http://www.diva-portal.org
This is the published version of a paper published in International Journal of Radiation
Oncology, Biology, Physics.
Citation for the original published paper (version of record):
Adjeiwaah, M., Bylund, M., Lundman, J A., Thellenberg Karlsson, C., Jonsson, J H. et
al. (2018)
Quantifying the effect of 3T MRI residual system and patient-induced susceptibility
distortions on radiotherapy treatment planning for prostate cancer.
International Journal of Radiation Oncology, Biology, Physics, 100(2): .317-324
https://doi.org/10.1016/j.ijrobp.2017.10.021
Access to the published version may require subscription.
N.B. When citing this work, cite the original published paper.
Permanent link to this version:
Physics Contribution
Quantifying the Effect of 3T Magnetic Resonance
Imaging Residual System Distortions and
Patient-Induced Susceptibility Distortions on
Radiation Therapy Treatment Planning for
Prostate Cancer
Mary Adjeiwaah, MSc,
*
Mikael Bylund, MSc,
*
Josef A. Lundman, MSc,
*
Camilla Thellenberg Karlsson, MD, PhD,
*
Joakim H. Jonsson, MSc, PhD,
*
and Tufve Nyholm, MSc, PhD
*
,y*Department of Radiation Sciences, Umea˚ University, Umea˚, Sweden; andyMedical Radiation Physics, Department of Immunology, Genetics and Pathology, Uppsala University, Uppsala, Sweden
Received May 26, 2017, and in revised form Aug 28, 2017. Accepted for publication Oct 12, 2017.
Summary
Despite the many advantages of using magnetic resonance imaging (MRI) in radiation therapy treatment planning, there are concerns about geometric distortions. Displacement fields from system distortions and patient-induced susceptibil-ities were used to distort 17 prostate patient CT images. Dose plans were optimized on the distorted CT and the plan parameters transferred to the CT images to calculate
Purpose: To investigate the effect of magnetic resonance system- and patient-induced susceptibility distortions from a 3T scanner on dose distributions for prostate cancers. Methods and Materials: Combined displacement fields from the residual system and patient-induced susceptibility distortions were used to distort 17 prostate patient CT images. VMAT dose plans were initially optimized on distorted CT images and the plan parameters transferred to the original patient CT images to calculate a new dose distribution.
Results: Maximum residual mean distortions of 3.19 mm at a radial distance of 25 cm and maximum mean patient-induced susceptibility shifts of 5.8 mm were found using the lowest bandwidth of 122 Hz per pixel. There was a dose difference of<0.5% be-tween distorted and undistorted treatment plans. The 90% confidence intervals of the mean difference between the dCT and CT treatment plans were all within an equiva-lence interval of (0.5, 0.5) for all investigated plan quality measures.
Conclusions: Patient-induced susceptibility distortions at high field strengths in closed bore magnetic resonance scanners are larger than residual system distortions after us-ing vendor-supplied 3-dimensional correction for the delineated regions studied.
Reprint requests to: Mary Adjeiwaah, MSc, Department of Radiation Sciences, Umea˚ University, SE-901 87 Umea˚, Sweden. Tel: (þ46) 720-47-4157; E-mail:mary.adjeiwaah@umu.se
The corresponding author’s PhD studies were supported by the Schlumberger Foundation Faculty for the Future.
Conflict of interest: none.
AcknowledgmentsdThe authors thank Johan Brynolfsson of Spec-tronics Medicals; Magnus G. Karlsson from University Hospital of Umea˚ for support during the treatment planning; Natalia Arteaga Marrero from Umea˚ University for proofreading the manuscript; and Mattias Berglund for support on the MICE Toolkit.
Int J Radiation Oncol Biol Phys, Vol. 100, No. 2, pp. 317e324, 2018
0360-3016/Ó 2017 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).
https://doi.org/10.1016/j.ijrobp.2017.10.021
biology physics
a new dose distribution. We found the dosimetric impact of distortions on treatment plans to be small.
However, errors in dose due to disturbed patient outline and shifts caused by patient-induced susceptibility effects are below 0.5%. Ó 2017 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/4.0/).
Introduction
The use of data from magnetic resonance imaging (MRI) for radiation therapy (RT) planning is increasing. This is primarily due to MRI’s superior soft tissue contrast compared with computed tomography (CT), leading to improved delineations. For prostate cancers, it results in better coverage of the target volume and rectal sparing(1, 2). However, the full integration of a magnetic resonance (MR)-only RT has been hindered by geometric distortions in MRI and lack of electron density information for dose calculations as well as MR-based digitally reconstructed radiographs (3). Recently the introduction of commercial products for MR-only treatment planning of the prostate is paving the way for clinical implementation of MR-only RT
(4). From a dosimetric point of view, excellent results have been achieved when using data from MRI instead of CT for dose calculations, with errors consistently below 2%
(5-7). Fiducial gold markers visible on images from MRI have been reliably used for patient positioning (8), and techniques for the reconstruction of MRI-based digitally reconstructed radiographs have been presented (9, 10). There are, however, concerns about the potential problems associated with geometric distortions from the MRI system and the patient to be imaged (11, 12). System-related distortions come from inhomogeneities in the static mag-netic field (B0 field) and nonlinearities in the gradient.
These distortions increase in magnitude with increasing radial distance from the isocenter of the MRI scanner. There are vendor-supplied correction algorithms to mini-mize gradient field nonlinearity distortions. However, re-sidual system-related distortions persist. Because of the different magnetic properties of the patient’s anatomic tissues, the patient also introduces distortions due to magnetic susceptibility effects and chemical shift effects. These are larger on high-field MR systems but can be reduced by utilizing high-bandwidth (BW) sequences. However, increasing the BW results in reduced signal-to-noise ratio, a potential image quality concern in RT treatment planning for organ sites with low signals (13).
There are several studies that have looked at the po-tential effects of MR geometric distortions on either MR-CT or MR-only RT treatment planning on anatomic sites such as the breast, brain, and prostate (5, 14-20). Particu-larly, the study on the breast concluded that even at high BWs the dosimetric impact of system- and patient-induced distortion could be clinically unacceptable. For the prostate, a recent study by Gustafsson et al (20) on the effect of system-related distortions on treatment plans reported no clinical dose difference when using 3-dimensional (3D)
correction in combination with a high-BW sequence. In addition, Sun et al(19)reported 0.01% differences between MR- and CT-based dose calculations using a pelvic phantom for distortion quantification. However, these in-vestigations on the prostate did not account for patient-induced susceptibility effects. In this study, we aimed to quantify the effect of residual system- and patient-induced susceptibility distortions on treatment plans in an MR-only workflow for different imaging BWs and gradient readout directions (GRDs).
Methods and Materials
Phantom measurement of system-related
distortions
A positron emission tomography MR/GE Signa 3T (Gen-eral Electric, Waukesha, WI) scanner was used in this study. We used a commercially manufactured phantom (Spectronic Medical, Helsingborg, Sweden) with a signal-producing volume of 350.7 470 450.7 mm3to measure residual system distortions. Susceptibility-induced distor-tions from the phantom have been reported to be negligible
(20). Axial MR images of the phantom were acquired using our clinical sequence for prostate cancer examinations. This is a T2-weighted fast spin-echo sequence with a repetition time of 1500 milliseconds, an echo time of 99 milliseconds, a 50-cm field of view (FoV), 2-mm slice thickness without spacing, a 512 512 image matrix, and 227 slices. Mul-tiple scans at BWs of 122, 244, and 488 Hz per pixel with anterior/posterior (A/P) and right/left (R/L) GRDs were obtained with the vendor-supplied 3D correction. The geometric center of the phantom was placed at the isocenter of the scanner for all measurements.
Simulation of patient-induced susceptibility
distortions
A method proposed by Lundman et al (21)that estimates susceptibility values from the bulk conversion of CT Hounsfield units was used to simulate patient-induced susceptibility distortions from 17 prostate cancer patients. The simulation process has been integrated in the Medical Interactive Creative Environment (MICE Toolkit), an image data analysis tool as part of the Swedish gentle RT project (http://gentleradiotherapy.se). All patient CT images had an in-plane resolution in a range of (0.97 0.97) to (1.097 1.097) mm2and an image matrix of 512 512, whereas the resolution for the MR images of the phantom
Adjeiwaah et al. International Journal of Radiation Oncology Biology Physics
was (0.98 0.98) mm2. We used the same BW and GRD arrangements from the phantom measurements in the sus-ceptibility simulations. However, distortions in millimeters were obtained by multiplying each patient’s displacement map by his or her individual image pixel sizes(21).
Accounting for marker-induced susceptibility distortions was beyond the scope of this study. Therefore, we removed streak artifacts around the intraprostatic fiducial markers on the CT images as well as marker-induced susceptibility effects on the distorted CT images by masking these areas with soft tissue during the simulation process, thereby excluding regions of unrealistic susceptibility-induced distortion data.
Creating the distorted CT
We added displacement vector fields from the phantom-measured residual system distortions and simulated patient-induced susceptibility distortions. This was used to deform the patient CT images using Bspline interpo-lation. A deform function in the MICE Toolkit, based on the Insight Segmentation and Registration Toolkit’s WarpImageFilter, warped the CT images with the supplied displacement field to produce distorted CT images (dCT). Each vector in the displacement field represented the distance between a geometric point in the CT image space and its corresponding point in the dCT data. Contrary to previous studies that did not take patient-induced sus-ceptibility effects into account, the delineated RT struc-tures were also distorted instead of directly copied from the CT to the dCT images(20, 22). Thus, for each of the 17 patient CTs, 6 distorted CT datasets were obtained on the basis of the different BW and gradient readout di-rections, as follows: (1) dCT datasets at BW of 122 Hz per pixel in the R/L gradient direction; (2) dCT datasets at BW of 122 Hz per pixel in the A/P gradient direction; (3) dCT datasets at BW of 244 Hz per pixel in the R/L gradient direction; (4) dCT datasets at BW of 244 Hz per pixel in the A/P gradient direction; (5) dCT datasets at BW of 488 Hz per pixel in the R/L gradient direction; and
(6) dCT datasets at BW of 488 Hz per pixel in the A/P gradient direction.
To eliminate unnecessary deformations as a result of distortion data from the peripherals of the phantom as well as simulate clinical scan conditions, the displacement fields were translated and resampled to match each patient’s co-ordinates.Figure 1shows the study workflow.
Treatment planning
Oncentra External Beam version 4.5 (Elekta, Stockholm, Sweden) was used for all optimizations and dose calcula-tions. Dual arc volumetric modulated arc therapy (VMAT)eoptimized treatment plans were initially gener-ated for all dCT images. We utilized 2 15-MV fields with a start angle of 178, an arc length of 356, and a 2 gantry spacing. The optimized field arrangements were transferred to the CT data (ie, the undistorted anatomy), and a “true” dose distribution was calculated. To isolate the effect of distortions on treatment plans, no further optimization was done during the calculation of the true dose distribution on the corresponding CT datasets. All patients had a pre-scribed absolute dose of 77 Gy in 35 fractions normalized to the planning target volume (PTV) and an integrated dose boost of 84 Gy for 6 patients with visible dominant lesions. The treatment isocenter was placed at the center of the PTV. A clinical target volume (CTV) to PTV margin of 7 mm, based on our clinical protocol, was used. In all, 204 (6 17 optimized dCT and 6 17 patient CT) treatment plans were calculated using a pencil beam algorithm with inhomogeneity correction. Table 1 shows the volumetric modulated arc therapy optimization objectives.
The clinical doseevolume acceptance priority criteria in order of descending importance used in this study were as follows:
Minimum dose to the CTV (CTV Dmin) should be95%
of the prescribed dose (73 Gy).
The 95% isodose (V95%) should cover at least 95% of the
PTV. Residual system related displacement vector field Combined displacement vector field Simulated patient-induced susceptibility displacement vector field dCT VMAT optimized plan on dCT CT CT doseplan dCT dose plan Difference
Fig. 1. Workflow. Schematic diagram of the distorted computed tomography (dCT) creation from the displacement vector fields and the treatment planning process. Abbreviation: VMATZ volumetric modulated arc therapy.
Less than 15% of the outlined rectal volume should receive more than 90% of the prescribed dose (V90%).
The “near minimum dose” to PTV (D98%) should be
90% of the prescribed dose (70 Gy).
Statistical analysis: equivalence testing
All plans were compared on the basis of the minimum dose to the CTV, V95%, and D98%of the PTV, as well as the V90%
of the rectum using the pairwise 1 2-sided equivalence test (TOST-P) (23). The null (H0) and alternative (H1)
hy-potheses are:
H0: ðy1 y2Þ <
d
or ðy1 y2Þ >d
H1:
d
< ðy1 y2Þ <d
ð1Þwhere (
d
,d
) is the equivalent interval (the clinically acceptable difference) between the 2 treatment plans, and y1and y2are the mean of the measured parameters. UsingTOST-P, equivalence was established at an alpha level (
a
) of 0.5% if the (1 2a
)% confidence interval for the dif-ference between the mean measured values of dCT and CT plans were found within (d
,d
). We chosed
values of (0.5, 0.5) Gy for dose differences and (0.5, 0.5)% for percentage volume differences. All TOST-P calculations were done in NCSS 11.0.9 (NCSS, Kaysville, UT).Results
Distortion quantification
The mean residual system distortions within a radial dis-tance of up to 25 cm (50-cm FoV) in the R/L readout were 3.16, 2.22, and 2.02 mm for BWs of 122, 244, and 488 Hz per pixel. The corresponding distortions in the A/P readout were 3.19, 2.52, and 2.08 mm, respectively. The phantom extended farther in the A/P than in the R/L direction, leading to the slightly higher distortions recorded in the A/P gradient readout. Susceptibility effects resulted in maximum shifts of 5.8, 2.9, and 1.5 mm at the BWs used. The extent and location of distortions from the scanner and the patient’s own anatomy are shown inFigure 2. This was obtained when distortion maps from patient-induced sus-ceptibility and system-related effects were overlaid on a representative CT image.
An estimation of geometric shifts occurring at some anatomic sites because of distortions was obtained when displacement fields from the residual system distortions and patient-induced susceptibility effects were masked with the contours of some delineated structures. The plot displayed inFigure 3shows the shifts occurring within the contours of the bladder, the femoral heads, PTV, and the rectum at our investigated BWs. These are the average shifts for all studied subjects.
Dosimetric evaluation
For the PTV, all plans were acceptable: the 95% isodose (PTV V95%) covered mean volumes of 99.9% 0.1% on
dCT and 99.8% 0.1% on CT, with a maximum mean difference of 0.1% 0.22% across all patients. The mean dose at the isocenter for all patients was 76.5 0.3 Gy and 76.4 0.3 Gy for dCT and CT plans, respectively. The maximum dose difference decreased with increasing BW in the R/L direction, whereas it remained almost the same in
Table 1 Optimizing conditions set on the planning target volume (PTV) and organs at risk
Structure Objective Weighting factor
PTV 77 Uniform 77 Gy 3000 PTV 84 Uniform 84 Gy 1000 Rectum Maximum 73.2 Gy 300 Rectum Maximum 40 Gy to 30% volume 1
Small bowel Maximum 53.2 Gy 1000
Urethra Maximum 75 Gy 300
External Surrounding dose falloff: 77 to 28 Gy in 1 cm
200
0
3 mm
a
b
Fig. 2. Distortion map overlaid on a representative computed tomography slice. Coronal computed tomography image of a representative patient overlaid with distortion field from (a) patient-induced susceptibility effects and (b) system-related distortions after 3-dimensional correction at a bandwidth of 244 Hz per pixel. High distortion shifts are observed within the patient owing to induced susceptibility (a) compared with that from the scanner (b). Gradient readout is in the right to left direction.
Adjeiwaah et al. International Journal of Radiation Oncology Biology Physics
the A/P direction. The mean percentage dose differences at the isocenter between distorted and undistorted treatment plans are shown inTable 2.
The dose distribution in the transverse plane for a representative patient at the isocenter between dCT and CT treatment plans, as well as the dose difference map, is given in Figure 4. Table 3 summarizes the findings for the different BWs and readout directions based on clinical dose volume criteria. The mean difference and the 90% CI for the PTV covered by the 95% isodose decreased with increasing BW in the R/L GRD. None of the calculated difference in mean values extended beyond the equivalence intervals used.
Discussions
This study investigated the dosimetric impact of geometric distortions on MRI-based prostate treatment plans. Previous
investigations have shown that residual system distortions have a small impact on prostate dosimetry(19, 20, 24). Our results show that patient-induced susceptibility distortions are larger than the system-specific distortions for the
Table 2 Mean percentage dose difference at the isocenter averaged over all patients among treatment plans on the dis-torted computed tomography and computed tomography im-ages at different BWs as well as GRDs in the R/L and A/P directions
BW (Hz per pixel) GRD Range (%) Mean (%) SD (%)
122 R/L (0.03, 0.43) 0.19 0.12 A/P (0.01, 0.32) 0.17 0.09 244 R/L (0.00, 0.41) 0.12 0.12 A/P (0.00, 0.32) 0.17 0.11 488 R/L (0.00, 0.35) 0.16 0.13 A/P (0.00, 0.33) 0.12 0.11
Abbreviations: A/P Z anterior/posterior; BW Z bandwidth; GRDZ gradient readout; R/L Z right/left; SD Z standard deviation.
Patient-induced related System ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 1 2 3 Bladder Femor alHead_L Femor alHead_R PTV Rectum Distor tion Shifts (mm) ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 1 2 3 Bladder Fe mor alHead_L Fe mor alHead_R PTV Rectum Distor tion Shifts (mm) ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 1 2 3 Bladder Femor alHead_L Femor alHead_R PTV Rectum Distor tion Shifts (mm)
a
b
c
Fig. 3. Shifts due to system- and patient-induced susceptibility distortions. Estimated shifts at the contours of the bladder, the femoral heads, planning target volume (PTV), and rectum at bandwidths of (a) 122, (b) 244, and (c) 488 Hz per pixel in the right to left gradient readout. The boxes show the average values of the median (centerline) and 25% and 75% percentiles (bottom and top of boxes) for all patients, in millimeters. The points at the top and bottom of each box represent the range in absolute shifts. LZ left; R Z right.
investigated BWs and FoVs. It might therefore not be sufficient to only refer to phantom-based distortion mea-surements when claiming adequate geometric accuracy for RT purposes. However, using a sufficient BW sequence the dosimetric impact is still within the 3% to 5% overall ac-curacy required for dose delivery and therefore may be considered acceptable for RT purposes(25).
Among the structures investigated, the femoral heads were the most sensitive to system-related distortions, whereas the prostate was the least affected, as shown in
Figure 3. Because the magnitude of system distortions in-creases with growing radial distance from the scanner’s isocenter, treatment plans for target volumes situated in the peripheral parts of the FoV, such as breast carcinomas, thoracic wall, and abdominal soft tissue sarcomas, may be more affected by system-related distortions(7, 14). For the same reason, distortions to the patient external contours might also be larger for larger patients.
Both system-specific and susceptibility-induced dis-tortions are reduced with increasing BW. The results show that increasing the BW was relevant in reducing patient-induced susceptibility shifts compared with system dis-tortions by a factor of 2, as can be seen inFigure 3. The reduction in system-specific distortions is probably related to the reduced impact of inherent B0inhomogeneity and
the small susceptibility effects caused by the phantom
(20). A benefit of using simulations and not direct mea-surements to estimate the patient-induced susceptibility effect is that contributions from inherent B0
in-homogeneity are not counted twice when adding the sys-tem- and patient-specific distortions. It has been previously described that susceptibility effects generate field perturbations of up to 10 ppm at 3T (26), corre-sponding to calculated shifts of up to 6 pixels, which are in line with our results. On the basis of the results of this study, we recommend a BW of 440 Hz per pixel for 3T scanners, corresponding to 440 Hz/mm using 1 1-mm2
a
b
c
0 84 Gy 0 84 Gy -1.5 1.5 Gy
Fig. 4. Dose distribution comparison. Axial slices at the isocenter showing dose distributions from a computed tomography (CT) (a) and distorted CT (b) plan for a representative patient at a bandwidth of 244 Hz per pixel in the right to left gradient readout direction. The difference map between the CT and distorted CT dose distributions is also shown (c).
Table 3 Equivalence test results between distorted and un-distorted dose distributions based on the doseevolume criteria within an equivalence interval of (0.5, 0.5) Gy for difference in dose and (0.5, 0.5)% for difference in percentage volumes ataZ 0.05 Bandwidth (Hz per pixel) GRD Difference in means SD 90% Confidence interval CTV Dmin 73 Gy 122 R/L 0.21 0.25 Gy (0.11, 0.32) 244 0.07 0.12 Gy (0.02, 0.12) 488 0.11 0.14 Gy (0.05, 0.17) 122 A/P 0.13 0.12 Gy (0.08, 0.18) 244 0.12 0.11 Gy (0.08, 0.17) 488 0.07 0.17 Gy (0.00, 0.14) PTV V95% 95% 122 R/L 0.10% 0.22% (0.01, 0.19) 244 0.04% 0.10% (0.00, 0.09) 488 0.01% 0.03% (0.00, 0.03) 122 A/P 0.01% 0.02% (0.00, 0.02) 244 0.09% 0.19% (0.01, 0.17) 488 0.03% 0.04% (0.01, 0.05) Rectum V90% 15% 122 R/L 0.04% 0.22% (0.13, 0.05) 244 0.02% 0.40% (0.15, 0.18) 488 0.07% 0.19% (0.01, 0.15) 122 A/P 0.01% 0.28% (0.13, 0.11) 244 0.01% 0.24% (0.09, 0.11) 488 0.03% 0.20% (0.12, 0.05) PTV D98% 70 Gy 122 R/L 0.21 0.16 Gy (0.14, 0.28) 244 0.15 0.20 Gy (0.06, 0.24) 488 0.12 0.12 Gy (0.07, 0.17) 122 A/P 0.22 0.15 Gy (0.16, 0.29) 244 0.14 0.23 Gy (0.05, 0.24) 488 0.13 0.14 Gy (0.08, 0.19) Abbreviations as inTable 2.
The P value for all calculations was .0001. These values were calculated for all voxels within the clinical target volume (CTV), planning target volume (PTV), and rectum.
Adjeiwaah et al. International Journal of Radiation Oncology Biology Physics
voxels in plane for prostate examinations in an RT context. At this BW, the water/fat shift is 1 voxel, and the distor-tions due to susceptibility effects will approximate to the same size as the system-specific distortions. Increasing BW beyond the recommended value would further decrease distortions, but the reduction in signal-to-noise ratio would limit its clinical applicability. It is worth noting that keeping the same BW per pixel while doubling the pixel size will also double the distortions.
The use of vendor-supplied distortion correction al-gorithms only reduces system-related distortions but not patient-induced distortions. Because our results show that patient-induced susceptibility effects were larger, utilizing correction algorithms to reduce these distor-tions in an MR-only workflow could be useful. Sug-gested methods such as B0mapping(27-29)and reversed
read-out gradient polarity (13) are currently to our knowledge not provided for routine clinical use from the major vendors and were not modeled in the present study. Intraprostatic fiducial markers introduce distor-tions in MR images due to the different magnetic prop-erties between the markers and soft tissues. These distortions are, however, local and result in deviations of <1 mm (8).
We found a relative dose difference at the PTV of <0.5%. This is within the findings from earlier studies by Petersch et al(24)and Mah et al(30), who reported a dose difference of <1% between corrected MR and CT plans. Gustafsson et al (20) used a similar method to generate distorted CT using displacement maps from system-related distortions and found a mean percentage dose difference of 0.02%. This lower value compared with our results might be because patient-induced susceptibility distortions were not quantified.
In an MR-only workflow, structures will be drawn on the basis of MR images. Therefore, by distorting the RT structures in this study, any shifts in the images caused by MR distortions would also be reflected in the delin-eated contours. Chen et al (22) reported large discrep-ancies for some patients when the CT contours were directly transferred to the MR images for treatment planning.
The patient within the scanner introduces distortions due to susceptibility effects as well as chemical shift effect. Chemical shift effects due to the difference in resonance frequencies between water and fat molecules (440 Hz at 3T) were not included in this study. On the basis of the BWs used in this study, calculated shifts of up to 3.6, 1.8, and 0.9 pixels could be attributed to chemical shift distortions. We did not take into account potential susceptibility-induced distortions introduced by the phantom, though it has been reported to be<0.5 mm
(20). Patients with hip implants were also not included in this study. However, our simulation method can be used to estimate susceptibility-induced effects if their geom-etry and susceptibility values are known.
Conclusion
By combining measured residual system-specific distor-tions after 3D correction and simulated patient-induced distortions for spin-echoebased sequences, we found minimal effect of distortions on prostate cancer treatment plans in an MR-only workflow. The magnitude of susceptibility-related distortions in this study was larger than system-specific distortions, even with a BW of 488 Hz per pixel in a 3T scanner. With the increasing use of MR scanners with high field strengths, incorporating methods to correct patient-induced susceptibility effects could mitigate their effect on treatment plans in an MR-only RT.
References
1. Debois M, Oyen R, Maes F, et al. The contribution of magnetic reso-nance imaging to the three-dimensional treatment planning of localized prostate cancer. Int J Radiat Oncol Biol Phys 1999;45:857-865. 2. Milosevic M, Voruganti S, Blend R, et al. Magnetic resonance imaging
(MRI) for localization of the prostatic apex: Comparison to computed tomography (CT) and urethrography. Radiother Oncol 1998;47:277-284. 3. Pasquier D, Betrouni N, Vermandel M, et al. MRI alone simulation for conformal radiation therapy of prostate cancer: Technical aspects. Conf Proc IEEE Eng Med Biol Soc 2006;1:160-163.
4. Ko¨hler M, Vaara T, Van Grootel M, et al. MR-only Simulation for Radiotherapy Planning Treatment Planning, White Pap. Philips MRCAT Prostate Dose Calc. Using Only MRI Data. Netherlands: Amsterdam; 2015;1-16. Available at: http://incenter.medical.philips. com/doclib/enc/fetch/2000/4504/577242/577251/587787/White_Paper_ MR-only_sim_LR.pdf%3Fnodeid%3D11147198%26vernum%3D-2. Accessed January 30, 2017.
5. Chen L, Price RA, Wang L, et al. MRI-based treatment planning for radiotherapy: Dosimetric verification for prostate IMRT. Int J Radiat Oncol Biol Phys 2004;60:636-647.
6. Jonsson JH, Karlsson MG, Karlsson M, et al. Treatment planning using MRI data: An analysis of the dose calculation accuracy for different treatment regions. Radiat Oncol 2010;5:62.
7. Prott FJ, Haverkamp U, Eich H, et al. Effect of distortions and asymmetry in MR images on radiotherapeutic treatment planning. Int J Cancer 2000;90:46-50.
8. Jonsson JH, Garpebring A, Karlsson MG, et al. Internal fiducial markers and susceptibility effects in MRI: Simulation and measurement of spatial accuracy. Int J Radiat Oncol Biol Phys 2012;82:1612-1618. 9. Chen L, Nguyen TB, Jones E, et al. Magnetic resonance-based
treatment planning for prostate intensity-modulated radiotherapy: Creation of digitally reconstructed radiographs. Int J Radiat Oncol Biol Phys 2007;68:903-911.
10. Ramsey CR, Oliver AL. Magnetic resonance imaging based digitally reconstructed radiographs, virtual simulation, and three-dimensional treatment planning for brain neoplasms. Med Phys 1998;25:1928. 11. Bakker CJG, Moerland MA, Bhagwandien R, et al. Analysis of
machine-dependent and object-induced geometric distortion in 2DFT MR imaging. Magn Reson Imaging 1992;10:597-608.
12. Chang H, Fitzpatrick JM. A technique for accurate magnetic reso-nance imaging in the presence of field inhomogeneities. IEE Trans Med Imaging 1992;11:319-329.
13. Moerland MA, Beersma R, Bhagwandien R, et al. Analysis and correction of geometric distortions in 1.5 T magnetic resonance im-ages for use in radiotherapy treatment planning. Phys Med Biol 1995; 40:1651-1664.
14. Walker A, Metcalfe P, Liney G, et al. MRI geometric distortion: Impact on tangential whole-breast IMRT. J Appl Clin Med Phys 2016;17:7-19. 15. Prabhakar R, Haresh KP, Ganesh T, et al. Comparison of computed tomography and magnetic resonance based target volume in brain tumors. J Cancer Res Ther 2015;3:121-123.
16. Kristensen BH, Laursen FJ, Løgager V, et al. Dosimetric and geo-metric evaluation of an open low-field magnetic resonance simulator for radiotherapy treatment planning of brain tumours. Radiother Oncol 2008;87:100-109.
17. Lee YK, Bollet M, Charles-Edwards G, et al. Radiotherapy treatment planning of prostate cancer using magnetic resonance imaging alone. Radiother Oncol 2003;66:203-216.
18. Wang D, Strugnell W, Cowin G, et al. Geometric distortion in clinical MRI systems. Part I: Evaluation using a 3D phantom. Magn Reson Imaging 2004;22:1211-1221.
19. Sun J, Dowling J, Pichler P, et al. MRI simulation: End-to-end testing for prostate radiation therapy using geometric pelvic MRI phantoms. Phys Med Biol 2015;60:3097-3109.
20. Gustafsson C, Nordstro¨m F, Persson E, et al. Assessment of dosimetric impact of system specific geometric distortion in an MRI only based radiotherapy workflow for prostate. Phys Med Biol 2017;62:2976-2989. 21. Lundman JA, Bylund M, Garpebring A, et al. Patient-induced sus-ceptibility effects simulation in magnetic resonance imaging. Phys Imaging Radiat Oncol 2017;1:41-45.
22. Chen L, Price RA Jr, Nguyen TB, et al. Dosimetric evaluation of MRI-based treatment planning for prostate cancer. Phys Med Biol 2004;49: 5157-5170.
23. Walker E, Nowacki AS. Understanding equivalence and noninferiority testing. J Gen Intern Med 2011;26:192-196.
24. Petersch B, Bogner J, Fransson A, et al. Effects of geometric distortion in 0.2T MRI on radiotherapy treatment planning of prostate cancer. Radiother Oncol 2004;71:55-64.
25. Thwaites D. Accuracy required and achievable in radiotherapy dosimetry: Have modern technology and techniques changed our views? J Phys Conf Ser 2013;444:12006.
26. Schenck JF. The role of magnetic susceptibility in magnetic resonance imaging: MRI magnetic compatibility of the first and second kinds. Med Phys 1996;23:815-850.
27. Jezzard P, Balaban RS. Correction for geometric distortion in echo planar images from B0 field variations. Magn Reson Med 1995;34:65-73.
28. Cusack R, Brett M, Osswald K. An evaluation of the use of magnetic field maps to undistort echo-planar images. Neuroimage 2003;142:127-142. 29. Matakos A, Balter J, Cao Y. Estimation of geometrically undistorted
B0 inhomogeneity maps. Phys Med Biol 2014;59:4945-4959. 30. Mah D, Steckner M, Hanlon A, et al. MRI simulation: Effect of
gradient distortions on three-dimensional prostate cancer plans. Int J Radiat Oncol 2002;53:757-765.
Adjeiwaah et al. International Journal of Radiation Oncology Biology Physics
