Australasian Physical & Engineering Sciences in Medicine Volume 28 Number 2, 2005 A fast, high spatial resolution optical tomographic scanner for measurement of absorption in gel dosimetry T van Doorn1,2, M Bhat1,2, T P Rutten2, T Tran2 and A Costanzo2 Department of Medical Physics, Royal Adelaide Hospital, Adelaide, Australia School of Chemistry and Physics, University of Adelaide, Adelaide, Australia Abstract A fast tomographic optical density measurement system has been constructed and evaluated for application in Fricke 3D gel dosimetry Although the potential for full three-dimensional radiation dosimetry with Fricke gel dosimeters has been extensively reported, its application has been limited due to a lack of fast optical density measurement systems In this work, the emphasis of the design has been to achieve a short scan time through the use of precision optics and minimal moving parts The system has been demonstrated in the laboratory to be able to achieve better than 1mm resolution and a scanning time per tomographic slice of 2.4 seconds Full volumetric sampling of a 10 cm diameter by 7cm long cylinder can be achieved in minutes When applied with a Fricke based gel dosimeter a linear response between reconstructed CT number and absolute dose was better than 3% interrogates the sample A range of optical scanners developed by different groups are included in Table While optical gel dosimetry has scored highly in Oldham’s spider plot for most parameters, it has ranked poorly for the time and cost factors There is also potential to improve the accuracy and sensitivity A fast optical CT scanning system is desirable to avoid degradation of the spatial dose information resulting from Fe3 ion diffusion, thereby preserving quality and the integrity of imaged data Moreover, in a clinical setting a fast readout system has practical advantage over slow readout systems In the work presented in this paper, a fast scanner based on precision optics and minimal moving parts has been proposed to tighten the RTAP criteria to RTAP(1,3,3,1) (resolution 1×1×1 mm, imaging time min, accuracy within 3%, and precision within 1%) Cost has also been proven to be able to be improved and the system is closer to Oldham’s optimal system It is appropriate to note however that the light attenuation through the gel is determined by changes in the optical absorption coefficient of the irradiated gels Whereas in irradiated polymer gels, the differential attenuation is the result of increased scatter24 The scanner is designed for absorption measurement in Fricke gels and has not been optimised for attenuation due to scatter The reconstructed image by this scanner is therefore free from scatter artefacts highlighted by Oldham and Kim with highly scattering gels25 Key words gel dosimetry, optical CT, 3D dosimetry, FBX dosimeter, conformal radiotherapy Introduction The need for reliable, three-dimensional dosimetry in modern radiotherapy has been extensively reported in conjunction with reports of developments in gel based dosimetry1-21 Oldham et al.19 have proposed that the 3-D dosimetry system should meet the RTAP(1,60,3,1) criteria (resolution 1×1×1 mm, imaging time 60 min, accuracy within 3%, and precision within 1%) Oldham et al20 further described the strength and limitations of different dosimeters using a spider web plot A rating of zero indicates that a dosimeter has no functionality for that specification; a rating of indicates highest relative functionality An ideal dosimeter with global application would rate on all radial axes In Figure 1, Oldham’s spider web plot has been extended to include a tissue equivalence factor Overall, gel dosimetry with optical scanners to measure the induced attenuation is currently ranked the highest Many optical gel scanning systems, including commercial units, have been reported in the literature5-7, 9-12,14,17-23 Primarily they differ in the way the light Corresponding author: Madhava Bhat, Department of Medical Physics, Cancer Centre, Royal Adelaide Hospital, SA 5000, Australia, Tel: 08-82222458, Fax: 08 82225937 Email: mbhat@mail.rah.sa.gov.au Received: 22 April 2004; Accepted: 24 March 2005 Copyright © 2005 ACPSEM/EA Materials - the optical scanner design criteria The scanner was designed to have the capacity to form cubic voxels of 1.0 mm sides over a cylindrical sample of 100 mm diameter and 70 mm length through the 76 Australas Phys Eng Sci Med Vol 28, No 2, 2005 Tissue Equivalence Sensitivity Accuracy Cost van Doorn et al RMS noise The narrow beam divergence of less than 0.99 m rad resulted in less than 1.8 mm increase in beam width over the optical path length of the system (~1.8 m) However, focussing lens L1 reduces the beam width to 0.1mm Figure is a photograph of the scanner with a sample gel in place The cylindrical gel sample was positioned in a refractive index matching bath10, midway between, and on the axis of, the twin lens pair The cylindrical sample housing provides for rotationally invariant sampling during the tomographic scanning process The bath measured 14.0 (W) ×14 (L)× 10 (H) cm and the transparent wall facing the laser beam was constructed of optical quality glass to minimize refraction The diameter of the gel cylinder was made larger (120 mm) than the target diameter (100 mm) to reduce refraction effects on laser beam near the edges of target cylinder (see Figure 2) The current through the photodiode detector with time was in proportion to the optical attenuation of the beam as it was swept across a path that is displaced from, but constantly parallel to, the optical axis of the lens pair For calibration purposes, a stationary mirror momentarily reflected the laser beam, un-attenuated by gel the sample, onto the sample detector, providing a reference signal for correction of laser beam intensity fluctuations This reflected beam is attenuated so that a reference signal is approximately equal to the sample signal Time Energy Resolution 3D Ion chamber Film TLD • A fast, high spatial resolution Gel Figure Spider plot of important 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, = poor and = high An ideal dosimeter would thus track along the outside ‘‘5’’ spider track Adapted from M Oldham, et al., Med Phys., 2003, pp: 623-634 Sampling volume and rate The cylindrical sample holder was mounted on a belt driven platform inside the refractive index matching bath, for rotation through known angular increments (1.25 degrees) With maximum mirror rotation speed of 120 revolutions per second, a 180° sinogram can be achieved in 2.4 seconds and the whole volume swept in approximately minutes (70 x 2.4 seconds) Rigid mounting of the rotation platform with respect to the optical axis of the twin lens system ensured correct alignment of the two centres of rotation for accurate CT reconstruction The relationship between sweep rate, projection width and sampling rate is illustrated with the help of figure 4a The number of samples required per sweep period determines the sampling rate The sweep period (S) is the product of the period of rotation of the mirror (T) and the arc of circle (C) subtended by the projection width (P) and the focal length (f): (1) S = C ×T Figure Schematic diagram of scanner reconstruction of 1.0 mm tomographic slices orthogonal to the axis of the cylinder at 1.0 mm intervals This volume is adequate for typical SRS treatments and small field IMRT Rapid projection scanning was implemented through the incidence of a laser on a rotating mirror at the focal point of a lens to create a parallel set of ray paths through a rotating sample (see figure 2) Whenever the reflected beam falls within the acceptance aperture of the first lens L1 the beam is refracted parallel to the optical axis of the lens pair (L1 and L2) (Rolyn Optics, model number 20.1328, 240 mm effective focal length) These lenses have both faces anti-reflection coating for the visible spectrum (400700nm) with single layer magnesium fluoride, (MgF2) The beam is then refracted by L2 to the ‘sample detector’ photodiode at the focal point of the second lens The laser source and single photo-detector remain stationary The HeNe laser (Coherent, Model Number 31-2772) generated a 2mW CW beam of approximately mm diameter at 543.5 nm wavelength The laser generator operates in TEM00 mode producing a low-divergence circular (Gaussian) collimated beam with less than 1% where P tan − 2× f C= (2) π With the component dimensions defined above and 500 samples per projection, each sample must be acquired in less than microsecond The scanning process is a translation of a rotational motion to a displacement off axis which is non-linear with time The angle subtended by the rotating mirror θs in the sampling interval ts is given by: 77 Australas Phys Eng Sci Med Vol 28, No 2, 2005 van Doorn et al • A fast, high spatial resolution Rotating mirror Lens Sample Water bath Lens Laser Detector Figure Photograph of optical CT scanner The laser beam sweeps across the gel in parallel lines and a projection of the gel matrix is recorded by the detector C f P Detector (a) θs w p θb θa f (b) Figure a) Relationship of sweep rate, projection width and sampling rate b) Lateral Displacement (w) of laser beam during sampling interval 78 Australas Phys Eng Sci Med Vol 28, No 2, 2005 van Doorn et al • A fast, high spatial resolution Table Authors Year Outcomes Appleby A and Leghrouz A.5 1991 Demonstrated visualisation of radiation dose distributions by naked eye using gel sections Tarte B.J and van Doorn T.6,7 1993 Linear optical tomographic scanning of ferrous sulphate gels for full three dimensional imaging of dose Gore et al10 1996 Optical CT scanner for polymer gel sample in a fixed phantom and translating mirrors for optical transmission data acquisition Transmission data required correction to account for deviation of laser beam from its normal incidence angle Total data acquisition time for a 60x60 pixel image was Tarte et al11 1997 Demonstrated linear optical response up to 10 Gy dose in laser scanned agarose gel sections Relative dose distributions in good agreement with ionisation chamber readings Kelly et al12 1998 Optical CT scanner resembling a “first generation” x-ray CT scanner developed with fixed beam and rotating FBX gel sample in a translating phantom for optical transmission data acquisition Transmission data required correction to account for deviation of laser beam from its normal incidence angle Data collection time per slice for 100x100 pixels was h Wolodzko et al14 1999 CCD camera based detector, manual rotation of the FBX gel phantom to acquire CT projection data Highlighted that non uniform pixel sensitivity is an issue to be resolved in CCD based optical CT reconstruction Oldham et al19 2001 Compared MR scanned image with laser scanned optical CT image of polymer gel Defined Resolution Time-Accuracy-Precision (RTAP) criteria.(resolution