Influence of tissue non-homogeneities on the accuracy of 3-D dose distribution monitoring during gamma-ray radiotherapy

Delivering precisely the desired dose distribution to the patient during radiotherapy is of crucial importance for treatment success and monitoring the actual distributed dose opens the possibility to correct for deposition inaccuracies and thus achieve better conformance to the treatment plan. Positron emission tomography has already been successfully used to measure the residual radioactivity from (γ, n) reactions in the patient after the treatment with hadron therapy or high-energy gamma-ray radiotherapy (above ~20 MV) [1], [2]. When trying to use a similar approach for gamma-rays of lower energies (most commonly 6 to 18 MV) the only source of positrons is prompt pair production in the beam, calling for measurements while the linac is operating and saturating the detector. The problems with saturation were alleviated to a formidable extent by using advanced digital processing techniques to measure energies of incident particles at rates beyond 10 Mcps per each scintillation crystal. The approach suggested by [3] builds on the discovery that there is a strong correlation between the delivered dose and the density of pair production. Our research confirmed the findings, but only as long as homogeneous objects were involved. In an inhomogeneous object, the annihilation density was no longer proportional to the dose, but rather to the density of deposited energy. This means that the intimate knowledge of the material density is required for proper reconstruction of the dose field image from the PET measurement. Additionally, one gets 2-4 mm thick regions at material boundaries where the charge particle equilibrium gets broken and the correlation no longer holds. This is, in principle, possible to account for based on ct data that is always available in such cases, but definitely calls for further work.