CT and CBCT imaging of gel dosimeters: a review of the state of the art

This review covers the main developments in x-ray CT-based polymer gel dosimetry (PGD) and more recently on-board cone-beam CT (CBCT) polymer gel dosimetry. Fundamental work as well as clinically-based applications are described and advantages and disadvantages of approaches highlighted. The body of work present within the literature points towards an overall feasibility of CT PGD and increasingly CBCT PGD as viable tools for the verification of highly complex, high dose radiotherapy treatments.


Introduction
Currently there exist three primary modalities for imaging polymer gel dosimeters (PGD): magnetic resonance imaging (MRI), optical computed tomography (optical CT) and x-ray CT [1,3].Additional modalities have been shown as proof-of-concept, including ultrasound imaging (US) and spectroscopic techniques [4,5], however, these have not been widely pursued in recent years.Of the three main imaging modalities used to image PGD, x-ray CT (herein referred to simply as CT) was the last to be introduced [3] and perhaps remains the least investigated.A primary reason for the low uptake of CT PGD is the overall low dose resolution of the system [6,7].Nonetheless, CT PGD has numerous practical advantages, including but not limited to: the access to CT imagers in the form of CT simulators within radiotherapy clinics; ease of use; integration within clinical workflow; and minimal additional operating cost.With these advantages in mind, this article provides a high-level review of the main areas of concentration within CT PGD research.Some effort is made to concentrate the review on more recent work, e.g.development of a deformable CT PGD system, as reviews of earlier works can be found in previous overviews [8][9][10].Recently, preliminary work has been undertaken to establish a novel PGD technique using on-board cone-beam CT (CBCT) for imaging and this review will also introduce this new area of PGD research.The review is organized as follows: summary of main technique developments in CT PGD (section 2); an introduction to CBCT PGD (section 3); applications of CT and CBCT PGD (section 4); followed by summary and conclusions (section 5).

CT PGD Technique Development
A primary focus of researchers in CT PGD has been on improving the dose resolution of the overall system.This effort has been multi-pronged and has covered areas of gel development (section 2.1), image acquisition optimization (section 2.2), and understanding the effects of CT dose on image quality and accuracy (section 2.3).Further investigations have involved characterizing the overall accuracy and

5-10mM
While it has been shown that the formulation of table 1 does not provide a linear dose response [19], the system does produce an overall ~17-20H change in CT number over 20Gy (i.e.global average response of ~0.8-1H/Gy) [19].It has been further shown that the formulation of table 1 does exhibit some dose rate dependence [20], although the severity of the variability is highly application dependent.Additional studies looking at alternate formulations have been conducted with the goal of improving dosimeter sensitivity, however, a more sensitive dosimeter than that described in table 1 has not, to the best of our knowledge, been reported nor widely adopted [21,22].Additionally, some work has been done in establishing the effects of reconstruction technique on subsequent CT PGD image quality [28].However, while promising, this avenue of investigation can require dedicated vendor agreements and support and has not been widely pursued.
2.2.2.Image filtering.Image filtering of averaged, background-subtracted CT images of PGDs has been widely investigated [22,[29][30][31].In summary, image filtering has a profound effect on image signal to noise ratio (SNR) and has thus been established as in important tool in CT PGD due to the low contrast inherent in this system.While a multitude of filters can be employed [30], the adaptive mean (AM) filter has been shown to be effective at SNR image improvements (over unfiltered images) while maintaining edge preservation [29].Figure 1 illustrates the potential efficacy of adaptive mean filtering on CT PGD images.
In addition to basic statistical noise filtering, some work has been done to establish a more dedicated, remnant artefact removal technique for the removal of mild baseline inhomogeneities within processed PGD images [32].The technique works via the application of a Savitsky-Golay filter of span s (width of filter), degree d (polynomial order) and number of iterations/applications n.The filter runs a first iteration that establishes the baseline signal.Artefacts above this baseline are removed, thus creating a new image to which the second iteration of the filter is applied.Figure 2 illustrates the efficacy of AM filtering and AM filtering + remnant artefact removal for irradiated PGD profiles.At present, AM filtering + remnant artefact removal is the standard for performing CT PGD.

2.3.CT Dose considerations.
A central question from the beginning of CT PGD was what, if any, was the impact of CT radiation dose delivered due to the process of imaging the dosimeter?This question has been studied by Baxter et al [33] in some detail.In summary of their work, it was established that under optimized imaging protocols (table 2) with 16-25 image averages a ~<0.2 H change in CT number can be expected for acylamide based PGDs imaged while still radiosensitive.Given NIPAM-based PGDs (table 1) exhibit similar or lower sensitivity to the acrylamide formulation therein studied, we would expect a similar impact of CT dose on the NIPAM dosimeters.However, the full extent of this postulation has not been investigated.Interestingly, Hill et al [14,34] turned the question of CT dose around and used PGDs to measure CTDI within typical CT imaging applications and over a range of CT scanning parameters.Figure 3 illustrates CTDI measurements using PGDs and MRI imaging.

2.4.Accuracy and precision of x-ray CT PGD
Maynard et al [19] undertook an exhaustive evaluation of the accuracy and precision achievable with NIPAM-based (table 1) CT PGD.Results suggest that the average relative dose error between Monte Carlo dose calculations and CT-based gel measurements is, generally, <2%, depending on the calibration method employed, with poorly calibrated dosimeters more in line with <3% error, as shown in figure 4. Furthermore, 3%/3mm gamma pass-rates for 2D gel images as compared with Monte Carlocalculated images were >97% for the cases shown in figure 4. It is clear that with careful experimental parameters, CT PGD can do well in accurately capturing radiotherapy-based radiation dose.

2.5.System modelling
Hayati et al [35] have undertaken Monte Carlo-based modelling of an x-ray CT PGD system, with the aim of creating a model that can be used for investigating and tuning CT and PGD parameters in silico for further optimization of the system.Figure 6

CBCT Gel Dosimetry
On-board imaging using CBCT has become a modern standard in linac based radiation therapy.The improved quality of CBCT images achievable with newer systems, along with the now relative ubiquity of these systems, has enabled the intriguing possibility of performing polymer gel dosimetry utilising CBCT as the imaging modality.CBCT image quality is reduced compared to traditional CT due to increased radiation scatter inherent in a cone-beam geometry, however there are unique and potentially highly advantageous features of CBCT imaging of gel dosimeters that make this pursuit interesting: (i) The ability to irradiate and image in the same location.With CBCT integrated directly with linear accelerators, you can irradiate and image a gel without having to move the dosimeter.The key advantage here is the removal of re-positioning uncertainties incurred in moving the dosimeter from linear accelerator couch to CT or MRI bed for example in more standard gel imaging approaches.(ii) The ability for close to real-time imaging of delivered dose distributions.With imaging integrated with the delivery system, the potential for near real-time imaging of dosimeters is apparent with CBCT.
Adamson et al [36] introduced CBCT PGD in their 2019 work where they outlined an on-line near realtime treatment verification workflow and demonstrated the technique's potential for characterizing high dose, steep dose-gradient deliveries.Dosimeter read-out by CBCT was performed directly following irradiation, 3 post-irradiation CBCTs were acquired for image averaging, and the total time between a pre-irradiation CBCT positioning scan and the final post-irradiation CBCT was 28min.This initial work was followed up by Jirasek et al [37] in 2020.In this work they showed that CBCT imaging 20-30 min post-irradiation allows for >90% polymer yield in NIPAM dosimeters and yields substantially improved image contrast when compared to imaging immediately following irradiation (see Figure 6).They employed a CBCT protocol optimized based on guiding principles in Hilts et al [27] which consisted of: 100kV, 1035mAs, head and DynamicGainFluoro kV modes, Titaniuum kV filter, full fan bowtie filter, 15 frames per second, smoothing filter and strong ring suppression.Up to 15 images averages were utilized to further improve image contrast to noise ratios.They achieved a 93% 3%,3mm Gamma pass rate for a 3 static field calibration measurement using 10 image averages and saw only marginal improvement at 15 averages.By comparison, using the standard CT PGD methodology achieved a 97% Gamma pass rate.As CBCT technology is still improving (e.g.inclusion of iterative reconstruction algorithms) the future may see image quality getting closer to CT and thus the potential for CBCT PGD an effective 3D dosimetry tool is very promising.[39], Maynard etal demonstrated the potential of CT PGD for deformable dose verification [40,41], as shown in figure 8. Deformable PGD is complicated by the fact that a deformable container housing the PGD must remain oxygen impermeable in order to maintain the radiation sensitivity of the dosimeter.However, the work of Liu et al and Maynard et al highlight the fact that pliable (deformable), oxygenimpermeable containers can be used in concert with PGD to create a highly flexible system for measuring organ-like dose deformation in a radiotherapy context.Deformable PGD remains one of the only ways in which deformed doses can be experimentally measured in a clinical setting and, hence, an exciting area for further investigation.

4.3.SRS Spatial Localization Accuracy
Verification of radiosurgery and other stereotactic treatment techniques offers excellent potential for CT based PGD given the high doses and steep dose-gradients that characterize these treatments.The first application of CT PGD was demonstrated in an early work that illustrated the potential of CT PGD to accurately localize a high dose irradiation achieved with linac, cone-based radiosurgery [11] Recently, Adamson et al [36] have illustrated this potential for CBCT PGD, as shown in figure 9.In this proof of principle study, CBCT PGD was able to verify the dose localization for a 6 target, multi-focal, 4-arc volumetric arc therapy (VMAT) stereotactic radiosurgery (SRS) treatment delivery to within 2mm of planned location The ability to measure and subsequently read-out the dose from a treatment delivery without moving the dosimeter, a feature inherent to CBCT, is demonstrated here and represents a clear advantage over other PGD techniques for cases such as this where an extremely high degree of spatial precision is required.

Summary and conclusions
It is clear that much work has been done in CT PGD since its first demonstration by Hilts et al [3].Work on PGD formulation optimization, imaging parameter optimization, CT dose considerations, and Monte Carlo-based system modelling have all pushed CT PGD into the realm of being feasible as a truly 3D dosimetry method for, typically, high dose treatment verifications.The accuracy and precision of CT PGD is now sufficiently high such that many current problems involving high dose in 3D dose verification can be realistically tackled with the technique.In particular, the introduction of CBCT PGD offers new and exciting possibilities in onboard, close to real-time 3D dose verification using polymer gels.

Figure 3 .
Figure 3. (a) demonstration of PGDs ability to measure CT dose within diagnostic CT scanners.(b) MRI profile through irradiated region of (a) for a nominal CT slice width of 8mm [14].

Figure 4 .
Figure 4. Relative dose error between Monte Carlo calculated 3-field dose delivery and CT PGD, as a function of delivered dose.Shown are several different calibration techniques, including self-calibration (blue), the average calibration curve generated over multiple gels (i.e. a "library" calibration file, red) and the largest error calibration curve (green) [19].

Figure 5 .
Figure 5. (a) Monte Carlo-simulated reconstruction of a profile through a water phantom with a 2mm air tube in its centre.Clearly shown is the air tube at the centre of the water phantom.(b) Monte Carlosimulated dose response of an irradiated polyacrylamide PGD imaged with a 140kVp scanner [35].

12th
International Conference on 3D and Advanced Dosimetry Journal of Physics: Conference Series 2630 (2023) 012004 (a) (b) Figure 8. a) a deformable gel dosimeter constructed in a latex balloon and sealed in low-density polyethylene.b) The measured displacement of wax-beads undergoing deformation within the gel dosimeter.Shown are the original position of the beads (red), their deformed position (green), and each bead's displacement vector (blue arrows) [40].

Figure 9 .
a) 4 arc VMAT plan for a 6 multi-focal target SRS treatment was delivered to a PGD.The resulting polymerization in the 6 distinct target regions is clearly visible in the irradiated gel.b) Onboard CBCT (greyscale image) with treatment plan overlay (colour contours) for a single irradiated target within the multi-focal delivery[36].

Table 1 .
PGD formulation for CT imaging.