MRI-guided Radiation Therapy
Cornelis van den Berg1

1Department of Radiation Oncology, Centre for Image Sciences, University Medical Center Utrecht, Netherlands


This educational discusses the use of MRI in radiation therapy with a focus on MRI-guided radiation therapy. It explains the technological development in relation to how this disruptive technology can change radiation therapy improving outcome and patient toxicity.

Target audience

Clinicians, physicist and engineers wanting to understand how MRI guidance is transforming modern radiation therapy.

Learning objectives

  • Understand the role of MRI in the pre-treatment radiation planning process.
  • Understand how the MR guidance of radiation treatment can reduce uncertainties and lead to better treatments.
  • Understand the MRI requirements for targeting, planning, tracking and response assessment in MRI-guided radiotherapy


External beam radiotherapy basically boils down to treating an invisible target (tumor) with an invisible high energy photon beam. This technological challenge has been the fundamental drive behind the introduction of imaging technology in radiotherapy. This development started with the use of x-ray projections revealing only bony anatomy, advancing to 3D images by means of CT imaging in the late nineties and finally now, to direct onboard MRI during radiation therapy. The fact that MRI by means of its 3D imaging capability and its superior soft tissue contrast allows direct visualization of target structures in 3D space, is transforming modern radiation therapy. In this talk I will explain the various functions and use of MR imaging technology in a modern workflow for MR guided radiotherapy treatment.

Pre-treatment MRI for target definition and radiation planning.

A radiotherapy process starts by identification of the target, i.e. delineation of the gross tumor volume (GTV) and organs-at-risk by a radiation oncologist. This process takes place prior to treatment and is often called the simulation process and includes imaging, delineation and the design of a patient tailored radiation plan. Accurate target delineation is essential as any error here will persists as a systematic error throughout the whole treatment [1]. The superior soft tissue contrast of MRI allows better tumor visibility aiding target definition[2]. Functional imaging like diffusion weighted and dynamic contrast enhanced MRI are nowadays routinely used in clinics to aid in GTV delineation [3]. CT images are still needed to provide electron density maps for radiation planning. First clinics have started to solely rely on MRI, i.e. MR-only simulation where synthetic CTs are derived from MR images [4-6]. This eliminates CT-MR registration errors, streamlines workflow and logistics and a patient only needs to undergo a single MRI exam. The main technological challenge is the generation of synthetic CTs with clear visualization of bony structures with MRI [7,8]. For this purpose specialized sequences (Ultra-short or zero TE sequences) as well as image processing (CT-MR atlas based segmentation) have been employed. However, more recently deep learning based synthetic CT generation in combination with standard 3D gradient echo sequences are emerging as the prevailing method [9,10]. 4D MRI is being used to characterize motion of target structures to determine respiratory motion margins of a patient specific basis. Various techniques are used for 4D MRI, however, a 3D golden-angle radial readout is becoming popular for abdominal and thoracic tumor sites [11-13]. In such a technique, k-space lines are retrospectively ordered based on a navigation signal (e.g. self navigation by central kz spoke signal) leading to 3D images for several phases of the respiration cycle. This information can be used to determine the tumor mid-vent position and personalized respiration motion margins. In a more advanced setup, registration is employed to determine 4D motion fields and construct a motion model that can be employed during radiation delivery[14].

MRI-guided Radiotherapy.

In conventional external beam radiation therapy, the plan designed in the simulation phase is replicated during each radiation fraction (going sometimes up to 35 fractions). Thus each time the patient has to be positioned on the radiation machine exactly the same as during the CT imaging in the simulation process on which the plan is based. The key advantage of MRI-guided radiation therapy is that by its soft tissue contrast direct visualization of tumor and organ-at-risk are possible at the time of radiation. This allows a radical new workflow and management of positioning uncertainties. Combined with progress in contour propagation and fast dose planning, pre-beam MR imaging allows localization of the target structures and subsequently the fast generation of a tailored plan based on the actual anatomy [15]. In this way, the traditional margins concerning position uncertainty can be eliminated minimizing irradiation of healthy tissue. This resulting lower toxicity can be exploited by switching to more hypo-fractionated schemes reducing costs and patient burden. Several integrated MRI-Linac systems with varying designs and field strengths are currently on the market or under development [16-19].

Moreover, the non-ionizing nature of MRI opens the possibility to perform cineMRI to track moving targets and accumulate retrospectively the deposited dose distribution given the 3D moving anatomy [20,21]. At this moment, real-time tracking can only be done sufficiently fast for 2D imaging. The combination of 2D and a motion model derived from 4D MRI, allows rapid tracking of 3D volumes [23]. Adaptation of other more recent image acceleration techniques such as simultaneous multi-slice [24] or highly undersampled non-cartesian readout combined with parallel imaging and compressed sensing reconstruction [25,26], open the way to perform sufficiently fast 3D motion tracking. A holy grail is the combination of real-time 3D imaging with real-time dose adaptation. For such a strategy low latency machine control and low latency image reconstruction and processing become essential. Very recent advances concerning deep learning based image reconstruction of undersampled k-space data enabling very rapid reconstruction are therefore very welcome [26]. After each radiation delivery, there will the possibility to perform response monitoring by measurement of tumor volume regression or characterization of the tumor tissue response by endogenous MR biomarkers (T1, T2, ADC,..). Longitudinal studies should indicate how any observed changes can be exploited during the course of the therapy to improve local and regional tumor control by escalating dose to non-responding tumor regions.


The use of MRI in radiation oncology allows better tumor definition, facilitates online and real-time adaptive treatments. MR guided Radiotherapy will be a disruptive technology offering the possibility to radically change workflow and eliminate uncertainties. This offers ways to improve therapy, toxicity and patient burden and provide effective treatment for tumors sites that are now treated with invasive therapies.


The research is funded by ZonMw IMDI Programme (project number: 1040030) and the research programme HTSM with project number 15354, which is (partly) financed by the Netherlands Organisation for Scientific Research (NWO).


1. Target definition in prostate, head, and neck. Rasch C, Steenbakkers R, van Herk M. Semin Radiat Oncol. 2005 Jul;15(3):136-4

2. Decreased 3D observer variation with matched CT-MRI, for target delineation in Nasopharynx cancer. Rasch CR, Steenbakkers RJ, Fitton I, Duppen JC, Nowak PJ, Pameijer FA, Eisbruch A, Kaanders JH, Paulsen F, van Herk M. Radiat Oncol. 2010 Mar 15;5:21

3. Role of Prostate MR Imaging in Radiation Oncology. Ménard C, Paulson E, Nyholm T, McLaughlin P, Liney G, Dirix P, van der Heide UA. Radiol Clin North Am. 2018 Mar;56(2):319-325.

4. Dosimetric and workflow evaluation of first commercial synthetic CT software for clinical use in pelvis.Tyagi N, Fontenla S, Zhang J, Cloutier M, Kadbi M, Mechalakos J, Zelefsky M, Deasy J, Hunt M. Phys Med Biol. 2017 Apr 21;62(8):2961-2975

5. A dual model HU conversion from MRI intensity values within and outside of bone segment for MRI-based radiotherapy treatment planning of prostate cancer.Korhonen J, Kapanen M, Keyriläinen J, Seppälä T, Tenhunen M. Med Phys. 2014 Jan;41(1):011704

6. MR-OPERA: A Multicenter/Multivendor Validation of Magnetic Resonance Imaging-Only Prostate Treatment Planning Using Synthetic Computed Tomography Images. Persson E, Gustafsson C, Nordström F, Sohlin M, Gunnlaugsson A, Petruson K, Rintelä N, Hed K, Blomqvist L, Zackrisson B, Nyholm T, Olsson LE, Siversson C, Jonsson J. Int J Radiat Oncol Biol Phys. 2017 Nov 1;99(3):692-700 7. Systematic Review of Synthetic Computed Tomography Generation Methodologies for Use in Magnetic Resonance Imaging-Only Radiation Therapy. Johnstone E, Wyatt JJ, Henry AM, Short SC, Sebag-Montefiore D, Murray L, Kelly CG, McCallum HM, Speight R. Int J Radiat Oncol Biol Phys. 2018 Jan 1;100(1):199-217.

8. A review of substitute CT generation for MRI-only radiation therapy. Edmund JM, Nyholm T. Radiat Oncol. 2017 Jan 26;12(1):28

9. MR-based synthetic CT generation using a deep convolutional neural network method. Han X. Med Phys. 2017 Apr;44(4):1408-1419

10. MR to CT Synthesis Using Unpaired Data. Wolterink J.M., Dinkla A.M., Savenije M.H.F., Seevinck P.R., van den Berg C.A.T., Išgum I. (2017) Deep In: Tsaftaris S., Gooya A., Frangi A., Prince J. (eds) Simulation and Synthesis in Medical Imaging. SASHIMI 2017. Lecture Notes in Computer Science, vol 1055. Also available at arXiv:1708.01155

11. Optimizing 4-dimensional magnetic resonance imaging data sampling for respiratory motion analysis of pancreatic tumors. Stemkens B, Tijssen RH, de Senneville BD, Heerkens HD, van Vulpen M, Lagendijk JJ, van den Berg CA. Int J Radiat Oncol Biol Phys. 2015 Mar 1;91(3):571-8

12. Investigation of undersampling and reconstruction algorithm dependence on respiratory correlated 4D-MRI for online MR-guided radiation therapy. Mickevicius NJ, Paulson ES. Phys Med Biol. 2017 Apr 21;62(8):2910-292

13. T2-Weighted 4D Magnetic Resonance Imaging for Application in Magnetic Resonance-Guided Radiotherapy Treatment Planning.Freedman JN, Collins DJ, Bainbridge H, Rank CM, Nill S, Kachelrieß M, Oelfke U, Leach MO, Wetscherek A. Invest Radiol. 2017 Oct;52(10):563-57

14. Image-driven, model-based 3D abdominal motion estimation for MR-guided radiotherapy. Stemkens B, Tijssen RH, de Senneville BD, Lagendijk JJ, van den Berg CA. Phys Med Biol. 2016 Jul 21;61(14):5335-55

15. A new methodology for inter- and intrafraction plan adaptation for the MR-linac. Kontaxis C, Bol GH, Lagendijk JJ, Raaymakers BW. Phys Med Biol. 2015 Oct 7;60(19):7485-97.

16. MRI/linac integration Lagendijk J J W, Raaymakers B W, Raaijmakers A J E, Overweg J, Brown K J, Kerkhof E M, van der Put R W, Hardemark B, van Vulpen M and van der Heide U A 2008 Radiother. Oncol. 86 25–9 3.

17. The ViewRay system: magnetic resonance-guided and controlled radiotherapy Mutic S and Dempsey J F 2014 Semin. Radiat. Oncol. 24 196–9

18. The rotating biplanar linac-magnetic resonance imaging system Fallone B G 2014 Semin. Radiat. Oncol.24 200–2

19. Australian magnetic resonance imaging-linac program Keall P J, Barton M, Crozier S and Australian MRI-Linac Program, including contributors from InghamInstitute, Illawarra Cancer Care Centre, Liverpool Hospital, Stanford University, Universities ofNewcastle, Queensland, Sydney, Western Sydney, and Wollongong 2014 The Semin. Radiat. Oncol. 24 203–6 6.

20. On-line MR imaging for dose validation of abdominal radiotherapy. Glitzner M, Crijns SP, de Senneville BD, Kontaxis C, Prins FM, Lagendijk JJ, Raaymakers BW. Phys Med Biol. 2015 Nov 21;60(22):8869-83 7

21. Effect of intra-fraction motion on the accumulated dose for free-breathing MR-guided stereotactic body radiation therapy of renal-cell carcinoma. Stemkens B, Glitzner M, Kontaxis C, de Senneville BD, Prins FM, Crijns SPM, Kerkmeijer LGW, Lagendijk JJW, van den Berg CAT, Tijssen RHN. Phys Med Biol. 2017 Sep 1;62(18):7407-7424

22. Image-driven, model-based 3D abdominal motion estimation for MR-guided radiotherapy. Stemkens B, Tijssen RH, de Senneville BD, Lagendijk JJ, van den Berg CA. Phys Med Biol. 2016 Jul 21;61(14):5335-55

23. Simultaneous multislice (SMS) imaging techniques. Barth M, Breuer F, Koopmans PJ, Norris DG, Poser BA. Magn Reson Med. 2016 Jan;75(1):63-81

24. Compressed sensing for body MRI. Feng L, Benkert T, Block KT, Sodickson DK, Otazo R, Chandarana H. J Magn Reson Imaging. 2017 Apr;45(4):966-987

25. XD-GRASP: Golden-angle radial MRI with reconstruction of extra motion-state dimensions using compressed sensing. Feng L, Axel L, Chandarana H, Block KT, Sodickson DK, Otazo R. Magn Reson Med. 2016 Feb;75(2):775-88

26. Image reconstruction by domain-transform manifold learning Bo Zhu, Jeremiah Z. Liu, Stephen Cauley, Bruce R. Rosen, Matthew S. Rosen Nature 555, 487–492

Proc. Intl. Soc. Mag. Reson. Med. 26 (2018)