On the comparison of two NIPAM gel dosimeters with high resolution 3D MRI sequences

Polymer gel dosimeters allow the measurement of a dose distribution in 3 dimensions. Irradiation induces polymerization reactions in the long chains of the gel molecules, which is characterized by a change in the transverse relaxation time T2. Gel dosimetry is mainly advantageous for its high resolution, but can also be expensive. The main goal of this project is to compare two normoxic polymer gel recipes made with NIPAM as monomer. The analysis is based on the quality of the images obtained for a fixed MRI scan time and a comparative cost of analysis of each 1L phantom. The ΔT2 maps are obtained by scanning the gels before and after irradiation with 3D sequences and the DESPOT2 technique is used to reconstruct the T2 maps. The phantoms were irradiated with 8 photon beams of 6 MV with a fixed 250 MU, 4 at two opposite sides of the jar (AP-PA configuration), to obtain 4 dose distributions of 4.7 Gy, 8.9 Gy, 13.2 Gy and 17.3 Gy. The SNR is then determined in the ΔT2 maps in function of the dose step and the concentration of NIPAM in the phantom. The first gel phantom is made with 15% NIPAM and has a total cost of $1561.65, including products and MRI scan costs. The SNR obtained for the 4 dose steps are 10.02, 31.18, 45.12 and 37.54. The contrast between the T2 before and after the irradiation in the 4 regions of dose are 0.45, 0.66, 0.78 and 0.85. The second phantom contained 5% NIPAM, cost 1186.85$ and the SNR obtained are 16.80, 27.08, 30.11 and 26.17 and the ΔT2 are of 0.61, 0.75, 0.79 and 0.77 for the 4 dose steps respectively. The comparison of the two recipes has shown that the increase of NIPAM concentration does not allow a significant increase in image quality for MRI. The recipe with 5% of NIPAM has a lower dynamic range but a better sensitivity at lower dose and is less expensive than the one with 15% of NIPAM. The results also show that by increasing the NIPAM concentration 3-fold, the cost of a 1L phantom increases by 30%.


Introduction
Gel dosimetry is a valuable tool for measuring 3D dose distributions and has many advantages for dosimetry: tissue equivalence, three-dimensional dose measurement, high spatial resolution and lack of energy dependence.There are two main categories of gel dosimeters: Fricke gels and polymer gels [1].Polymer gels can characterize complex and non-uniform dose distributions with a large dose range and no diffusion effects, which could be interesting to study complex treatment plans, but their fabrication is difficult and the access to MRI is limited.However, their fabrication is more challenging, their reading via optical CT is limited by scatter conditions, while MRI-based reading may require long (costly) scan times given low contrast-to-noise efficiency.Polymer gel includes normoxic gels, which means that they are less sensitive to oxygen contamination and can be manufactured under standard atmospheric conditions [1].One such type of gel is the NIPAM polymer gel which is useful for MRI analysis.During irradiation, a polymerization reaction is initiated in the long molecular chains, decreasing the transverse relaxation time (T2) proportionally to the delivered dose.By measuring the change in the T2 relaxation time, before and after the irradiation, a 3D dose map can be obtained with a calibration curve.The Driven Equilibrium Single Pulse Observation of T2 (DESPOT2) technique is used to produce the T2 maps from MRI sequences because of its high SNR efficiency per unit voxel size [2].
In 2006, Senden et al. [3] developed a polymer gel dosimeter using the NIPAM as monomer for his low toxicity and his high solubility in water.More recent studies have proposed new polymer gel recipes based on NIPAM monomer in significantly different concentrations.The goal of this work is to compare two new NIPAM gel recipes by analyzing the SNR and the ∆T2 value in the dose distribution in function of the cost of each dosimeter.Indeed, NIPAM is the most expensive product in the manufacture of the gel and the goal of the project is to analyze the cost of the contrast-to-noise ratio for each dosimeter while maintaining a constant reasonable MRI scan time of 30 min pre-and 30 min postirradiation.

Gel preparation
Two phantoms are manufactured based on previous NIPAM recipes presented in the articles of Maynard et al. (2020) and Chiang et al. (2013).The recipes are shown in Table 1.The fabrication procedure is based on the one presented by Olding et al. ( 2010) [6].The gelatine (300 type A, Sigma-Aldrich, Oakville, ON) is first dissolved in 80% of the distilled water and the solution is heated to 50°C.This temperature is maintained till the gelatine is completely dissolve.The BIS (N,N'-Methylenebis acrylamide, Sigma-Aldrich, Oakville, ON) product is then dissolved in the water-gelatine solution.At the same time, in another beaker the NIPAM (N-Isopropylacrylamide, Sigma-Aldrich, Oakville, ON) is dissolved in the 20% of distilled water remaining at ambient temperature without any heating.When NIPAM is completely dissolved in the water, the THPC (Tetrakis (hydroxymethyl) phosphonium chloride solution, Sigma-Aldrich, Oakville, ON) product is added, and the solution is stirred for about 1 to 2 minutes.Meanwhile, the water-gelatine-BIS solution is cooled to 34°C in running cold water.When the solution is cooled to the desired temperature, the water-NIPAM-THPC solution is added and the gel is then stirred for about 1 to 2 minutes and then is put in the 1L Polyethylene Terephthalate plastic jars (Modus Medical Devices Inc., London, ON).

Treatment planning and irradiation
The two phantoms are irradiated on a Truebeam Linac with a 6 MV photon beam and 250 MU.Four photon beams of 5x10 cm 2 , 5x7.5 cm 2 , 5x5 cm2 and 5x2.5 cm 2 are used at gantry angles in an AP-PA configuration (90° and 270°) for a total of 8 photon beams for each phantom

T2 maps production
The two phantoms are first imaged with 3D sequences on MRI.The technique used is based on the method on T2 maps reconstruction DESPOT2 [2].Two sequences FLASH multi-echo with optimized flip angles combined with a B1 map are used to obtained T1 maps.Then, T1 and B1 maps, with four bSSFP sequences with multiple phase and flip angles are used to solve for T2 voxel per voxel.The two phantoms were imaged before and after the irradiation to further improve uniformity.The ∆T2 = (T2 pre − T2 post )/2  map is obtained by the subtraction of the T2 map before and after the irradiation and normalized by the T2 pre-irradiation.The T2 map pre-irradiation is also filtered via Gaussian anisotropic diffusion in 3DSlicer (conductance of 0.5 and 5 iterations) to reduce noise propagation in the final ∆T2.The SNR in the ∆T2 map (which is also essentially a contrast-to-noise ratio in this case) is measured in region of interest (ROIs) positioned at the center of each dose step.

∆T2 maps production
The figure 3 shows the ∆T2 maps obtained for the two recipes.The 4 regions of dose of 4.7 Gy, 8.9 Gy, 13.2 Gy and 17.3 Gy can be seen in the sagittal and coronal views.

SNR and cost of gel phantoms
The signal to noise ratio (SNR) is calculated in 4 ROIs in the ∆T2 maps obtained using the technique of Preibisch and Deichmann [7], and its value is related to the dose predicted in each step.The Figure 2a shows the 4 measures of SNR for the recipes vs the dose, while Figure 2b shows the contrast vs the dose.The SNR achieved for both phantom recipes looks very promising.However, both recipes appear to have saturated at lower dose than we anticipated (about 10 Gy for the 5% NIPAM and 15 Gy for the 15% NIPAM).More investigation will be required to evaluate the response of each recipe at lower dose of 0 to 5 Gy, and whether a threshold dose response occurs for the gels.In Table 2, the cost of each of the 1L phantom is compared.This includes the cost of the products, the shipping, the jars and the MRI time scan of 32 minutes done before and after the irradiation [7].The MRI scan time is kept constant and the main difference in the cost of each phantom is the NIPAM and BIS products.The cost of a 1L jars of gel is about 375$ higher for a NIPAM concentration of 3 times higher and a BIS concentration 1.5 times higher.However, the cost of the SNR at 13 Gy is higher for the 5% NIPAM (1186.85$/30=39.6$ per unit CNR) than for the 15% NIPAM (1561.64$/45=34.7$ per unit CNR), while the opposite is true at low dose of 4 Gy.However, a fair assessment would require considering a range of cost in MRI scanning, as opposed to a mere average value of $500 per hour.We plan to do that in our final assessment.

Conclusion
A comparison between two gel dosimeters composed of NIPAM monomer allows to study the benefits of increasing NIPAM concentration in the gel by measuring the image quality for different doses for a fixed MRI time scan.The ultimate goal is to minimize the cost of the contrast-to-noise ratio at a dose of interest.A first recipe with a NIPAM concentration of 15% shows a higher SNR for high doses while the second gel has a better dynamic range and contrast at lower dose.The study of the cost of each phantom shows that a recipe with 3 times more NIPAM and 1.5 times more BIS is 1.3 more expensive and does not upgrade significantly the image quality.Gel dosimeters are known to be hard to manufacture but gel based on NIPAM monomer has shown progress and potential for increased clinical interest while they are less toxic.Even if MRI is the gold standard imaging, CT imaging can be used to characterize the change of opacity and density in the gel [1].One of the main problems remaining for clinical application of gel dosimeters is the cost of the fabrication and their required imaging time at MRI.The next steps are to adjust the recipe by playing on the concentrations of the ingredients and increase or decrease the scan time in the MRI.Other properties of the gel are also important, including the dynamic range, dose rate response, whether there is a threshold dose response, linearity, etc.We are still in the early stage of optimizing our own recipe and the MRI pulse sequences.

Figure 2 .
a) Signal to noise ratio and b) contrast according to the estimated dose steps delivered to the phantom for the recipe made with 15% NIPAM (recipe 1) and the 5% NIPAM (recipe 2).

Table 1 .
List of the ingredients (distilled water, gelatin, NIPAM, BIS and THPC) used in the two recipes tested.

Table 2 .
Calculation of the cost of each phantom of 1L from recipe 1 and recipe 2 presented in Table1.