A deep learning method for image-based subject-specific local SAR assessment.

PURPOSE: Local specific absorption rate (SAR) cannot be measured and is usually evaluated by offline numerical simulations using generic body models that of course will differ from the patient's anatomy. An additional safety margin is needed to include this intersubject variability. In this work, we present a deep learning-based method for image-based subject-specific local SAR assessment. We propose to train a convolutional neural network to learn a "surrogate SAR model" to map the relation between subject-specific B1+ maps and the corresponding local SAR. METHOD: Our database of 23 subject-specific models with an 8-transmit channel body array for prostate imaging at 7 T was used to build 5750 training samples. These synthetic complex B1+ maps and local SAR distributions were used to train a conditional generative adversarial network. Extra penalization for local SAR underestimation errors was included in the loss function. In silico and in vivo validation were performed. RESULTS: In silico cross-validation shows a good qualitative and quantitative match between predicted and ground-truth local SAR distributions. The peak local SAR estimation error distribution shows a mean overestimation error of 15% with 13% probability of underestimation. The higher accuracy of the proposed method allows the use of less conservative safety factors compared with standard procedures. In vivo validation shows that the method is applicable with realistic measurement data with impressively good qualitative and quantitative agreement to simulations. CONCLUSION: The proposed deep learning method allows online image-based subject-specific local SAR assessment. It greatly reduces the uncertainty in current state-of-the-art SAR assessment methods, reducing the time in the examination protocol by almost 25%.

Understanding the physical relations governing the noise navigator.

PURPOSE: The noise navigator is a passive way to detect physiological motion occurring in a patient through thermal noise modulations measured by standard clinical radiofrequency receive coils. The aim is to gain a deeper understanding of the potential and applications of physiologically induced thermal noise modulations. METHODS: Numerical electromagnetic simulations and MR measurements were performed to investigate the relative contribution of tissue displacement versus modulation of the dielectric lung properties over the respiratory cycle, the impact of coil diameter and position with respect to the body. Furthermore, the spatial motion sensitivity of specific noise covariance matrix elements of a receive array was investigated. RESULTS: The influence of dielectric lung property variations on the noise variance is negligible compared to tissue displacement. Coil size affected the thermal noise variance modulation, but the location of the coil with respect to the body had a larger impact. The modulation depth of a 15 cm diameter stationary coil approximately 3 cm away from the chest (i.e. radiotherapy setup) was 39.7% compared to 4.2% for a coil of the same size on the chest, moving along with respiratory motion. A combination of particular noise covariance matrix elements creates a specific spatial sensitivity for motion. CONCLUSIONS: The insight gained on the physical relations governing the noise navigator will allow for optimized use and development of new applications. An optimized combination of elements from the noise covariance matrix offer new ways of performing, e.g. motion tracking.

Proton MRS of cervical cancer at 7 T.

The differentiation grade of cervical cancer is histologically assessed by examining biopsies or surgical specimens. MRS is a highly sensitive technique that images tissue metabolism and can be used to increase the specificity of tissue characterization in a non-invasive manner. We aim to explore the feasibility of using in vivo 1 H-MRS at 7 T in women with cervical cancer to study tissue fatty acid composition. 10 women with histologically proven Stage IB1-IIB cervical cancer were scanned with a whole-body 7 T MR system with a multi-transmit system and an internal receive only monopole antenna. A STEAM sequence was used to obtain 1 H-MRS data. Fatty acid resonances were fitted with Lorentzian curves and the 2.1 ppm/1.3 ppm ratios were calculated. 1 H-MRS data showed fatty acid signals resonating at 2.1 ppm, 1.9 ppm, 1.5 ppm, 1.3 ppm and 0.9 ppm. Mean 2.1/1.3 ppm ratios were 0.019 ± 0.01, 0.021 ± 0.006, 0.12 ± 0.089 and 0.39 ± 0.27 for normal, Grade I, Grade II and Grade III groups respectively. Poorly differentiated tumor tissue (Grade III) showed elevated fatty acid ratios when compared with the well differentiated tumor (Grade I) or normal tissue. 1 H-MRS in cervical cancer at 7 T is feasible and individual fatty acid signals were detected. In addition, poorly differentiated tumors show more fatty acid unsaturation. The 2.1 ppm/1.3 ppm ratio has potential for tumor characterization in a non-invasive manner for uterine cervical cancer.

Validating subject-specific RF and thermal simulations in the calf muscle using MR-based temperature measurements.

PURPOSE: Ongoing discussions occur to translate the safety restrictions on MR scanners from specific absorption rate (SAR) to thermal dose. Therefore, this research focuses on the accuracy of thermal simulations in human subjects during an MR exam, which is fundamental information in that debate. METHODS: Radiofrequency (RF) heating experiments were performed on the calves of 13 healthy subjects using a dedicated transmit-receive coil while monitoring the temperature with proton resonance frequency shift (PRFS) thermometry. Subject-specific models and one generic model were used for electromagnetic and thermal simulations using Pennes' bioheat equation, with the blood equilibration constant equaling zero. The simulations were subsequently compared with the experimental results. RESULTS: The mean B1+ equaled 15 µT in the center slice of all volunteers, and 95% of the voxels had errors smaller than 2.8 µT between the simulation and measurement. The intersubject variation in RF power to achieve the required B1+ was 11%. The resulting intersubject variation in median temperature rise was 14%. Thermal simulations underestimated the median temperature increase on average, with 34% in subject-specific models and 28% in the generic model. CONCLUSIONS: Although thermal measures are directly coupled to tissue damage and therefore suitable for RF safety assessment, insecurities in the applied thermal modeling limit their estimation accuracy. Magn Reson Med 77:1691-1700, 2017. © 2016 International Society for Magnetic Resonance in Medicine.

MRI-based transfer function determination for the assessment of implant safety.

PURPOSE: We introduce a new MR-based method to determine the transfer function (TF) for radiofrequency (RF) safety assessment of active implantable medical devices. Transfer functions are implant-specific measures that relate the incident tangential electric field on an (elongated) implant to a scattered electric field at its tip. The proposed method allows for TF determination with a high spatial resolution in relatively fast measurements without requiring dedicated bench setups from MRI images. THEORY AND METHODS: The principle of reciprocity is used in conjunction with the potential to measure currents with MRI to determine TF. Low-flip angle 3D dual gradient echo MRI data are acquired with an implant as transceive antenna, which requires minimal hardware adaptations. The implant-specific TF is determined from the acquired MRI data, with two different postprocessing methods for comparison. RESULTS: TFs of linear and helical implants can be determined accurately (with a Pearson correlation coefficient R ≥ 0.7 between measurements and simulations, and a difference in field at the tip ΔEtip ≤ 19%) from relatively quick (t < 20 minutes) MRI acquisitions with (several) millimeter spatial resolution. CONCLUSION: Transfer function determination with MRI for RF safety assessment of implantable medical devices is possible. The proposed MR-based method allows for TF determination in more realistic exposure scenarios and solid media. Magn Reson Med 78:2449-2459, 2017. © 2017 International Society for Magnetic Resonance in Medicine.

Thermal noise variance of a receive radiofrequency coil as a respiratory motion sensor.

PURPOSE: Development of a passive respiratory motion sensor based on the noise variance of the receive coil array. METHODS: Respiratory motion alters the body resistance. The noise variance of an RF coil depends on the body resistance and, thus, is also modulated by respiration. For the noise variance monitoring, the noise samples were acquired without and with MR signal excitation on clinical 1.5/3 T MR scanners. The performance of the noise sensor was compared with the respiratory bellow and with the diaphragm displacement visible on MR images. Several breathing patterns were tested. RESULTS: The noise variance demonstrated a periodic, temporal modulation that was synchronized with the respiratory bellow signal. The modulation depth of the noise variance resulting from the respiration varied between the channels of the array and depended on the channel's location with respect to the body. The noise sensor combined with MR acquisition was able to detect the respiratory motion for every k-space read-out line. CONCLUSION: Within clinical MR systems, the respiratory motion can be detected by the noise in receive array. The noise sensor does not require careful positioning unlike the bellow, any additional hardware, and/or MR acquisition. Magn Reson Med 77:221-228, 2017. © 2016 Wiley Periodicals, Inc.

Boosting the SNR by adding a receive-only endorectal monopole to an external antenna array for high-resolution, T2 -weighted imaging of early-stage cervical cancer with 7-T MRI.

The aim of this study was to investigate the signal-to-noise ratio (SNR) gain in early-stage cervical cancer at ultrahigh-field MRI (e.g. 7 T) using a combination of multiple external antennas and a single endorectal antenna. In particular, we used an endorectal monopole antenna to increase the SNR in cervical magnetic resonance imaging (MRI). This should allow high-resolution, T2 -weighted imaging and magnetic resonance spectroscopy (MRS) for metabolic staging, which could facilitate the local tumor status assessment. In a prospective feasibility study, five healthy female volunteers and six patients with histologically proven stage IB1-IIB cervical cancer were scanned at 7 T. We used seven external fractionated dipole antennas for transmit-receive (transceive) and an endorectally placed monopole antenna for reception only. A region of interest, containing both normal cervix and tumor tissue, was selected for the SNR measurement. Separated signal and noise measurements were obtained in the region of the cervix for each element and in the near field of the monopole antenna (radius < 30 mm) to calculate the SNR gain of the endorectal antenna in each patient. We obtained high-resolution, T2 -weighted images with a voxel size of 0.7 × 0.8 × 3.0 mm3 . In four cases with optimal placement of the endorectal antenna (verified on the T2 -weighted images), a mean gain of 2.2 in SNR was obtained at the overall cervix and tumor tissue area. Within a radius of 30 mm from the monopole antenna, a mean SNR gain of 3.7 was achieved in the four optimal cases. Overlap between the two different regions of the SNR calculations was around 24%. We have demonstrated that the use of an endorectal monopole antenna substantially increases the SNR of 7-T MRI at the cervical anatomy. Combined with the intrinsically high SNR of ultrahigh-field MRI, this gain may be employed to obtain metabolic information using MRS and to enhance spatial resolutions to assess tumor invasion.

Dipole antennas for ultrahigh-field body imaging: a comparison with loop coils.

Although the potential of dipole antennas for ultrahigh-field (UHF) MRI is largely recognized, they are still relatively unknown to the larger part of the MRI community. This article intends to provide electromagnetic insight into the general operating principles of dipole antennas by numerical simulations. The major part focuses on a comparison study of dipole antennas and loop coils at frequencies of 128, 298 and 400 MHz. This study shows that dipole antennas are only efficient radiofrequency (RF) coils in the presence of a dielectric and/or conducting load. In addition, the conservative electric fields (E-fields) at the ends of a dipole are negligible in comparison with the induced E-fields in the center. Like loop coils, long dipole antennas perform better than short dipoles for deeply located imaging targets and vice versa. When the optimal element is chosen for each depth, loop coils have higher B1 (+) efficiency for shallow depths, whereas dipole antennas have higher B1 (+) efficiency for large depths. The cross-over point depth decreases with increasing frequency: 11.6, 6.2 and 5.0 cm for 128, 298 and 400 MHz, respectively. For single elements, loop coils demonstrate a better B1 (+) /√SARmax ratio for any target depth and any frequency. However, one example study shows that, in an array setup with loop coil overlap for decoupling, this relationship is not straightforward. The overlapping loop coils may generate increased specific absorption rate (SAR) levels under the overlapping parts of the loops, depending on the drive phase settings. Copyright © 2015 John Wiley & Sons, Ltd.

Improved RF performance of travelling wave MR with a high permittivity dielectric lining of the bore.

Application of travelling wave MR to human body imaging is restricted by the limited peak power of the available RF amplifiers. Nevertheless, travelling wave MR advantages like a large field of view excitation and distant location of transmit elements would be desirable for whole body MRI. In this work, improvement of the B1+ efficiency of travelling wave MR is demonstrated. High permittivity dielectric lining placed next to the scanner bore wall effectively reduces attenuation of the travelling wave in the longitudinal direction and at the same time directs the radial power flow toward the load. First, this is shown with an analytical model of a metallic cylindrical waveguide with the dielectric lining next to the wall and loaded with a cylindrical phantom. Simulations and experiments also reveal an increase of B1+ efficiency in the center of the bore for travelling wave MR with a dielectric lining. Phantom experiments show up to a 2-fold gain in B1+ with the dielectric lining. This corresponds to a 4-fold increase in power efficiency of travelling wave MR. In vivo experiments demonstrate an 8-fold signal-to-noise ratio gain with the dielectric lining. Overall, it is shown that dielectric lining is a constructive method to improve efficacy of travelling wave MR.

Introduction of ultra-high-field MR brain imaging in infants: vital parameters, temperature and comfort.

Background: Brain MRI in infants at ultra-high-field scanners might improve diagnostic quality, but safety should be evaluated first. In our previous study, we reported simulated specific absorption rates and acoustic noise data at 7 Tesla. Methods: In this study, we included twenty infants between term-equivalent age and three months of age. The infants were scanned on a 7 Tesla MRI directly after their clinically indicated 3 Tesla brain MRI scan. Vital parameters, temperature, and comfort were monitored throughout the process. Brain temperature was estimated during the MRI scans using proton MR spectroscopy. Results: We found no significant differences in vital parameters, temperature, and comfort during and after 7 Tesla MRI scans, compared to 3 Tesla MRI scans. Conclusions: These data confirm our hypothesis that scanning infants at 7 Tesla MRI appears to be safe and we identified no additional risks from scanning at 3 Tesla MRI.

Design of a radiative surface coil array element at 7 T: the single-side adapted dipole antenna.

Ultra high field MR imaging (≥7 T) of deeply located targets in the body is facing some radiofrequency-field related challenges: interference patterns, reduced penetration depth, and higher Specific Absorbtion Ratio (SAR) levels. These can be alleviated by redesigning the elements of the transmit or transceive array. This is because at these high excitation field (B(1) ) frequencies, conventional array element designs may have become suboptimal. In this work, an alternative design approach is presented, regarding coil array elements as antennas. Following this approach, the Poynting vector of the element should be oriented towards the imaging target region. The single-side adapted dipole antenna is a novel design that fulfills this requirement. The performance of this design as a transmit coil array element has been characterized by comparison with three other, more conventional designs using finite difference time domain (FDTD) simulations and B +1 measurements on a phantom. Results show that the B +1 level at the deeper regions is higher while maintaining relatively low SAR levels. Also, the B +1 field distribution is more symmetrical and more uniform, promising better image homogeneity. Eight radiative antennas have been combined into a belt-like surface array for prostate imaging. T(1) -weighted (T1W) and T(2) -weighted (T2W) volunteer images are presented along with B +1 measurements to demonstrate the improved efficiency.

Fast contour propagation for MR-guided prostate radiotherapy using convolutional neural networks.

PURPOSE: To quickly and automatically propagate organ contours from pretreatment to fraction images in magnetic resonance (MR)-guided prostate external-beam radiotherapy. METHODS: Five prostate cancer patients underwent 20 fractions of image-guided external-beam radiotherapy on a 1.5 T MR-Linac system. For each patient, a pretreatment T2-weighted three-dimensional (3D) MR imaging (MRI) scan was used to delineate the clinical target volume (CTV) contours. The same scan was repeated during each fraction, with the CTV contour being manually adapted if necessary. A convolutional neural network (CNN) was trained for combined image registration and contour propagation. The network estimated the propagated contour and a deformation field between the two input images. The training set consisted of a synthetically generated ground truth of randomly deformed images and prostate segmentations. We performed a leave-one-out cross-validation on the five patients and propagated the prostate segmentations from the pretreatment to the fraction scans. Three variants of the CNN, aimed at investigating supervision based on optimizing segmentation overlap, optimizing the registration, and a combination of the two were compared to results of the open-source deformable registration software package Elastix. RESULTS: The neural networks trained on segmentation overlap or the combined objective achieved significantly better Hausdorff distances between predicted and ground truth contours than Elastix, at the much faster registration speed of 0.5 s. The CNN variant trained to optimize both the prostate overlap and deformation field, and the variant trained to only maximize the prostate overlap, produced the best propagation results. CONCLUSIONS: A CNN trained on maximizing prostate overlap and minimizing registration errors provides a fast and accurate method for deformable contour propagation for prostate MR-guided radiotherapy.

Introduction of Ultra-High-Field MR Imaging in Infants: Preparations and Feasibility.

BACKGROUND AND PURPOSE: Cerebral MR imaging in infants is usually performed with a field strength of up to 3T. In adults, a growing number of studies have shown added diagnostic value of 7T MR imaging. 7T MR imaging might be of additional value in infants with unexplained seizures, for example. The aim of this study was to investigate the feasibility of 7T MR imaging in infants. We provide information about the safety preparations and show the first MR images of infants at 7T. MATERIALS AND METHODS: Specific absorption rate levels during 7T were simulated in Sim4life using infant and adult models. A newly developed acoustic hood was used to guarantee hearing protection. Acoustic noise damping of this hood was measured and compared with the 3T Nordell hood and no hood. In this prospective pilot study, clinically stable infants, between term-equivalent age and the corrected age of 3 months, underwent 7T MR imaging immediately after their standard 3T MR imaging. The 7T scan protocols were developed and optimized while scanning this cohort. RESULTS: Global and peak specific absorption rate levels in the infant model in the centered position and 50-mm feet direction did not exceed the levels in the adult model. Hearing protection was guaranteed with the new hood. Twelve infants were scanned. No MR imaging-related adverse events occurred. It was feasible to obtain good-quality imaging at 7T for MRA, MRV, SWI, single-shot T2WI, and MR spectroscopy. T1WI had lower quality at 7T. CONCLUSIONS: 7T MR imaging is feasible in infants, and good-quality scans could be obtained.

Dosimetry for the MRI accelerator: the impact of a magnetic field on the response of a Farmer NE2571 ionization chamber.

The UMC Utrecht is constructing a 1.5 T MRI scanner integrated with a linear accelerator (Lagendijk et al 2008 Radiother. Oncol. 86 25-9). The goal of this device is to facilitate soft-tissue contrast based image-guided radiotherapy, in order to escalate the dose to the tumour while sparing surrounding normal tissues. Dosimetry for the MRI accelerator has to be performed in the presence of a magnetic field. This paper investigates the feasibility of using a Farmer NE2571 ionization chamber for absolute dosimetry. The impact of the mcagnetic field on the response of this ionization chamber has been measured and simulated using GEANT4 Monte Carlo simulations. Two orientations of the ionization chamber with respect to the incident beam and the magnetic field which are feasible in the MRI accelerator configuration are taken into account. Measurements are performed using a laboratory magnet ranging from 0 to 1.2 T. In the simulations a range from 0 to 2 T is used. For both orientations, the measurements and simulations agreed within the uncertainty of the measurements and simulations. In conclusion, the response of the ionization chamber as a function of the magnetic field is understood and can be simulated using GEANT4 Monte Carlo simulations.

Integrating a 1.5 T MRI scanner with a 6 MV accelerator: proof of concept.

At the UMC Utrecht, The Netherlands, we have constructed a prototype MRI accelerator. The prototype is a modified 6 MV Elekta (Crawley, UK) accelerator next to a modified 1.5 T Philips Achieva (Best, The Netherlands) MRI system. From the initial design onwards, modifications to both systems were aimed to yield simultaneous and unhampered operation of the MRI and the accelerator. Indeed, the simultaneous operation is shown by performing diagnostic quality 1.5 T MRI with the radiation beam on. No degradation of the performance of either system was found. The integrated 1.5 T MRI system and radiotherapy accelerator allow simultaneous irradiation and MR imaging. The full diagnostic imaging capacities of the MRI can be used; dedicated sequences for MRI-guided radiotherapy treatments will be developed. This proof of concept opens the door towards a clinical prototype to start testing MRI-guided radiation therapy (MRIgRT) in the clinic.

Magnetic-field-induced dose effects in MR-guided radiotherapy systems: dependence on the magnetic field strength.

Several institutes are currently working on the development of a radiotherapy treatment system with online MR imaging (MRI) modality. The main difference between their designs is the magnetic field strength of the MRI system. While we have chosen a 1.5 Tesla (T) magnetic field strength, the Cross Cancer Institute in Edmonton will be using a 0.2 T MRI scanner and the company Viewray aims to use 0.3 T. The magnetic field strength will affect the severity of magnetic field dose effects, such as the electron return effect (ERE): considerable dose increase at tissue air boundaries due to returning electrons. This paper has investigated how the ERE dose increase depends on the magnetic field strength. Therefore, four situations where the ERE occurs have been simulated: ERE at the distal side of the beam, the lateral ERE, ERE in cylindrical air cavities and ERE in the lungs. The magnetic field comparison values were 0.2, 0.75, 1.5 and 3 T. Results show that, in general, magnetic field dose effects are reduced at lower magnetic field strengths. At the distal side, the ERE dose increase is largest for B = 0.75 T and depends on the irradiation field size for B = 0.2 T. The lateral ERE is strongest for B = 3 T but shows no effect for B = 0.2 T. Around cylindrical air cavities, dose inhomogeneities disappear if the radius of the cavity becomes small relative to the in-air radius of the secondary electron trajectories. At larger cavities (r > 1 cm), dose inhomogeneities exist for all magnetic field strengths. In water-lung-water phantoms, the ERE dose increase takes place at the water-lung transition and the dose decreases at the lung-water transition, but these effects are minimal for B = 0.2 T. These results will contribute to evaluating the trade-off between magnetic field dose effects and image quality of MR-guided radiotherapy systems.

Feasibility of MRI guided proton therapy: magnetic field dose effects.

Many methods exist to improve treatment outcome in radiotherapy. Two of these are image-guided radiotherapy (IGRT) and proton therapy. IGRT aims at a more precise delivery of the radiation, while proton therapy is able to achieve more conformal dose distributions. In order to maximally exploit the sharp dose gradients from proton therapy it has to be combined with soft-tissue based IGRT. MRI-guided photon therapy (currently under development) offers unequalled soft-tissue contrast and real-time image guidance. A hybrid MRI proton therapy system would combine these advantages with the advantageous dose steering capacity of proton therapy. This paper addresses a first technical feasibility issue of this concept, namely the impact of a 0.5 T magnetic field on the dose distribution from a 90 MeV proton beam. In contrast to photon therapy, for MR-guided proton therapy the impact of the magnetic field on the dose distribution is very small. At tissue-air interfaces no effect of the magnetic field on the dose distribution can be detected. This is due to the low-energy of the secondary electrons released by the heavy protons.

Experimental verification of magnetic field dose effects for the MRI-accelerator.

The MRI-linear accelerator system, currently being developed, is designed such that the patient is irradiated in the presence of a magnetic field. This influences the dose distribution due to the Lorentz force working on the secondary electrons. Simulations have shown that the following dose effects occur: the build-up distance is reduced, the lateral profile becomes asymmetric in the direction orthogonal to the magnetic field and at tissue-air interfaces the dose increases due to returning electrons. In this work, GafChromic film measurements were performed in the presence of a magnetic field to experimentally quantify these dose effects. Depth-dose curves were measured in a PMMA-air-PMMA phantom and the lateral profiles were measured in a homogeneous PMMA phantom with the photon beam protruding over the edges of the phantom. The measurement results confirmed the magnetic field dose effects that were predicted by simulations. This enabled us to verify Geant4 Monte Carlo simulations of these MRI-linac specific dose effects: the relative agreement for the depth-dose curves between measurements and simulations was within 2.2%/1.8 mm. The relative agreement for the lateral profiles was 2.3%/1.7 mm. Overall, the magnetic field dose effects that are expected for irradiation with the MRI-linac can be modelled using Geant4 Monte Carlo simulations within measurement accuracy.

Integrating a MRI scanner with a 6 MV radiotherapy accelerator: impact of the surface orientation on the entrance and exit dose due to the transverse magnetic field.

At the UMC Utrecht, in collaboration with Elekta and Philips Research Hamburg, we are developing a radiotherapy accelerator with integrated MRI functionality. The radiation dose will be delivered in the presence of a lateral 1.5 T field. Although the photon beam is not affected by the magnetic field, the actual dose deposition is done by a cascade of secondary electrons and these electrons are affected by the Lorentz force. The magnetic field causes a reduced build-up distance: because the trajectory of the electrons between collisions is curved, the entrance depth in tissue decreases. Also, at tissue-air interfaces an increased dose occurs due to the so-called electron return effect (ERE): electrons leaving tissue will describe a circular path in air and re-enter the tissue yielding a local dose increase. In this paper the impact of a 1.5 T magnetic field on both the build-up distance and the dose increase due to the ERE will be investigated as a function of the angle between the surface and the incident beam. Monte Carlo simulations demonstrate that in the presence of a 1.5 T magnetic field, the surface dose, the build-up distance and the exit dose depend more heavily on the surface orientation than in the case without magnetic field. This is caused by the asymmetrical pointspread kernel in the presence of 1.5 T and the directional behaviour of the re-entering electrons. Simulations on geometrical phantoms show that ERE dose increase at air cavities can be avoided using opposing beams, also when the air-tissue boundary is not perpendicular to the beam. For the more general case in patient anatomies, more problems may arise. Future work will address the possibilities and limitations of opposing beams in combination with IMRT in a magnetic field.