Stoichiometric Calibration and its adaptation to Dual Energy CT for Proton-therapy treatment planning

In the field of oncology, proton therapy emerges as an improvement on the precision previously offered by photon beam therapy, known as conventional radiotherapy. With photons, the doses are progressively diminished on their way through the tissues, while the doses provided by the protons are almost completely discharged at a specific point, the so-called Bragg Peak. The latter offers the advantage of concentrating the doses in the tumour tissues, avoiding damage to the surrounding healthy tissues. The method consists in the emission of proton beams with different directions and energies in order to achieve the complete treatment of the tumour. These beams have to be defined correctly in advance taking into account the conformation and composition of the patient's tissues. An algorithm based on the Monte Carlo method is used to compute the dose deposited by these beams in the patient anatomy. This method uses physical interaction statistics to predict the particle trajectories and their energy loss. Therefore, the MCsquare algorithm, which is based on this method, uses information on the nature of the particles, in this case protons, to define their behaviour when they come into contact with biological tissues. At the same time, it considers the information provided on the characteristics of the tissue with which it will interact. This information are the electronic densities and Stopping Powers. Both can be extracted from Computed Tomography images of the patients. In these images the tissues of interest can be identified and by measuring the value of the pixels that represent these tissues, the Hounsfield Units or HU are obtained. This thesis focuses on the calculation and choice of the best relationship between the values of HU obtained from the images and the human tissues that they represent in reality. This relationship is a calibration and in this case the one used is the Stoichiometric Calibration, which defines the dependency of the loss of proton energy represented by the HU as a function of the atomic number of the elemental compositions of the tissues. Specifically, in this thesis different calibration curves for CT scanners are obtained from measurements of the Gammex 467 phantom and various liquid mixtures of water, salt, sugar and ethanol in different proportions prepared in the laboratory. From them, CT images are obtained at different energies for stoichiometric calibration. Finally, the data provided by the phantom is chosen to study the application of the joint calibration of the information provided by several images obtained at different energies, a method known as double-energy or multi-energy. Following the method of Bourque et al, the information of HU obtained at both energies is taken and joined in a same coefficient to later calculate the Effective Atomic Number and the relative electronic densities associated with the data distribution of the two images. Thus, the EAN data and mass densities are achieved, giving rise to the calibration that can be introduced in the dose calculation algorithm of the proton therapy treatment. Therefore, the associated Stopping Power is also obtained, and a 3D map of an anthropomorphic phantom is created with the information of how much stopping force is going to be assumed by each voxel of the image.