Preliminary commissioning of a prototype solid tank optical CT scanner for 3D radiation dosimetry

Optical computed tomography (CT) is one of the leading modalities for imaging gel dosimeters. In previous research, it was shown that a design could significantly reduce the volume of the refractive index baths that are commonly found in optical CT systems. The proposed scanner has been manufactured and is in process of being commissioned. The rays refract through the system and their paths are estimated using a ray-tracing simulator. Reconstruction is preformed through algebraic reconstruction technique and iterated using the FISTA-TV algorithm. A Sylgard®184 edge phantom was created to commission the scanner’s spatial resolution. The scanner is capable of a 0.929 mm−1 spatial resolution, although the spatial resolution is highly dependent on the number of iterations, and method of processing the edge spread function.


Introduction
Optical computed tomography emerged as a potential method for scanning three-dimensional dosimeters in 1996 when Gore et al. demonstrated its applicability in scanning PAG polymer gel dosimeters [1].Since that initial work, a variety of scanner designs and methods have emerged [2][3][4][5][6].Typically, these systems employ a standard filtered back projection (FBP) technique for reconstruction.Back projection reconstruction operates under the assumption of a direct path from signal source to the detector, which necessitates that these systems carefully index match the optical path.The requirement of a direct ray path can make researchers go to great length to reduce the refraction in optical CT systems.The primary method that most optical CT scanner systems use to minimizing refractive artifacts is an index matching bath [1,[7][8][9][10].Managing a refractive index bath can be inconvenient, mess, and unappealing, as they often involve handling a relatively large volume (1-15L) of chemical mixtures (water, glycerol, propylene glycol, etc) that require constant maintenance and monitoring.The fluid can evaporate or separate over long scanning period, thus altering the remaining fluid's refractive index.In the event of the chemical mixture separating, Schlieren patterns may emerge in the viewing area as the result of motion in the fluid [11].
Previously published work has demonstrated that a solid tank scanner with a severely reduced need for index matching can be built and optimized given a set of design criteria [12].The purpose of this work is to commission the design found from that previous research.The prototype scanner incorporates a solid acrylic block with minimal index matching fluid (5-30mL) and eliminates the need for rays to follow a direct path from source to detector.The scanner utilizes an altered version of the ray-tracing simulator, developed during design phase, to generate a ray-path system matrix.The Fast Iterative Shrinkage-Thresholding Algorithm Total Variation (FISTA-TV) algorithm is employed to perform reconstruction using the algebraic reconstruction technique.The commissioning of this scanner is still a work-in-progress.

Materials and methods
The rationale behind the design, as well as the procedure have been discussed in-depth in a previously published paper [12].The prototype fan-beam scanner features a 635nm laser diode with a 60° linegenerating Powell lens.The fan-beam enters a 290mm long solid acrylic (PMMA) block.The front face of the block is machined into a 61mm radius semi-circle to direct the beam towards a 104mm diameter hole that has been bored through the block (Fig. 1).A gel dosimeter is placed in the cavity of the bore, and through the use of an O-ring (CR39932, SKF; Toronto, ON), a very small volume of refractive index matching fluid is suspended in the gap between the dosimeter and the block (Fig. 2).The rear wall of the block was specially machined to allow five 64-element photodiode arrays (S8865 Series, Hamamatsu; Hamamatsu City Japan) to sit flush up against the acrylic without a collimator.Each of the 320 photodiode elements has an active area of 0.8mm by 0.7mm, with a 0.1mm spacer between each active area.The gel containers are mounted to a pair of motion controlling stages (ESP300, Newport; Irvine, CA), allowing for rotational and vertical movement.This scanner is intended to be used on FlexyDos3D dosimeters, developed by De Deene et al. in 2015 [13].The design geometry is specific to the refractive index of the base dosimeter material, Sylgard®184 which is made from a 10:1 ratio of Sylgard®184 Silicone Elastomer and Sylgard®184 Curing Agent respectively.If dosimeters with other optical properties were to be tested, a separate block geometry and apparatus setup would have to be calculated as outlined in the previously mentioned design optimization paper [12].The overall design would remain relatively unchanged, outside of a new solid tank, and different laser set up.
Image reconstruction is performed using algebraic reconstruction technique (ART).The system matrix used during the reconstruction is a geometry matrix that was created using a retooled fork of the ray-tracer created for the scanner design optimization.Reconstruction iterations were performed using the FISTA-TV algorithm.Many other algorithms were tested on 24 synthetic test problems and 12 clinical test problems by Guenter et al. in 2021, and FISTA-TV was a top performer in both solution quality and low runtime [14].

Results and discussion
Spatial resolution was commissioned by creating a Sylgard®184 phantom, where a circular column was molded from a precisely machined rod.The mold was then back-filled with a dyed Sylgard®184 mixture (Fig. 3).After scanning the phantom and reconstructing the image, the centroid of the high-attenuation insert is found and the distance from the centroid is calculated for each pixel (Fig. 4).The edge spread function (ESF) of the region of circular attenuation with well-defined geometry is found by fitting a cubic spline to the edge data (Fig. 5), and the derivative of the ESF returns a line spread function (LSF).Applying a fast-Fourier transform to the LSF results in the modulation transfer function (MTF), and the 50% cut-off frequency is reported as the system's spatial resolution.With iterative reconstruction the number of iterations can greatly affect the clarity of an edge, as well as the noise in a system.During this experiment reconstructions were performed with a varied number of iterations (Table 1), and as expected a larger number of iterations corresponded to a higher spatial resolution up to 200 iterations.Due to the highly-varied nature of data on either side of the edge, some improvement to the spatial resolution can be found by cropping the data around the edge.The data was cropped using the KARI method, as outline in Viallefont-Robinet et al. in 2018 [15], resulting in significant improvements in spatial resolution.The highest spatial resolution result (MTF50) was found to be 0.929 mm -1 after 200 iterations (Fig. 6).More iterations resulted in a loss of spatial resolution.Reconstructions with less than 80 iterations did not result in the conditions required for the KARI method of cropping, so those iteration numbers were disregarded.

Conclusion
In previous research, an optimal scanner was designed that would lessen the need for index matching and implement algebraic reconstruction technique with the FISTA-TV algorithm to reconstruct the images.That previously designed scanner has been manufactured, and commissioning is a work-inprogress.Currently, it was found that the scanner is capable of achieving a spatial resolution of 0.929 mm -1 .Best results were achieved at 200 iterations when cropping the data around the ESF.The future of this work is to finish commissioning the prototype scanner and characterize any artifacts that emerge during scanning or reconstruction.

Figure 1 .
Figure 1.Shown is a bird's eye schematic representation of the scanner apparatus.The block, container, bore gap, and ray paths are all shown toscale if a Sylgard®184 based dosimeter is held in the container.

Figure 2 .
Figure2.Shown is a not-to-scale crosssectional schematic of the scanner apparatus.It can be seen that the dosimeter container is fixed to the motors from the base, and that an O-ring is used to allow the bore gap to be filled with index matching fluid.

Figure 3 .
Figure 3.A schematic of a back-filled phantom used to measure spatial resolution.

Figure 4 .
Figure 4.The reconstructed region of interest (200 iterations) of the back-filled spatial resolution phantom.

Figure 5 .
Figure 5. Edge data for the ROI in Figure 4 that has been fitted with a cubic spline.

Figure 6 .
Figure 6.Shown are the reconstructed slices at a) 80 iterations, and b) 200 iterations.The square region of attenuation in the images was not used for this work.The c) profiles through the center of the circular spatial resolution phantom (row 342) shows a steeper edge for as iterations increase to 200.The d) MTF at 80, 140, 200 and 300 iterations from the spatial resolution phantom once ESF cropping has been implemented.Highlighted is the MTF50 for each data set.

Table 1 .
The MTF50 spatial resolution as the number of iterations increases.