Optical-CT gel-dosimetry I: basic investigations

Abstract

Comprehensive verification of the intricate dose distributions associated with advanced radiation treatments is now an immediate and substantial problem. The task is challenging using traditional dosimeters because of restrictions to point measurements (ion chambers, diodes, TLD, etc.) or planar measurements (film). In essence, rapid advances in the technology to deliver radiation treatments have not been paralleled by corresponding advances in the ability to verify these treatments. A potential solution has emerged in the form of water equivalent three dimensional (3D) gel-dosimetry. In this paper we present basic characterization and performance studies of a prototype optical-CT scanning system developed in our laboratory. An analysis of the potential role or scope of gel dosimetry, in relation to other dosimeters, and to verification across the spectrum of therapeutic techniques is also given. The characterization studies enabled the determination of nominal operating conditions for optical-CT scanning. "Finger" phantoms are introduced as a powerful and flexible tool for the investigation of optical-CT performance. The modulation-transfer function (MTF) of the system is determined to be better than 10% out to 1 mm(-1), confirming sub-mm imaging ability. System performance is demonstrated by the acquisition of a 1 x 1 x 1 mm3 dataset through the dose distribution delivered by an x-ray lens that focuses x rays in the energy range 40-80 KeV. This 3D measurement would be extremely difficult to achieve with other dosimetry techniques and highlights some of the strengths of gel dosimetry. Finally, an optical Monte Carlo model is introduced and shown to have potential to model light transport through gel-dosimetry systems, and to provide a tool for the study and optimization of optical-CT gel dosimetry. The model utilizes Mie scattering theory and requires knowledge of the variation of the particle size distribution with dose. The latter was determined here using the technique of dynamic-light-scattering.

FIG. 1

Spider plot of prominent dosimetry specifications common to all radiation therapies. Radial axes correspond to the different specifications and exhibit a scale from 0-5. The performance of a dosimeter(or the requirements of a particular treatment technique)is rated for each specification by a number 0-5 (0=poor and 5=high). An ideal dosimeter would thus track along the outside “5” spider track.

FIG. 2

Modified optical-CT scanner incorporating automated multislice acquisition. Components are A=laser, B=mirrors, C=beamsplitter, D=reference diode, E=linear “stepping” translation stage, F=optical-bath containing immersed gel-dosimeter, G=vertical translation stage mounting rotation stage and suspending gel-dosimeter in the optical-bath, H=field photodiode, and I=X95 support. The two mirrors (B)on opposite sides of the water-bath (F)translate the laster translate the laser beam across the optical bath for acquisition of each projection.

FIG. 3

Photograph of a 12 cm diameter finger phantom containing four variably attenuating blue gelatin fingers. Test-tube depressions in a gel can be filled with solutions(or gel), that have a modified refractive index, attenuation, or scattering power. Here, the blue fingers were gelatin gel that had been dyed with precise concentrations of blue food coloring to effect controlled variations in the attenuation coefficients of each finger. This approach is a versatile method to create known optical conditions to validate the optical-CT technique.

FIG. 4

Prototype x-ray lens from Osmic Inc. The lens emits focused x rays from 40-80 keV.

FIG. 5

(a)Illustrative LightTools simulation of 200 photon tracks from a 632 nm laser incident on a 32 mm thick slab of PAG gel, which has been irradiated to a significant dose(Wire frame box). Scattered rays hitting the sides of the gel are automatically terminated. The laser spot size at the measurement plane is significantly broadened by scattering inside the gel. The number of histories is limited for display purposes. (b)Measurement geometry matching the simulation in(a). The laser is collimated at(a) and then incident orthogonally on the gel flask, where it is observed to scatter in the polymerized gel region (5 cm×5 cm extent). A collimated photodiode point detector (b)translates behind the dosimeter as indicated, to measure the cross-plane profile of the laser beam 3.5 cm after exiting the gel.

FIG. 6

Sample optical-CT images of the same finger phantom, illustrating the effect of varying key acquisition parameters on image quality: ADC averaging (N_{ADC read}), rotational increment between projections (dtheta=360/N_projections), and linear step size between line integrals (dx). Figures are presented in pairs corresponding to the lowest and highest quality images for a particular acquisition parameter (further details given in the main text). Units are attenuation coefficient (mm^-1) relative to the optically matched bath.

FIG. 7

Quantitative analysis of the optical-CT images presented in Fig. 6, together with other scans not shown. Fingers are numbered according to increasing attenuation. Figures i, ii, and iii are standard deviations in the reconstructed pixel value for small regions located centrally in the fingers. Figures iv, v, and vi are mean pixel values in the same regions. Pixel values are attenuation coefficients relative to the matched bath in mm^-1.

FIG. 8

(a)Axial optical-CT scan of a finger phantom containing fingers of varying optical attenuation. (b)A comparison of the attenuation coefficient measured in optical-CT versus spectrophotometer values. The close agreement and near unity slope verify that the system is capable of measuring the absolute attenuation coefficient to within 3%-4%.

FIG. 9

Measurement of spatial resolution for the optical-CT system.(a)A magnified view of the optical-CT image of a 0.25 mm diameter steel wire (black circle superimposed to illustrate the wire cross section), showing the full-width at half-maximum of 0.66 mm (b)The MTF curve for the reconstruction in (a), showing better than 10% MTF out to 1 mm^-1. The MTF was corrected for the diameter of the wire by dividing by the Fourier transform of the wire circ function.

FIG. 10

Variation in optical-CT profiles through the fingers in Fig. 8(a)over a 4 h period.(a)Cross-plane profiles through the four fingers for scans acquired at time zero(solid curves) and t=4 h (dotted curves). (b)Plots of the image pixel value in the central, shoulder, and toe regions of Cell #1 versus time.

FIG. 11

Optical-CT images of the relative dose distribution delivered in a PAG after irradiation by focused photons (40-80 KeV)emitted by the x-ray lens (shown in Fig. 4). The cylindrical gel flask is ∼8 cm in diameter and the slices are 1 mm apart. The in-plane resolution is 1×1 mm². The comprehensive 3D measurement presented here would be very difficult to achieve with conventional dosimeters because of the low energy of the radiation and the small dimensions of the field.

FIG. 12

Optical-CT images of reconstructed attenuation coefficients in 120 kVp irradiations. In the left image the unfocused x-ray beam is collimated to 5 mm diameter utilizing a lead sheet. In the right image, the x-rays have been focused to a point in the gel dosimeter utilizing the x-ray lens in Fig. 4.

FIG. 13

PAG gel polymer microparticle size distributions, as determined by dynamic light scattering (DLS), for gels irradiated to different doses.

FIG. 14

Measured(a) and Monte Carlo calculated (b)profiles across the laser beam on the exit of PAG gels irradiated to different dose levels [measurement setup shown in Fig. 5(b). The Monte Carlo model is observed to reproduce the general behavior of the measured profiles but the limitations of the Mie model are observed in the failure to accurately model the penumbral tails of the PAG gels receiving a low and medium dose.(Note: discrepancies are highlighted by the log scale.