The RADPOS (Best Medical Canada, Ottawa, ON) consists of a MOSFET radiation dosimeter with an active area of 0.04 mm² and a small cylindrical electromagnetic positioning device (8-mm length, 1.3-mm diameter). To avoid radiation attenuation and disturbance, the positioning sensor and dosimeter are separated by 8 mm. Dose measurement is based on the difference in threshold voltages ΔVth before and after irradiation. In this study, TN-252RD MOSFET detectors with 0.25-mm oxide thickness and a high-sensitivity bias voltage setting were used.

The RADPOS probe is connected to a mobile MOSFET reader to record the threshold dosimeter voltage. The probe is also connected to a 3D Guidance preamplifier and 3D Guidance tracker (Ascension Technology Corporation, Burlington, VT). The 3D Guidance DC magnetic field transmitter, which is connected to the tracker, generates a pulsed 3D magnetic field with well-defined characteristics. The sensor’s response to this magnetic field is monitored by the position tracker and analyzed to determine the x, y, and z coordinates, as well as the probe’s azimuth, elevation, and roll rotation angle.

The MOSFET reader and 3D Guidance tracker are connected directly or wirelessly to a host computer. Special software allows the user to record the ΔVth of the MOSFET and the spatial coordinates of the position sensor manually or automatically at user-defined intervals.

In vivo dosimetry was performed using the therapeutic proton beam line at National Cancer Center Hospital East. The beam line employs a dual-ring double-scattering method for PBT [14] . The thickness of the first scatter and shape of the second scatter were determined by the energy of the proton beams. The maximum diameter of the system’s irradiation field was 200 mmφ. The 190-MeV proton beam was tested daily to ensure that the proton range was within 0.5 mm [15] .

For accurate comparison, MOSFET detector outputs were converted to dose values. Measurements were performed in a PMMA dose calibration phantom [9] [16] . A calibrated 0.6-cc Farmer ionization chamber (FIC, type 30,013; PTW, Freiburg, Germany) and the MOSFET detector were placed along a line perpendicular to the beam axis. The proton energy in the calibration point was 157 MeV, and linear energy transfer (LET) was 0.5 keV/μm. Protons in this point are in the proximal region of the Bragg curve, and the MOSFET detector response has no LET dependence. To obtain the dose calibration factor for the MOSFET detector, the detector and FIC were exposed five times to 200 cGy. The dose calibration factor was determined from the average output.

The dose for the MOSFET detector was obtained as the product of the MOSFET reading (mV), dose calibration factor, and LET correction factor. As the MOSFET response depends strongly on the LET of the proton beam [9] [16] , it was corrected using the highly precise simplified Monte Carlo (SMC) method [17] [18] . This method enables calculation only of dose deposition determined by the experimental depth-dose distribution and lateral displacement of protons due to the multiple scattering effect in materials and the incident angle. When applied to a complex anthropomorphic phantom, the SMC method reproduced the measured dose distribution well, satisfying an accuracy tolerance of 3 mm and 3% in the gamma index analysis [19] . The computation time using the SMC method on graphics processing unit architecture under the computer-unified device architecture platform for the clinical cases is around 1 minute [20] . As a result, the SMC method enabled rapid and accurate dose calculation in even heterogeneities.

To evaluate the usefulness of the RADPOS under more realistic conditions, in-vivo proton dosimetry was performed using the MOSFET detector with an anthropomorphic phantom (The Phantom Laboratory, Salem, CA, USA; Figure 1). The phantom’s head and neck region contains representations of complex inhomogeneous tissues, with bone, soft tissue, and various materials and shapes. The phantom was immobilized with a mold and mask. The transmitter was positioned so that the x, y, and z axes corresponded to the head-foot, left-right, and anterior-posterior axes in sagittal, coronal, and transverse planes, respectively (Figure 1). The RADPOS origin point was defined as the point 200 mm from the transmitter’s center along the x axis. Figure 2 shows the target area for which a treatment plan was designed. We assumed that the target was rectangular and solid, allowing straightforward evaluation of in vivo dosimetry. The isocenter was located at the center of the planning target volume (PTV) region. An irradiation condition for the PTV was determined using a treatment planning system developed in house. This system calculates the dose using the SMC method, and the correction factor for the MOSFET response to account for LET effects. A calculation grid size of 1.172-mm was used for the CT image

(Asteion; Toshiba Medical Systems, Otawara, Japan). The pixel size of the CT image was based on a 0.586 × 0.586 × 3-mm³.

The estimated mean statistical error of the calculated dose in the target volume region was within 1% rms. A gantry angle of 0˚ was arranged on the PBT planning system. For the PTV, a bolus and patient collimator were designed using the planning system. A 190-MeV proton beam, a ridge filter of 60-mm spread-out Bragg peak width, and a range shifter of 2.0-mm thickness were selected. As Kohno et al. reported that the snout must be lo- cated >400 mm from the RADPOS to obtain position accuracy within 1 mm [13] , the snout was positioned at −500 mm along the z axis from the isocenter.
The isocenter was exposed three times to 200 cGy as a point prescription. Evaluation points are marked on Figure 2. Three measurement points were selected to evaluate the dose delivered by protons passing through the inhomogeneities. The dose distribution in the target region was not uniform, but was characterized by steep gradients (up to >5%/mm). Given the presence of complex hot and cold spots around the boundary of inhomogeneity, a precise dose calculation algorithm is desirable in situations involving tissues with significant inhomogeneity.

We measured points A and C (Figure 2) three times and point B twice using the RADPOS. To identify each evaluation point, one RADPOS was fixed at the reference point. Using the coordinates of the reference point and each evaluation point, positional relationships were determined. Although in vivo dosimetry has conventionally been performed by predetermining evaluation points on a CT image, we obtained these points directly by position measurement with the RADPOS. Thus, we expect that high-precision in vivo dosimetry can be performed.