Skip to Content

Yves De Deene Homepage

Quantitative MRI

Introduction

Innovative radiation delivery techniques have been developed over the last two decades that enable a better delivery of ionizing radiation to the tumour volume while sparing the healthy surrounding tissue [1-3]. Dedicated software developments combined with increasing computer power enable the prediction of dose distributions obtained with these different treatment modalities. Uncertainties with respect to the position of the treatment target are further reduced by incorporating imaging techniques on the treatment machine (Image Guided Radiotherapy) [4]. The increased complexity and capability of high-precision conformal radiation treatments has consequently increased the need for three dimensional (3D) dosimetric quality assurance [5] to ensure the delivered radiation dose matches the intended radiation dose. The combined use of these new technologies increases tumour control and reduces complications for the patient [6].

As the radiation dose can now be delivered to the tumour volume more precisely with steep dose gradients between the tumour and the surrounding healthy tissue, other important challenges arise. Until now, in the practice of radiation therapy, the tumour has been considered as a solid homogeneous body that is invariant over time. However, it is well recognized by the radiation oncology community that this does not correspond with the actual situation for two reasons: (i) A physical tumour is biologically heterogeneous [7]. Different physiological regions can be distinguished. Some regions may contain proliferating cells while other regions may be necrotic. The degree of oxygenation of tumour cells can vary considerably as tumours are provided with an abnormal vascular network. Abnormal angiogenesis is responsible for anoxic and hypoxic tumour regions. Also well known is that hypoxic tumour cells are more resistant to radiation than normoxic tumour cells [8]. (ii) Most radiation therapy is fractionated (i.e. the patient receives incremental doses on a daily basis, over the course of several weeks). During this period, the tumour changes, partially as a response to the fractionated radiation treatment [9]. Re-oxygenation, redistribution, repopulation and tumour shrinkage may all modify the initial spatial and functional pattern on which the prescribed radiation dose distribution is based.

Theoretical studies and accumulated clinical data strongly suggest that treatment plans need to be designed to exploit patient- and tumour-specific biological features on an individual basis [10-14]. Anatomy based medical imaging modalities (such as X-ray CT, conventional MRI and ultrasound) are limited in detecting biological changes when the gross morphology of the tumour (anatomically delineated tumour volume) does not change during therapy. Functional biomedical imaging, defined as the non-invasive characterization of tissues in terms of biochemical pathways or physiology, has recently become more available in clinical practice. Functional imaging modalities such as positron emission tomography (PET), single-photon emission computed tomography (SPECT), magnetic resonance spectroscopy (MRS), and dynamic contrast-enhanced MRI and CT may reveal occult carcinoma and provide the radiation oncologist with additional information on tumour physiology such as blood flow, vascular permeability, proliferation rate and tumour oxygenation. Some early attempts have been undertaken to include biological targets into the radiation treatment planning. However, in these preliminary studies, because of a lack of quantitative physiological imaging, a linear relation is considered between target dose and the image signal intensity in the functional images [13]. It is recognized that the choice to use a linear relation between the target dose and the image signal intensity in a positron emission tomography (PET) image after uptake of radioactive fluoro-de-oxy-glucose 18FDG (18FDG-PET) does not have a sound base but is the result of a limited availability of quantitative physiological imaging techniques. In order to translate the biological data to a prescription radiation dose, more quantitative detail is required (e.g. cellular density, oxygen concentrations (pO2), pH, vascular density, blood fraction, molar metabolite concentrations).

Quantitative MRI Guided Cancer Treatment

Magnetic Resonance Imaging (MRI) is a promising candidate for mapping essential bio-functional information of the tumor. Moreover, MRI is non-invasive and does not expose the patient to additional ionizing radiation. MRI relies on the evolution of nuclear spin magnetization after the excitation of the spin system by resonant radiofrequency pulses (typically with a frequency of 120 MHz in a 3 Tesla MRI scanner). The time evolution of the nuclear spin magnetization is determined by molecular interactions. As a result, the image contrast in MR images is determined by the molecular environment from which the signal is received.

Quantitative MRI

Figure 1 - Schematic representation of the different steps in quantitative MRI guided cancer therapy. By use of adequate image sequence design, functional MR images are recorded in which the contrast is determined by specific biological processes (a). The contrast weighted MR images are translated into quantitative images that display a specific MRI physical parameter (such as relaxation times, diffusion coefficients, etc.) (b). These scanner independent MRI physical parameters can be translated into physical parameters such as cell size, pH, metabolite concentrations and oxygen pressure (pO2) (c) that serve as input to radio-biological models that can predict the effectiveness of ionising radiation onto cancer cell death in each pixel. The output of this model is a map of prescribed radiation dose (d) that serves as template for radiation treatment optimization (e). Target biomedical engineering research actions are indicated (encircled).

A general outline of the approach in quantitative MRI guided cancer therapy is illustrated in figure 1. In a first step, MRI pulse sequences are developed and optimized to acquire contrast-weighted images (figure 1a) that are correlated with tumor biology. Both endogenous and exogenous MRI biomarkers can be explored. In order to increase the sensitivity of MRI, hyperpolarization techniques are being developed. The signal intensity in each image voxel of the contrast-weighted images is also scaled by the MRI scanner by an arbitrary gain factor in order to obtain an optimal dynamic range of pixel intensity values. A quantitative MR image can be calculated (figure 1b) from a set of different contrast-weighted images. The pixel intensity in each voxel of the quantitative MR images is uniquely correlated with an MRI physical property (such as T1 and T2 relaxation times, molecular diffusion coefficients, etc.) and is in general independent of the scanner type. Hence, quantitative MR images are well-suited in multi-center studies as the data acquired in different centers is comparable. The correlation between the pixel intensity in (‘semi-quantitative’) contrast-weighted images and MRI physical properties is described by the theory of the evolution of nuclear magnetization under the influence of a time sequence of radiofrequency pulses and magnetic field gradients that are applied during image acquisition. Algebraic calculations or computer simulations of the evolution of the nuclear magnetization in an imaging voxel are used to correlate the signal intensity in the contrast-weighted MR images with the MRI physical property. The inverse relation can then be applied to derive quantitative MR images. Modelling the relation between the molecular interactions and the evolution of the nuclear magnetization enable the translation of quantitative MR images into quantitative parametric microstructure images (displaying cellular density and size distributions) and quantitative biochemical parametric images (displaying endogenous metabolite and exogenous tracer concentrations, acidity (pH) and oxygen concentrations) (figure 1c). The prescribed radiation dose distribution in radiotherapy can be uniquely assigned on the basis of the quantitative physiological and biochemical images through radiobiological models that describe the probability of cell death with radiation taking into account the specific cellular environment (figure 1d). The prescribed radiation dose to each voxel of tissue serves as input to treatment planning software that determines an optimal radiotherapy treatment protocol (figure 1e).

Due to its non-invasive harmless character, bio-functional MRI can be repeated several times during the course of a fractionated radiation treatment enabling adaptive cancer treatments without additional exposure of the patient to ionizing radiation.

Quantitative MRI Tools

Non-invasive analysis of tissue microstructure

Upon external stimuli, such as radiation or chemotherapeutics, direct cellular mechanisms such as damage to the cell membrane and other cell structures or indirect cellular mechanisms through expression and inhibition of metabolic specific signalling pathways [15-17] lead to changes in cell structural properties. These cell structural properties comprise changes in cell volume, cell density, interstitial water and cell permeability [18-19]. Changes in MRI contrast properties in tumours such as relaxation and diffusion upon radiation treatment or chemotherapy have been observed by several authors [20-24] and are suggested as potential biomarkers. Although changes in MRI contrast reflect cell structural changes, one single MRI contrast is determined by several cell structural properties. In addition, the obtained image contrast and even quantitative measured MR properties may depend significantly on MRI pulse sequence parameters. Moreover, in human tissue, several different tissue components are encountered. This may obscure any generalizations on cell response based on individual MRI contrast properties, such as T1, T2 and apparent diffusion coefficient (ADC). For example, in diffusion MRI, the dependence of the ADC on the diffusion time is often overlooked in several studies. Another approach is needed to separate the different cell structural properties in quantitative MRI measurements.

Quantitative MRI

Figure 2 - Two examples of simple microstructures for which agreement is found between computational simulations and experimentally derived MRI diffusion properties. A first structure consists of a synthetic porous polymer with polygonal micro-capillaries visible with optical microscopy (a). The image is segmented and a random walk Monte Carlo simulation with 1000 random walkers is run (b). The measured ADC reveals directional anisotropy and dependence on diffusion time which is in accordance with the simulated results (c). The molecular self-diffusion of water in yeast cells (d) is modelled in 3D (e) and demonstrates a non-linear behaviour in the diffusion-weighted signal which is strongly influenced by the cell-membrane permeability (f).

Random walk models of molecular self-diffusion of water are being developed that help in correlating the ADC acquired as a function of diffusion time with the cellular microstructure. It has been found crucial to validate these computational models against experimental diffusion data obtained in well-designed and controlled laboratory conditions.

Non-invasive mapping of tumour oxygenation and vascularization

It is well-known that the vasculature in tumours is perturbed [25] which leads to an inefficient supply of oxygen and nutrients to certain regions in the tumour volume resulting in necrotic and hypoxic tumour regions. Hypoxic tumour cells are characterized by an increased radiotherapy- and chemotherapy-resistance. Intra-tumour hypoxic regions may require three times the radiation dose needed to achieve comparable tumour control as in normoxic tissue. The reduced radiation response in hypoxic tumour cells is attributed to a decreased contribution of indirect cell damage by oxygen radicals and to gene expression leading to a delay in cell proliferation and to the production of stress proteins and an increase in tumour progression [8]. Different strategies are being developed to increase the effectiveness of radiation treatment in hypoxic regions [26]. Different oxymetry techniques have been applied, including BOLD MRI and 19F MRI relaxometry, but the inter-relation between invasive and non-invasive techniques needs to be better defined [27]. Quantitative hypoxia imaging may help in discriminating patients that would benefit from anti-hypoxia directed therapies [28,29]. A profound knowledge of the evolution of hypoxic tumour regions during cancer treatment may have a significant influence on treatment strategies and on the understanding of radiobiological mechanisms.

The correlation between blood flow, vascular density, oxygen consumption and oxygen concentration in tissue can be best modelled in vitro under well-controlled laboratory conditions. Therefore, a biologically active perfusion phantom has been constructed that consists of small capillaries that are perfused with a blood mimicking fluid (figure 3). The bio-perfusion phantom is MRI compatible and can be scanned on a clinical MRI scanner with Dynamic Contrast Enhanced (DCE) MRI sequences and can be used to validate experimental methods to map oxygen concentration with MRI [30]. This experimental model can then serve as a test model of computational modelling that takes the vascularization and diffusion of oxygen into the tissue into account [31].

DCE and oximetry

Figure 3 - In vitro DCE and multi-nuclear 19F MRI oximetry study with a perfused bioreactor helps in optimizing the MRI imaging protocol and in validating computational simulations of vascularization and oxygenation of tissue. A perfused 3D scaffold of cells embedded in a collagen matrix is placed in the MRI scanner and is perfused with a blood-mimicking fluid through small capillaries using a peristaltic pump. At a certain moment in time, a contrast agent is injected and images are recorded over time. The evolution of the signal intensity in each voxel is a measure of the permeation of the contrast agent in the tissue and can be analysed using kinetic models. In addition, the fluid can be replaced with a perfluorocarbon solution enabling regional mapping of oxygen pressure through the 19F MRI signal.

For in vivo measurements, Fluor-containing artificial blood substitutes can be administered in large quantities to the patient. The artificial blood substitute Fluosol® contains perfluorocarbons suspended in an albumin solution and is FDA approved. While the permeability, size and density of the capillaries in the perfusion phantom can be measured independently using alternate techniques, these parameters are unknown in the in vivo situation but can be determined from DCE MRI compartmental modelling. A computational model of radiobiological effectiveness incorporating the effect of oxygen on cell survival [32] can then be developed.

Non-invasive mapping of metabolite concentrations

In vivo magnetic resonance spectroscopy (MRS) has been widely used in the investigation of metabolic changes that originate from an alteration of the molecular pathways during carcinogenesis [33,34]. It has also been suggested that in vivo MRS may have potential in assessing early tumour treatment response non-invasively [35-37]. Only a few studies have been focused on the quantitative determination of metabolite concentrations in cancer. Quantitative MRS involves a wide range of processing steps before ultimately determining ‘absolute’ concentration values [38,39]. Normally, several pre-processing steps are performed on the acquired raw MR spectra. All these ‘fine-tuning’ procedures can have a substantial effect on the derived concentrations. In cancer, the baseline correction can be a tricky endeavour because of the signal originating from lipids. However, these lipid signals and other macromolecules which create the baseline of the spectrum, also comprise tumour specific information [40]. Therefore, it is challenging to quantify these lipid fractions. Our research group has gained experience in absolute quantification of metabolites by use of in vivo MRS [41-42]. Several methods have been applied to correlate the acquired MR spectrum with metabolite concentrations (figure 5).

Spectroscopy

Figure 4 -Absolute in vivo MR spectroscopy can be performed in different ways but requires several corrections that when not accounted for can result in significant errors.

The acidity (pH) of tumours has also been found to be an important biomarker of carcinogenesis. The pH can be determined by spectroscopy as the chemical shift between certain resonances (e.g. between inorganic phosphate and ATP) is pH dependent. Another method to determine pH relies on the chemical exchange between hydrogen atoms of amide groups on proteins and hydrogen atoms on water molecules [43].

Conclusions

Quantitative Magnetic Resonance Imaging has great potential in providing quantitative input into radio-biological models that can be used to predict treatment efficiency. In vitro studies are very useful in the quantification of the MRI methods as the microstructure and chemical properties of the test phantoms can be easily controlled and varied. They are indispensable in the validation of computational biophysical models. Quantitative diffusion MRI can be used to determine microstructure properties such as cellular densities and cell membrane permeability. Multi-nuclear 19F MRI can be used to determine oxygen concentrations and absolute MR spectroscopy can provide metabolite concentrations in vivo. These quantitative biological properties can serve as input to radiobiological models that can help in sophisticating the existing radiation dose prescription to tissue in radiotherapy treatment planning. We advocate that the incorporation of quantitative MRI will enable patient individualized biologically adaptive therapy and increase the efficiency of cancer treatments.

References

  1. Webb S, “The physical basis of IMRT and inverse planning”, Br. J. Radiology 2003; 76: 678-89.
  2. Palma DA, Verbakel W, Otto K and Senan S 2010, “New developments in arc radiation therapy: A review”, Cancer Treatment Reviews 2010; 36: 393-9.
  3. Mackie TR,”History of tomotherapy” (Review), Phys. Med. Biol. 2006: 51; R427-53.
  4. Verellen D, De Ridder M, Linthout N, Tournel K, Soete G and Storme G, “Innovations in image-guided radiotherapy”, Nature Reviews: Cancer 7, 949-60.
  5. Baldock C†, De Deene Y†, Doran S, Ibbott G, Jirasek A, Lepage M, McAuley KB, Oldham M, Schreiner LJ. Polymer gel dosimetry (Review). Phys Med Biol, 2010; 55(1): R1-R63.
  6. Veldeman L, Madani I, Hulstaert F, De Meerleer G, Mareel M and De Neve W, “Evidence behind use of intensity-modulated radiotherapy: a systematic review of comparative clinical studies”, Lancet Oncol. 2008; 9: 367-75.
  7. Siemann DW, “The tumor microenvironment: A double-edged sword”, Int. J. Rad Oncol. Biol. Phys. 1998: 42; 697-9.
  8. Harris AL 2002, “Hypoxia – A key regulatory factor in tumour growth”, Nature Reviews: Cancer 2002: 2; 38-47.
  9. Turesson I, Carlsson J, Brahme A, Glimelius B, Zackrison B and Stenerlow B, “Biologically optimized radiation therapy”, Acta Oncol 2003; 42, 92-106.
  10. Stewart RD and Li XA, “BGRT: Biologically guided radiation therapy – The future is fast approaching!”, Med Phys 2007; 34: 3739-51.
  11. Brahme A, Nilsson J and Belkic D, “Biologically optimized radiation therapy”, Acta Oncol 2001; 40: 725-34.
  12. Ling CC, and Li XA, “Over the next decade the success of radiation treatment planning will be judged by the immediate biological response of tumor cells rather than by surrogate measures such as dose maximization and uniformity”, Med Phys 2005; 32: 2189-92.
  13. Vanderstraeten B, Duthoy W, De Gersem W, De Neve W and Thierens H 2006, “[18F]fluoro-deoxy-glucose positron emission tomography ([18F]FDG-PET) voxel intensity-based intensity-modulated radiation therapy (IMRT) for head and neck cancer”, Radiother Oncol 2006; 79: 249-58.
  14. Cao Y, Tsien CI, Nagesh VY, Junck L, Ten Haken R, Ross BD, Chenevert TL and Lawrence TS 2006, “Clinical investigation survival prediction in high grade gliomas by MRI perfusion before and during early stage of RT”, Int J Radiat Oncol Biol Phys 2006; 64: 876-85.
  15. Szumiel I 1994, “Review: Ionizing radiation-induced cell death”, Int J Radiat Oncol Biol Phys 66, 329-41.
  16. Valerie K, Yacoub A, Hagan MP, Curiel DT, Fisher PB, Grant S and Dent P 2007, “Radiation-induced cell signaling: inside-out and inside-in”, Mol Cancer Ther 2007; 6: 789-801.
  17. Okada H and Mak TW 2004, “Pathways of apoptotic and non-apoptotic death in tumour cells”, Nat Rev Cancer 4, 592-603.
  18. Minard KR, Holtom GR, Kathmann LE, Majors PD, Thrall BD and Wind RA, “Simultaneous 1H PFG-NMR and confocal microscopy of monolayer cell cultures: Effects of apoptosis and necrosis on water diffusion and compartmentalization”, Magn Reson Med 2004; 52: 495-505.
  19. Lyng H, Haraldseth O and Rofstad EK 2000, “Measurement of cell density and necrotic fraction in human melanoma xenografts by diffusion weighted magnetic resonance imaging”, Magn Reson Med 2000; 43: 828-36.
  20. Belfi CA, Medendorp SV and Ngo FQH 1991, “The response of the KHT sarcoma to radiotherapy as measured by water proton NMR relaxation times: relationships with tumor volume and water content”, Int J Radiat Oncol Biol Phys 1991: 20; 497-507.
  21. Chenevert TL, McKeever PE and Ross BD 1997, “Monitoring early response of experimental brain tumors to therapy using diffusion magnetic resonance imaging”, Clinical Cancer Research 1997: 3; 1457-66.
  22. Mardor Y, Pfeffer R, Spiegelmann R, Roth Y, Maier SE, Nissim O, Berger R, Glicksman A, Baram J, Orenstein A, Cohen J S, Tichler T 2003, “Early Detection of Response to Radiation Therapy in Patients With Brain Malignancies Using Conventional and High b-Value Diffusion-Weighted Magnetic Resonance Imaging”, J Clin Oncol 2003; 21: 1094-100.
  23. Rodrigues LM, Howe FA, Griffiths JR and Robinson SP 2004, “Tumor R2* is a prognostic indicator of acute radiotherapeutic response in rodent tumors”, J Magn Reson Imaging 2004; 19: 482-8.
  24. Sierra A, Michaeli S, Niskanen JP, Valonen PK, Gröhn HI, Ylä-Herttuala S, Garwood M and Gröhn OH 2008, “Water spin dynamics during apoptotic cell death in glioma gene therapy probed by T1rho and T2rho”, Magn Reson Med 2008; 59: 1311-9.
  25. Jain RK, “Normalization of tumor vasculature: An emerging concept in antiangiogenic therapy”, Science 2005; 307: 58-62.
  26. Overgaard J and Horsman MR, “Modification of hypoxia-induced radioresistance in tumors by the use of oxygen and sensitizers”, Seminars in Radiation Oncology 1996; 6: 10-21.
  27. Padhani AR, Krohn KA, Lewis JS and Alber M, “Imaging oxygenation of human tumours”, European Radiology 2007; 17: 861-72.
  28. Jain RK, “Normalization of tumor vasculature: An emerging concept in antiangiogenic therapy”, Science 2005; 307: 58-62.
  29. Overgaard J and Horsman MR, “Modification of hypoxia-induced radioresistance in tumors by the use of oxygen and sensitizers”, Seminars in Radiation Oncology 1996; 6: 10-21.
  30. Baete SH, Vandecasteele J, Colman L, De Neve W, De Deene Y, “An oxygen-consuming phantom simulating perfused tissue to explore oxygen dynamics and F-19 MRI oximetry.” Magn Reson Mat Phys Biol Med 2010; 23(4): 217-26.
  31. Baete SH, Vandecasteele J, De Deene Y, “19F MRI oximetry: simulation of perfluorocarbon distribution impact.” Physics in Medicine and Biology 2011; 56(8): 2535-57.
  32. Dasu A and Toma-Dasu I, “Treatment modeling: The influence of micro-environmental conditions”, Acta Oncologica 2008; 47: 896-905.
  33. Gillies RJ and Morse DL, “In vivo MRS in cancer”, Annu. Rev. Biomed. Eng. 2005; 7: 287-326.
  34. Kwock L, Smith JK, Castillo M, Ewend MG, Collichio F, Morris DE, Bouldin DE and Cush S, “Clinical role of proton magnetic resonance spectroscopy in oncology: brain, breast and prostate cancer”, Lancet Oncol. 2006; 7: 859-68.
  35. Manton DJ, Chaturvedi A, Hubbard A, Lind MJ, Lowry M, Maraveyas A, Pickles MD, Tozer DJ and Turnbull LW, “Neoasjuvant chemotherapy in breast cancer: early response prediction with quantitative MR imaging and spectroscopy”, Br. J. Cancer 2006; 94, 427-35.
  36. Vaidya SJ, Payne GS, Leach MO and Pinkerton CR, “Potential role of magnetic resonance spectroscopy in assessment of tumour response in childhood cancer”, Eur. J Cancer 2003; 33: 728-35.
  37. Lee SC, Poptani H, Pickup S, Jenkins WT, Kim S, Koch CJ, Delikatny EJ and Glickson JD, “Early detection of radiation therapy response in non-Hodgkin’s lymphoma xenografts by in vivo 1H magnetic resonance spectroscopy and imaging”, NMR in Biomedicine 2009; 23(6): 624-32.
  38. Tofts P, “Quantitative MRI of the brain: measuring changes caused by disease”, ed. John Wiley & Sons; 2003.
  39. de Graaf RA, “In vivo NMR spectroscopy: Principles and Techniques”, John Wiley & Sons; 2007.
  40. Schmitz JE, Kettunen MI, Hu D and Brible KM, “1H MRS-visible lipids accumulate during apoptosis of lymphoma cells in vitro and in vivo”, Magn. Reson. Med. 2005; 54: 43-50.
  41. Ozdemir MSR, Reyngoudt H, De Deene Y, Sazak HS, Fieremans E, Delputte S, D'Asseler Y, Derave W, Lemahieu I, Achten E 2007. “Absolute quantification of carnosine in human calf muscle by proton magnetic resonance spectroscopy”, Phys. Med. Biol. 2007; 52(23): 6781-94.
  42. Reyngoudt H, De Deene Y, Descamps B, Paemeleire K, Achten E, “H-1-MRS of brain metabolites in migraine without aura: absolute quantification using the phantom replacement technique”, Magnetic Resonance Materials in Physics Biology and Medicine 2010; 23(4): 227-41.
  43. Sun PZ and Sorensen AG, Imaging pH using the chemical exchange saturation transfer (CEST) MRI: Correction of concomitant RF irradiation effects to quantify CEST MRI for chemical exchange rate and pH, Magnetic Resonance in Medicine 2008; 60(2): 390-7.