Megavoltage (MV) photon beams are typically used to treat deep-seated tumours because of the skin-sparing effect. In recent years image-guided radiation therapy (IGRT) on MV photon treatment systems has developed, and online magnetic resonance-guided radiation therapy (MRgRT), offering superior soft tissue contrast imaging, is now available. Such treatments are delivered by machines known as magnetic resonance (MR)-linacs that provide a non-ionizing form of imaging. One of the two commercially available MR-linac systems is the Elekta Unity (Elekta, Stockholm, Sweden) which integrates a 1.5 T magnetic resonance imaging unit and a linac with a flattening-filter-free (FFF) 7 MV photon beam. For these MR-linacs the radiation beam is perpendicular to the magnetic field direction.
The impact of the transverse magnetic field on secondary electron transport is well established in literature [1]–[3]. Dose perturbations include a lateral shift in the dose distribution and asymmetric penumbra [1], [4], [5], a reduction in the depth of maximum dose (dmax) [1], [2], [4]–[7], and the electron return effect (ERE) [2], [4], [8]. There is also evidence that skin-sparing can be enhanced, compared to conventional linac (0 T) treatments, due to the magnetic field sweeping contaminant secondary electrons away from the treatment area [5]–[13]. However, in some situations, such as when the entry and exit surfaces are oblique, this is not the case [8], [9], [11], and the corresponding dose due to the ERE is non-trivial [9], [11].
Previous investigations of near-surface dose in a transverse MR-linac have used dosimeters with varying effective point of measurement (EPOM). These include radiochromic film [6], [14], [15], PTW 31021 Semiflex 3D [16], optically stimulated luminescence dosimeters (OSLDs) [17]–[19], thermoluminescent dosimeters (TLDs) [20], metal-oxide-semiconductor field-effect transistor (MOSFET) [21], gel [22], PTW 60019 microDiamond [6], [14], [16], and a PTW 34045 Advanced Markus chamber [6], [12]. Parallel-plate chambers, such as the Advanced Markus chamber, are commonly used for dose measurements in the build-up region on conventional linacs. For MR-linac dosimetry, the magnetic field influences charge collection in the air-filled sensitive volume (SV) of the ionization chamber [23]. For reference dosimetry, near constant correction factors, specific to the magnetic field, can be applied to ionisation chamber measurements beyond dmax; however, in the build-up region correction factors become depth-dependent [24] due to a loss of charged-particle equilibrium (CPE) conditions. With a variable magnetic field correction factor in the build-up region and a SV thickness in the order of millimetres [25], the ionisation chamber is not an ideal dosimeter to accurately measure skin dose in a transverse MR-linac.
The human skin depth recommended for practical dose estimates is 0.07 mm, and corresponds to the average depth of the basal layer that is responsible for producing new skin cells [26]. To accurately measure skin dose, a dosimeter with a small SV and reproducible water equivalent depth (WED) of 0.07 mm should be considered, to avoid volume averaging that can occur in heterogeneous dose regions. Similarly for small-field dosimetry, the size of the SV should be considerably smaller than the field size as a loss of CPE impacts the detectors readings [27]. With skin dose specified at 0.07 mm, previous near-surface dose investigations are lacking as dosimeters with larger EPOMs were used [6], [14]–[19], [21], [22], [28]. The dose averaged across the SV of an OSLD corresponds to a near-skin water equivalent depth (WED) of 0.16 mm, with the external casing of the OSLD removed [17]. Similarly, film positioned at the surface has a WED of 0.14 mm (i.e. half the thickness of a sheet of EBT-3 film).
Of interest is a study that investigated surface and near-surface dose measurements, at the beam entry and exit, in a 1.5 T transverse MR-linac using nanoDot OSLDs (Landauer, Glenwood, USA) [17]. The study reported surface doses, relative to dmax, of 15.7%, 16.7%, and 18.0%, at the beam entry, and 56.0%, 62.8%, and 63.4% at the beam exit, for 5 × 5 cm2, 10 × 10 cm2, and 22 × 22 cm2 fields, respectively [17]. The authors commented that further experimental investigations were required since previous film data reported entry surface doses of 34.6% and 35.8%, relative to dmax, for field sizes 5 × 5 cm2 and 10 × 10 cm2, respectively [16].
For accurate Monte Carlo simulations of skin dose, high-resolution scoring is recommended [8]. Unlike experimental dosimeters with a fixed SV, Monte Carlo simulations can be scored according to the user’s specifications. Dose calculations at the beam entry and exit for a 6 MV linear accelerator with a 1.5 T transverse magnetic field, have previously been investigated [8]. Using 0.01 mm thick voxels, Monte Carlo calculations determined a sharp increase in dose from 10.0–40.0%, normalised to the dose at dmax, in the first millimetre of the build-up region for a 10 × 10 cm² field. At the beam exit for the same field size, the dose increased from 40.0–55.0%, relative to dmax, as a direct consequence of the ERE [8]. Using high-resolution scoring geometry at the beam entry and exit reveals the extent of the dose gradient that otherwise would be masked using a larger dose voxel size. Likewise, using a dosimeter with a large SV, relative to the steep gradient, would cause volume averaging and inaccurate dosimetry.
We propose using a MOSFET detector, called the MOSkin™, for surface dosimetry in a 1.5 T transverse MR-linac. MOSFET detectors often have an epoxy bubble to protect the SV; however, the MOSkin™ utilises a thin and reproducible polyimide build-up. With a 4.8 × 10− 6 mm3 SV, thickness of 3.0 × 10− 4 mm, reproducible water equivalent build-up equal to 0.07 mm, and overall thickness of 0.4 mm (which includes a thin silicon substrate positioned at the rear of the detector [29]), the MOSkin™ is an ideal candidate for skin and surface dosimetry. Another benefit of MOSFET technology is the real-time readout of measurements, while dosimeters such as film and OSLD require more stringent preparation and read-out protocols. Additionally, the MOSkin™ reportedly experiences no significant readout changes in the presence of a 1.0 T static magnetic field [30]. Previous MR-linac measurements with the MOSkin™ on a 1.0 T inline MR-linac show comparable results to Geant4 simulations, film, and a microDiamond detector, demonstrating the suitability of the detector in an MR environment [31]–[33]. Readers wishing for greater detail on the mechanisms behind general MOSFET and MOSkin™ dosimetry readout are referred elsewhere [34].
The objective of this study is to use the MOSkin™ to experimentally characterise surface and near-surface dose on a 1.5 T transverse MR-linac. The term ‘skin dose’ will herein be synonymous with ‘surface dose’, measured on a water phantom, and measurement points beyond a surface depth of 0.07 mm will be referred to as ‘near-surface dose’. Based on our review of available literature, this would be the first published work of MOSFET’s in a 1.5 T transverse MR-linac and the first published work of experimental skin dose, at a depth of 0.07 mm, in a 1.5 T transverse MR-linac. Experimental measurements with Gafchromic EBT-3 film (Ashland ISP Advanced Materials, NJ, USA) and simulations were also performed to compare to the MOSkin™ measurements.