct locations in the compartment phantom (Fig. 4). PRF temperature readings in locations 1 and 2 (averaged over a 3x3 voxel region) are displayed with a red and a black square respectively. The error bars represent the standard deviation. Temperature drops indicate MR image acquisition, where RF heating has been interrupted. The temperature evolution of the probe located in the central compartment was measured with a fiber optic probe (location 3, blue line).

Proc. Intl. Soc. Mag. Reson. Med. 25 (2017)

5439