Development and evaluation of a novel MR‐compatible pelvic end‐to‐end phantom

Abstract MR‐only treatment planning and MR‐IGRT leverage MRI's powerful soft tissue contrast for high‐precision radiation therapy. However, anthropomorphic MR‐compatible phantoms are currently limited. This work describes the development and evaluation of a custom‐designed, modular, pelvic end‐to‐end (PETE) MR‐compatible phantom to benchmark MR‐only and MR‐IGRT workflows. For construction considerations, subject data were assessed for phantom/skeletal geometry and internal organ kinematics to simulate average male pelvis anatomy. Various materials for the bone, bladder, and rectum were evaluated for utility within the phantom. Once constructed, PETE underwent CT‐SIM, MR‐Linac, and MR‐SIM imaging to qualitatively assess organ visibility. Scans were acquired with various bladder and rectal volumes to assess component interactions, filling capabilities, and filling reproducibility via volume and centroid differences. PETE simulates average male pelvis anatomy and comprises an acrylic body oval (height/width = 23.0/38.1 cm) and a cast‐mold urethane skeleton, with silicone balloons simulating bladder and rectum, a silicone sponge prostate, and hydrophilic poly(vinyl alcohol) foam to simulate fat/tissue separation between organs. Access ports enable retrofitting the phantom with other inserts including point/film‐based dosimetry options. Acceptable contrast was achievable in CT‐SIM and MR‐Linac images. However, the bladder was challenging to distinguish from background in CT‐SIM. The desired contrast for T1‐weighted and T2‐weighted MR‐SIM (dark and bright bladders, respectively) was achieved. Rectum and bone exhibited no MR signal. Inputted volumes differed by <5 and <10 mL from delineated rectum (CT‐SIM) and bladder (MR‐SIM) volumes. Increasing bladder and rectal volumes induced organ displacements and shape variations. Reproduced volumes differed by <4.5 mL, with centroid displacements <1.4 mm. A point dose measurement with an MR‐compatible ion chamber in an MR‐Linac was within 1.5% of expected. A novel, modular phantom was developed with suitable materials and properties that accurately and reproducibly simulate status changes with multiple dosimetry options. Future work includes integrating more realistic organ models to further expand phantom options.


| INTRODUCTION
Radiation therapy (RT) treatments are traditionally planned using computed-tomography simulation (CT-SIM) images, with interfraction patient setup verification performed using x-ray-based on-board (OB) imaging techniques. Although CT offers strengths such as providing a direct measurement of electron density for dose calculation and geometric image accuracy, it lacks the excellent soft tissue contrast achievable from magnetic resonance imaging (MRI). 1 In the male pelvis, MRI has been shown to improve prostate delineation accuracy, [2][3][4][5] reduce interobserver contouring variability, [3][4][5][6] improve the localization of the prostate apex, [2][3][4][5][6] and increase differentiation of the seminal vesicles from the base of the prostate. 2,4,6 Consequently, treatment planning using MR/CT coregistered images, where delineated soft tissue structures on MR images are transferred onto fused CT images, is often utilized. 2,6 However, uncertainties on the order of 2 mm are introduced due to this coregistration in the pelvis 7,8 that may be reduced with MR-only radiation treatment planning. 9 MR-only treatment planning eliminates this coregistration uncertainty while streamlining clinical efficiency.
Furthermore, the implementation of on-board MR image-guided radiation therapy (IGRT) systems allows for daily image guidance and real-time imaging throughout a treatment fraction, which is ideal for managing and monitoring both interfraction and intrafraction motion, respectively, without additional radiation exposure. 10,11 It has been shown that critical structures and targets within the pelvis are better visualized on MR-IGRT systems than OB-CT, 10 allowing for superior target localization. This, in turn, may lead to improved accuracy of MR-to-MR registration and facilitate dose escalation while also offering potential to reduce treatment margins and toxicity to organs at risk. 11 However, MRI and MR simulation (MR-SIM) acquisitions typically require longer scanning times than CT, which may result in additional errors being introduced because of anatomical or patient movement. 12 In the pelvis, multiple uncertainties may arise as a result of patient motion, changes in anatomical structure (position/deformation) due to intrasession bladder filling, and the introduction of patient-specific distortions due to air in the rectum. 13 Because of this transient nature, it is currently difficult to characterize the geometric and dosimetric uncertainties that may arise in these new workflows. MR-compatible phantoms are currently limited for benchmarking these new workflows. Recently, an anthropomorphic multimodality prostate phantom was developed to compare MR-SIM to CT-SIM. 14 The phantom was custom designed with organs (prostate, rectum, bladder, and femoral heads) that adequately generated signal in MR for end-to-end testing. However, it was unable to simulate organ filling. Niebuhr  Patients were positioned supine and head-first, aligned using external LAP lasers (LAP GmbH Laser Applications, Lüneberg, Germany) to right central axis (CAX), left CAX, and anterior CAX tattoos. They were immobilized with their hands placed on chest, feet banded, and knees immobilized in a black leg sponge. The phantom habitus was determined by evaluating data taken in treatment position at the marked isocenter across the cohort. Measurements of the pelvis width, pelvis height, and sacrum external spacing (distance between the posterior edge of the sacrum and the exterior body surface) were taken using the distance measurement tool in the Eclipse Treatment Planning System (Varian Medical Systems, Palo Alto, CA).
Because of the high visibility of the pelvic bones in the CT-SIM dataset, the pelvic skeleton dimensions across the same 19 subjects were also determined using the treatment planning CT for each patient.
Measurements for the iliac crest width, femoral head width, greater trochanter width, pelvic skeleton depth, and pelvic skeleton height were also obtained using the distance measurement tool in Eclipse.

2.B | Internal organ kinematics
To quantify the impact of systematic bladder filling on organ volume, location, and displacement, 10 immobilized healthy volunteers underwent a~45-min MR-SIM imaging session using the bladder filling protocol outlined in Fig. 1. Subjects voided their bladder prior to consuming 600 mL of water, T2W sequences were acquired immediately with empty bladders and~15 min postconsumption with partially full bladders. An additional 300-600 mL of water were consumed with no subject repositioning, and one to two more time points were acquired with full bladders. A single physician delineated the bladder and rectum at each time point following RTOG 0815 criteria. 16 Temporal datasets were evaluated for the center of mass, shape, and volume of the rectum and bladder with varied filling conditions. To characterize the rectum shape, measurements were obtained as shown in Fig. 2(a) for the width of the anterior, posterior, and middle of the rectum, the length, and the distance from the coccyx to the posterior of the rectum. with distilled water using a 20-mL syringe to increasing volumes up to 350 mL while the silicone balloon was filled to 250 mL and imaged in a 1.0-T Panorama High-Field Open (HFO) system (Philips Medical Systems, Best, the Netherlands). The general shape, longterm stability, and filling capacity of each of these balloon assemblies were assessed. To evaluate the potential of filling the silicone balloon to >300 mL, the filling integrity was assessed by filling it repeatedly to 500 mL and visually inspecting the balloon for mechanical/physical changes.  17 Therefore, both pelvises were imaged using a UTE-Dixon sequence (repetition time (TR)/echo time (TE)/flip F I G . 1. Bladder filling protocol: the patient originally voided their bladder then consumed~600 mL of water. A T2W MR-SIM scan was acquired immediately after drinking and 15 min later. After which, the subject consumed an additional 300-600 mL of water and was imaged again a total of 30 min after the initial bladder void. A 3D modeling is shown for each time point, where a much larger longitudinal than lateral growth of the bladder is observed. angle (α) = 11.5/(0.14/3.45/6.9) ms/25°, voxel size = 0.96 × 0.96 × 1.3 mm 3 , and bandwidth = 994 Hz/pixel) to determine if any measurable signal was detected.

2.D.1 | Scan acquisition
The final phantom build was evaluated across three platforms. CT images were acquired using a Philips Brilliance Big Bore CT-SIM with the following settings: 120 kVp, 244 mAs, 512 × 512 mm 2 field of view (FOV), 0.98 × 0.98 mm 2 resolution, and 2.0-mm slice thickness. To ensure consistent positioning within each imaging modality, the phantom was aligned via external LAP lasers to the external markings (anterior CAX, left CAX, and right CAX) made during CT-SIM. MR images were acquired on the MRIdian Linac (ViewRay Inc., Oakwood Village, OH) using a true fast imaging and steady precession (TrueFISP) sequence. This is a fully refocused (refocusing occurs in all three axes) steady-state sequence with shorter acquisition time, high contrast-to-noise, and signal-to-noise ratios. 18

2.D.2 | Simulating bladder and rectal status changes
Across all three imaging platforms, scans were acquired with a constant bladder volume of 250 mL and varying inputted rectal volumes (30, 60, 90, 120, and 150 mL). In the MR-SIM, additional scans were acquired with a constant rectal volume of 60 mL and varying bladder volumes (90, 150, 250, and 350 mL) ranging from a mostly empty bladder to a mostly full bladder. The fixed rectal and bladder volumes of 60 and 250 mL, respectively, were selected as prostate cancer treatments that are ideally simulated/delivered with a mostly empty rectum and mostly full bladder. 19

2.D.3 | Organ interactions
Contours were generated for the bladder, rectum, and prostate on all MR-SIM T2W datasets in MIM. Centroid displacements due to volume changes were assessed for each contoured organ. Bladder and rectum diameters [left-right (L-R) and anterior-posterior (A-P)] were measured with increased bladder and rectum filling to assess shape changes due to differing filling conditions. Associations between bladder and rectum volumes and resulting centroid displacements were assessed via linear regression.

3.B | Internal organ kinematics
To quantify internal status changes, bladder filling data from 10 human subjects yielded an average bladder volume difference of 87% between empty (81.9 ± 66.9 mL) and full (383.0 ± 346.7 mL) bladders. Figure 1 shows be good for long-term use within the phantom. The silicone balloon was filled to >380 mL with little to no air bubbles present. Therefore, the silicone balloon was chosen for use within the phantom as it addressed the concern of deterioration of polyisoprene over time with exposure to water and oxygen and was able to achieve the desired volumes. Additionally, silicone, in terms of electron density and MR signal is a suitable substitute material for exterior organ shells. 15 The barbed fitting used to tether the balloon to the catheter at each end withstood balloon pressure at maximum volume.

3.C.2 | Bone considerations
Both mold-cast pelvises had anthropomorphic shapes, but the clear urethane pelvis structure did not include femoral heads and was >12% different than desired in pelvic width and depth. The clear pelvis also had a CT number of 60 HU, which is considerably less than the CT number of cancellous bone (262 HU). 20 The blue dyed urethane pelvis included femoral heads, was appropriately sized, and exhibited a CT number closer to that of cancellous bone (213 HU).
Neither pelvis generated a signal in MRI sequences, particularly in the UTE acquisition, and neither approximated cortical bone (1454 HU). 20 Nevertheless, the blue urethane was selected for the initial phantom build due to its higher CT number.   Fig. 4(a). The rectum, urethra, bone, and phantom filling exhibited contrasts as expected for these tissue types. However, the bladder was almost indistinguishable from sur-

3.F | Dosimetry verification
The    | 273 marked and in clinical use for the treatment of prostate cancer. 26,27 While the modular phantom does enable the introduction of 3D printed organs for female anatomy (uterus, cervix, etc.), as other deformable phantoms have, 21 the current pelvic skeleton geometry does not accurately represent the average female anatomy. The pelvis structure was modeled after the male pelvis, which is taller and narrower than the female pelvis, with a 7% difference in width to height ratios. 28 Future extensions of this work include incorporating a penile bulb and reorganization of internal organs so that the rectum interacts more closely with the prostate.
Despite some of the above limitations, a novel, anthropomorphic, and modular pelvic phantom with the ability to simulate bladder and rectal status changes was developed and validated. Potential future clinical applications of this phantom include the benchmarking of MR-to-MR deformable image registration algorithms, evaluation of MR-based adaptive workflows, quantifying distortions in MR images due to susceptibility effects at air-tissue interfaces, and evaluating the electron return effect for MR-IGRT.

| CONCLUSION
A novel end-to-end pelvis phantom has been developed to validate MR-only and MR-IGRT workflows, with the ability to perform both dosimetric and geometric evaluations.