Image reconstruction for ion computed tomography

Abstract

Ion beam therapy is a well-established method to treat deep-seated tumors in cancer therapy. In contrast to x-rays, the energy deposition of ions in matter shows a characteristic maximum at the end of their range in a material. This maximum in dose deposition, referred to as Bragg-peak, allows for a high dose in the tumorous area while healthy tissue can be spared. For treatment planning, the location of the Bragg peak has to be known with high precision. Therefore, the stopping power of the irradiated materials, i.e. the energy deposition of a particle per unit path length, has to be extracted from prior measurements. Usually, an x-ray CT scan is generated, which results in a 3D image given in Hounsfield units, defined by the linear attenuation coefficients of the materials. For ion beam therapy, these values have to be converted to stopping power values of the materials via a calibration curve. However, this conversion is a main source of range errors in the treatment planning.Ion computed tomography allows to directly measure the stopping power of a material. Therefore, this imaging modality offers the potential to significantly reduce range errors for treatment planning. However, ions are affected by multiple Coulomb scattering and do not pass through a material in a straight line. Assuming a straight-line path and using conventional reconstruction techniques from x-ray CT without adaption therefore leads to significant limitations in spatial resolution and stopping power accuracy in the reconstructed3D image.To address this problem, a potential ion CT apparatus consists of position-resolving detector modules (tracking modules), to measure the entry and exit position and direction of each particle into the object to be imaged (phantom or patient) as well as a calorimeter to measure the particle’s residual energy. This setup is rotated around the patient or phantom and measurements are taken at multiple angles. From the tracking measurement, a path estimate inside the object is made for each ion and used in the subsequent reconstruction process. Here, an optimized cubic spline and a most likely path estimate have already been shown to generate the most accurate results.An ion CT demonstrator consisting of the already addressed elements is operated by the Institute of High Energy Physics (HEPHY) and the Technische Universität Wien (TU Wien). This thesis aimed at establishing a full reconstruction workflow for the demonstrator and testing it with simulated and measured data. Furthermore, this thesis aimed at investigating and improving the reconstruction workflow with phantoms of clinically relevant size regarding spatial resolution and stopping power accuracy. Therefore, an ion CT setup was modelled in Geant4 Monte Carlo simulations and the simulated data were taken as input for the subsequent reconstruction. For the reconstruction itself, possibilities to use the CUDA-based x-ray CT reconstruction framework TIGRE for ion CT were investigated for the first time. Different path estimates were used in the reconstruction in order to compare the accuracy of the reconstruction result as well as the reconstruction time, which is a crucial factor when it comes to clinical application.Additionally, the demonstrator system was used to investigate other potential ion imaging modalities. Therefore, radiographic images based on elastic and inelastic scattering of ions with the atomic nuclei of the target material were recorded and analysed.