Thesis

Artificial Intelligence Frameworks in quality control in Proton Therapy

• Call:

  ProtoTera Call 2020

• Academic Year:

  2020/2021

• Supervisor:

  Pedro Teles

• Co-Supervisor:

• Host Institution:

  IPO - Porto

• Granting Degree Institution:

  Universidade do Porto

• Typology:

  National

• Abstract:

  In radiation oncology, at some point among the various processes that exist to treat the disease, procedures involving the use of radiation in the execution of the treatment will certainly be an option to be considered. Currently, among the treatments widely available commercially, the use of photon and electron beams stands out, these proven to be effective in their role in delivering radiation doses. However, the use of ions, such as protons, carbon ions and others, still in research phase, classified in the scientific world as a new branch of radiotherapy, called Particle Therapy, sometimes also called Hadron Therapy, it presents itself as a potentially more accurate and precise option in the process of delivering absorbed dose to the tumor. Within these possibilities, we consider the widely spread use of protons in clinical radiation oncology, Proton Therapy. Proton therapy is now moving away from static collimators and compensators toward greater use of Intensity Modulated Proton Therapy - IMPT, a new modality in which a pencil beam is scanned across the patient while the beam energy and intensity are modulated according to the treatment program, which can include treatment from multiple directions. IMPT not only allows proton therapy to be applied to tumors that could not be accessed via the older proton-therapy treatment methods but also allows better control of the dose distribution and, therefore, greater success in sparing critical organs from harmful radiation dose. Proton therapy has to be evaluated not just in terms of the success of the irradiation in destroying the cancer but most importantly in terms of how well the effects of the radiation on the rest of the body are minimized. Even with the advent of IMPT the field of proton therapy has almost certainly not yet reached its full potential. One problem that remains to be satisfactorily resolved is the so-called range uncertainty. For proton therapy to be successful, the proton therapy range in the tissue of the patient must be accurately known in advance to plan the treatment and then machine the compensator or to program the IMPT. Up to now, the average energy depposited by proton in a medium per unit path length, the stopping power SP, has been estimated from X-ray CT images xCT by converting the X-ray absorption (linear attenuation coefficient) measured in Hounsfield Units HU to relative proton stopping power RSP, the stopping power expressed with respect to water. This entire process, which involve some specific procedures, is called calibration. The main interest is to build a curve relating HU and RSP directly. Therefore, as a result we obtain a so called calibration curve between HU and RSP. There are different methodologies to obtaining RSP values. One well recognized is called Stoichiometric calibration. Stoichiometric calibration is based on a data table (ICRU-44) which contains mean anatomic composition values from human tissues. It’s worth to highlight that the tabulated values are different from real tissue. To strengthen this argument, it is a fact that persons present different percentages of chemical composition a different phase of life (e.g, a young patient, from the percentage composition point of view, shows different values than middle age or old patients). Furthermore, the composition of a given type of tissue may depend on its location in the body(e.g, pelvic kidney). An alternative methodology to theory estimate RSP values was proposed by Bethe-Bloch's formulation (ICRU-49) to obtain those values from quantities relating to the medium and physics parameters from beam used. One of the most important parameters used into their arguments is defined as mean excitation energy (Im) from a specific substance or a medium. This quantity, which is essentially an experimentalal measurement, presents slightly different values depending on the publication. After the ICRU work was completed, new values of 80 +/- 2, 81.8, and 77 eV were reported. The difference in the Im values between 75 and 80 eV results in 0.8–1.2% differences in the SP in the energy range of 10–250 MeV, which implies the same impact on absorbed doses. Consequently, different levels of accuracy are present and influence on the range calculations for the proton beam. Besides the problem with these different approaches, X-rays interact very differently in material compared with protons, resulting in relations between HU and RSP that are not unique and can therefore be ambiguous. If, on the one hand, the calculation of the RSP has inaccurate aspects, on the other hand, the CT number is a scale that depends essentially on the electronic and electrotechnical control relating to the manufacturer of the CT device. CT number directly relates to the scanner spectrum emitted for each CT device, originally built and installed. Therefore, an empirical performance between HU and a tissue structure is established for each specific device. Artifacts in X-ray images can produce additional errors in calibration curve of HU to RSP. Range errors up to 11% in the head have been reported. Recent works predicts typical errors of 1.8% and 1.1% for bone and soft tissue respectively, although the presence of higher density materials, and the resulting beam-hardening artifacts, can result in larger errors for specific cases, depending on the position of the sample within the body and the size of the body. In general, errors in the range prediction increase with the complexity of the geometry through which the protons must pass, but the use of Monte Carlo simulations can be a significant aid in reducing these uncertainties. Thereby, inherent uncertainties relating HU and RSP values also influence various other aspects that constitute the whole treatment plannning delivery dose, from patient positioning and immobilization, passing through critical intermediate steps, such as target volume definition and healthy tissue delineation to the final plan approval. These aspects have a considerable influence over the target structure's margin choices and the final absorbed dose calculation.