MR Thermometry for guidance of thermal therapy

S.M. Sprinkhuizen

Research output: ThesisDoctoral thesis 1 (Research UU / Graduation UU)

Abstract

The research described in this thesis has aimed to further investigate magnetic resonance (MR) thermometry (MRT) techniques as to become a reliable guidance tool for thermal therapy. The various physical processes allowing for temperature measurements based upon the MR signal are described. It is demonstrated that proton resonance frequency shift (PRFS)-based MRT, which employs the temperature dependence of the proton electron screening constant of water, is the method of choice. Potential sources of error in PRFS-based MRT are identified, in particular time varying magnetic field inhomogeneities and heat-induced changes of the magnetic susceptibility of tissue. A source of time varying magnetic field inhomogeneities in vivo is respiration. A study was conducted to quantify respiration-induced field inhomogeneities in the human breast, to assess its impact on PRFS-based MRT. The average field fluctuation due to regular respiration was 0.13 ppm and due to maximum capacity respiration 0.16 ppm. These numbers can be misinterpreted as temperature changes of 13 ºC and 16 ºC, respectively, when PRFS based MR thermometry is used during thermal treatment of breast cancer. In conclusion, respiration causes significant field fluctuations in the breast that should be corrected for to allow accurate MR thermometry in the human breast under free breathing circumstances. Furthermore, the influence of the temperature dependence of the susceptibility of tissue on PRFS-based MRT during thermal therapy was studied. Heating experiments were performed in a controlled phantom set-up to show the impact of temperature-induced susceptibility changes on PRFS-based temperature maps. To study the implications for a clinical patient, simulations were performed in a 3D breast model for the specific clinical case of MR-guided High Intensity Focused Ultrasound (HIFU) in the breast. The simulations showed that an ellipsoidal HIFU thermal spot of diameter 8 mm and length 20 mm and a maximum temperature increase of ΔT = 30 ºC led to temperature errors in the glandular tissue ranging between -8.6 ºC and +6.2 ºC, depending on the orientation of the focal spot. It was concluded that the influence of susceptibility changes may lead to significant temperature errors in PRFS-based MRT that are not to be neglected. Rather, a more thorough understanding of the relation between temperature and magnetic susceptibility was found to be required. Especially the temperature dependence of the susceptibility of fat tissue is of interest, since it is reported to be in the same order of magnitude as the temperature dependence of the proton electron screening constant of water. However, no values have been reported in literature on the temperature dependence of the susceptibility of human fat tissue. Therefore, a study was conducted in which this temperature dependence was measured for human breast fat tissue. In this work ex vivo measurements were performed on a 14.1 T five millimeter narrow bore NMR spectrometer. Breast fat tissue samples were collected from six subjects, directly post-mortem. The susceptibility was measured over a temperature range from 24 ºC to 65 ºC. A linear behaviour of the susceptibility over temperature was observed in all six samples. The resulting temperature dependence of the susceptibility of human breast fat ranged between 0.0038 and 0.0076 ppm/ºC. The outcome of this study supports the previous findings that the impact of heat-induced susceptibility changes of fat during thermal therapy in the breast may not be neglected. There is an additional issue regarding tissues containing fat. The electron screening constant of protons in fat is near independent of temperature which hampers PRFS-based temperature mapping in fatty tissues. However, the presence of fat may be used to our benefit by using it as a temperature independent reference signal which allows for absolute rather than relative MR temperature measurements. Multiple resonances can be separately detected in spectroscopic data. An image-based MR technique which allows for the acquisition of spectroscopic data at high temporal and spatial resolution is the multi gradient-echo (mGE) sequence. The application of the mGE sequence for MR thermometry purposes was further developed. The possibility of post-processing the mGE data into absolute temperature maps using time domain analysis of the magnitude of the mGE signals was investigated. In vitro experiments were performed to provide proof of concept for retrieving absolute temperature maps from the time domain analysis of mGE magnitude images. It is shown that this technique is insensitive to both field drift and local field disturbances. Furthermore, ex vivo bone marrow experiments were performed, using the fat resonance as a reference for absolute temperature mapping. It was shown that the post-processing based on the magnitude signal in the time domain allows for the determination of the resonance frequency difference between water and fat in bone marrow. However, it was concluded that the relation between this frequency difference and absolute temperature has to be examined more extensively because temperature dependent electron screening might not be the only factor influencing the frequency difference between water and fat in tissue. The post-processing of the MR signal to obtain temperature maps is based on certain assumptions with regard to the acquired signal. Specific situation exists in which the assumed properties of the acquired signal are incorrect. MR thermometry techniques that are based on the temperature dependence of the water proton resonance frequency are generally considered to be insensitive to static field inhomogeneities, because they either employ subtraction of successive phase images (PRFS-based MRT) or use an internal temperature-independent reference component (mGE-based MRT). It is shown that MR thermometry measurements are most certainly affected by the presence of static background field gradients, and in particular by background gradients that are aligned with the read-out gradient. Both theoretically and in phantom experiments it was shown that static background field gradients alter the effective echo time in gradient-echo acquisitions, which induces temperature errors in PRFS-based and mGE-based MR thermometry. The impact of static background gradients on MRT in vivo was assessed by static field gradient mapping in the breast of a volunteer. It was concluded that accurate field gradient mapping would facilitate better choices with regard to the direction and strength of the read-out gradient employed, to avoid errors due to background gradients in vivo. Finally, it is hypothesized that polyethylene glycol signals from pegylated liposomes may provide a temperature insensitive proton resonance frequency component than can serve as a reference for absolute MR temperature measurements. The feasibility of performing dynamic absolute MR thermometry using a multi-gradient echo sequence in combination with pegylated liposomes is evaluated. It was shown by a dilution experiment that at clinical relevant concentrations, the liposomal formulation and MR imaging parameters used in this study do not allow for precise absolute mGE-based MR temperature measurements with the PEG resonance as reference, due to insufficient signal-to-noise ratio. For translation to in vivo applications of this PEG-referenced MR thermometry technique, increased signal from the PEG resonance is thus required, which may be achieved by active targeting or by chemical adjustments to increase the PEG-load per liposome
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Utrecht University
Supervisors/Advisors
  • Viergever, Max, Primary supervisor
  • Bartels, Wilbert, Co-supervisor
Award date4 Nov 2010
Publisher
Print ISBNs978-90-393-5405-6
Publication statusPublished - 4 Nov 2010

Keywords

  • Econometric and Statistical Methods: General
  • Geneeskunde (GENK)
  • Geneeskunde(GENK)
  • Medical sciences
  • Bescherming en bevordering van de menselijke gezondheid

Fingerprint

Dive into the research topics of 'MR Thermometry for guidance of thermal therapy'. Together they form a unique fingerprint.

Cite this