Essential steps within a RT course were analyzed for phantoms and patients with CFP-T implants.
Phantom and patient data
An academic phantom was created by molding a CFP-T implant (Icotec AG, Altstätten, Switzerland) and numerous fiducial markers into epoxy resin (Figure 1 left). The resulting slab can be added to the commercially available RW3 solid water phantom (PTW, Freiburg, Germany). This offers the flexibility to change the depth of the implant as well as carrying out measurements at multiple distally located planes. For this work the CFP-T implant was positioned at a depth 4.5 cm and measurements were enabled at 0 mm, 2 mm, 10 mm and 50 mm distance distal to the slab with the CFP-T implant in it.
A more realistic and complex situation is represented by an anthropomorphic torso phantom with interchangeable spine inserts (Icotec AG, Altstätten, Switzerland and CIRS, Norfolk, USA)23 (Figure 1 right). In the present study, the spine inserts, representing normal bone structure and CFP-T implants were employed. This phantom consists of multiple coronal slabs, which allows carrying out measurements at different planes. Measurements were performed at two different planes in proximity to the CFP-T implant. These planes are indicated and labelled in Figure 2.
Furthermore, planning-CT data sets were available at our institution for two patients with CFP-T implants, who were irradiated postoperatively with cRT and had both pre- and post-operative MRI examinations.
CT and contouring
CT data sets were acquired for all phantoms as well as for real patients using a Brilliance Big Bore device (Philips, Netherland). Slice thickness was set to 1 and 3 mm for phantoms and patients, respectively.
All artifacts as well as the high-density titanium parts were contoured and the following densities were assigned to the corresponding regions: 1 g/cm3 for normal soft tissue, 1.29 g/cm3 for bone and 4.45 g/cm3 for titanium.
For the anthropomorphic phantom, two clinically realistic planning target volumes (PTVs) were delineated with consideration of preoperative tumor infiltration, namely PTV1 and PTV2. While the PTV1 encompasses the vertebral body, pedicles and both transverse processes, PTV2 is only confined to vertebral body, unilateral pedicle and left transverse process (Figure 2). In addition, the spinal cord was contoured as OAR.
The planning-CTs, pre- and post-operative spinal MRIs from two patient cases were imported into the treatment planning software Precision (Accuray, Sunnyvale, US) and fused together. The delineation of target volumes on the planning-CTs of patients was done according to consensus contouring guidelines for post-operative SBRT24. Briefly, the clinical target volume (CTV) includes the gross residual tumor on postoperative imaging modalities, adjacent anatomical components of the vertebra that are at risk of microscopically spread, areas with preoperative tumor involvement and finally stabilizing implants if the risk of involvement is high24. The adjacent relevant OARs are contoured and spinal cord planning risk volume (PRV) was generated with 2 mm expansion. PTV was created adding 2 mm margin to CTV in all directions and cropped from spinal cord PRV in order to respect the dose constraints.
Treatment planning and dosimetry
To assess the dosimetric accuracy of the treatment planning system, it is essential to verify that the calculated dose corresponds to the actually delivered dose. For dose delivery, the CK system employs three different beam collimator types, namely fixed collimators, Iris collimator and a multileaf collimator (MLC). In this work the Monte Carlo (MC) dose calculation algorithm, which is available for all collimator types, is used. Statistical uncertainties were set at 1% for all calculations. For dosimetric comparisons, plans with different collimator types are considered. Due to the two-dimensional (2D) measurement with a high spatial resolution and in accordance with the measurement possibilities within the utilized phantoms, film dosimetry using Gafchromic EBT3 films (Ashland, US) was chosen as dosimetry tool.
Plans with a single perpendicular beam impinging on the academic phantom were created for the Iris collimator and the MLC separately. For both collimators the maximal field size were used. While the 115x100-mm2 field size of the MLC allows covering the whole CFP-T implant, the 60 mm diameter Iris collimated field was centered on the densest part of the screw.
For the anthropomorphic phantom with inserts, representing CFP-T implants, treatment plans were generated using the Iris collimator and the MLC collimator for both PTVs separately. The four plans were applied and for each delivery, film measurements were carried out at the indicated positions (Figure 2 in Section A).
The same set of four plans was also applied on the anthropomorphic phantom without using the CFP-T implants and the same measurements were carried out as described above.
All irradiated films were digitized using an 10000XL (Epson, JP) scanner, corrected for lateral response artifacts of the scanner25 and compared with the calculated dose distributions within the software FilmQA Pro (Ashland, US). In order to convert the grey values on the film into dose values, calibration stripes with a known applied dose were used to carry out a triple channel calibration25 within FilmQA Pro. The subsequent comparisons were carried out using the green color channel.
A gamma evaluation with a dose difference criterion of 5% of the global maximal dose, a distance to agreement criterion of 1 mm and a 20% dose threshold was carried out in order to compare the calculated with the measured dose distribution.
In order to better visualize the differences between calculated and measured dose, the dose values for the calibrated films were exported from FilmQA Pro and compared to the calculated dose values using python 3.626.
IGRT and delivery
The CK employs a matching algorithm that matches an orthogonal kV image pair (actual position of the patient/phantom) to the reference planning CT (2D-3D match) resulting in a correction for the setup error in 6D (translational and rotational setup errors). Inter- and intra-fraction matching is done automatically; this procedure puts a high demand on high quality CT and planar images.
Although, accurate patient/phantom positioning is per se a requirement for the comparisons in the precedent section, a dedicated scenario was created in order to check if the accuracy of the IGRT controlled delivery of the CK can be maintained in presence of CFP-T implants:
Spherically shaped dose distributions are delivered to a cube containing two orthogonal gafchromic films as part of the standard machine QA of the CK. On both films, the deviation of the circular isodose lines from the intended positions are registered as targeting error using the E2E-software (Accuray, Sunnyvale, US). This cube is now affixed to the anthropomorphic phantom and a plan, delivering a spherical dose distribution on the cube is created. However, for patient positioning the spine-tracking matching was used in the area of the phantom where the hybrid implants are positioned. The resulting targeting error was then compared to the results of the regular standard machine QA.