Development of a three-dimensional radiation dosimetry system

The direct non-destructive measurement of the radiation absorbed dose in three dimensions is considered to be technically difficult. Accurate determination of the spatial distribution of absorbed dose plays an important role in many applications particularly in medicine. In radiotherapy computer calculations are frequently used to estimate three-dimensional dose distributions in complex geometry, hence a practical dosimetry system able to provide three-dimensional (3-D) integrated measurements is highly desirable for verifying such dose predictions. Magnetic Resonance Imaging (MRI) has been used to visualise 3-D dose distributions, inside two different detector materials, namely the ferrous sulphate gel (Fricke gel) and the polymer gel system. Each of these procedures has its own drawbacks and limitations, and this research project sought to find improvements and alternatives to overcome these problems. Work on the Fricke gel led to an improved preparation procedure employing gelatin gel whose lower melting point reduces the possibility of dissolved oxygen loss. The role of each component was clarified which led to the omission of all unnecessary chemicals such as the sodium chloride and benzoic acid. Initially MRI was the only 3-D readout technique available, however simple relaxometry was used to characterise the detector quantitatively with each modification before employing an MRI scanner to obtain images. Optimisation of the active constituents saves time and effort, and minimises the cost of equipment as well as materials. A serious drawback of the Fricke gel is ion diffusion, which causes blurring of the recorded spatial distribution and much effort was given to attempts to reduce this. However it was concluded that it is possible to slow down ion diffusion but at the cost of detector sensitivity. Therefore the best way of dealing with this problem is by introducing a fast readout technique so that the dose distribution can be recorded before serious diffusion blurring has occurred. The polymer gel system normally includes acrylamide but this was avoided for safety reasons. In order to resolve the dissolved oxygen problem which makes the preparation and storage of the dosemeter difficult a modification was proposed by which oxygen would interact with other chemicals and not with the monomers in this type of dosemeter. However, research in this direction was halted as attention turned to manufacturing a convenient, safe, highly sensitive and reliable radiochromic gel detector based on a modified version of the Fricke dosemeter. This material turned out to be extremely transparent which led to the construction of a simple, inexpensive optical tomography imaging device capable of very fast readout. A radiochromic gel dosemeter, in which ionising radiation causes a colour change in the visible wavelength region, was produced by modifying the conventional Fricke gel dosemeter. The roles of the three active components in the Ferrous sulphate Xylenol orange Gelatin (FXG) gel dosemeter were quantified with special consideration given to their effect on the system sensitivity and stability. Optimal composition was found to be 0.5 mM ferrous sulphate, 0.1 mM xylenol orange and 25 mM sulphuric acid. The dose response is linear in the range 0.1 to 30 Gy. Its sensitivity was found to be ΔA = 0.084 cm-1 per Gy, where A is the optical absorbance measured at a wavelength of 585 nm. Radiation induced colour changes varied by less than 5% over 24 hours providing that samples are stored in a cool dark place such as inside a refrigerator. An optical tomography scanner with a parallel light beam and a CCD detector was designed, built and used to interrogate cylindrical samples of FXG. The final version of the scanner can cope with samples up to 10 cm in diameter limited by the width of the light beam. A maximum of 512x512 pixels per slice images were possible, however readings from neighbouring CCD channels were averaged together to yield a cross-sectional image on a 128x128 array. Each 3-D optical data set consists of 402 2-D projections each averaged from 10 CCD frames. Slice thickness as small as 0.14 mm can be achieved in this case, which is much narrower than the typical MRI value of about 6 mm. The time to acquire a stack of a 100 cross-sectional images with a resolution of 128x128 pixels per slice is about 20 minutes. This can be compared with the few minutes required to acquire a good quality image of a single slice by MRI. The 3-D imaging performance was tested with collimated X-ray beams which gave encouraging results. (author)