< Back to previous page

Publication

Multi-modal image analysis for selective internal radiation therapy dosimetry

Book - Dissertation

The role of external beam radiation therapy in liver cancer management remains restricted because of poor tolerance of normal liver parenchyma to radiation. One solution is to follow the Paul Ehrlich (1854-1915) proposal, "we must learn to shoot microbes with magic bullets'", to develop a radionuclide magic bullet to maximize the tumor irradiation while sparing healthy liver parenchyma as much as possible. The liver has an dual blood flow mechanism; the difference in blood supply between liver malignancies and the normal liver parenchyma, which is predominantly arterial and portal, respectively. Selective internal radiation therapy (SIRT) which is utilizing microspheres loaded with a high-energy ß-emitting radioisotope (e.g. yttrium-90 or holmium-166) benefits from this mechanism; the microspheres target the tumors passively, when infused into the hepatic artery, and consequently deliver lethal tumor irradiation while sparing a significant portion of the non-tumoral liver tissue. As mentioned before, despite the advantageous targeting of the targeted tumors, a fraction of the microspheres can be accumulated within the non-targeted tissues. In Europe, to perform a successful treatment, quantifying the tumor and non-tumor tissue irradiation is compulsory for treatment planning and for treatment verification. Because of a considerable inter- and intra-patient variation in tumor and liver tissue vascular anatomy, classical models that assume "the same relative tracer uptake in the tumor and the normal liver parenchyma in all patients" are over-simplistic for this task. Radionuclide therapies are historically prescribed in a (semi-)empirical manner and some evidence shows personalizing treatment planning could significantly enhance the therapeutic outcomes. As a result, in each individual patient, a careful treatment planning by employing a simulation workup to estimate the distribution of the therapeutic microspheres is essential. In addition, a more precise treatment evaluation is necessary to accurately determine the localization of the radionuclide in the patient's body and to identify potential adverse effects in terms of treatment efficacy and safety. The tumor and non-tumor irradiation are usually represented by absorbed dose or absorbed energy per unit of mass; this scheme is called dosimetry. The existing dosimetric tools do not adequately extract all required information. The most commonly used method, initially developed for diagnostic applications in nuclear medicine, is widely accepted to be insufficient for internal radionuclide therapy due to several questionable assumptions: (i) the person is represented by a standard mathematical model, and the morphology of each person is not taken into account, (ii) activity is assumed to be distributed uniformly in the source organ (or sub-organ). In contrast, voxel-level dose calculation addresses the patient-specific morphology and activity heterogeneity. Voxel-level dose estimation also could be employed in treatment planning to determine a maximum injectable activity tailored for each patient based on healthy tissue tolerance criteria and tumor dose coverage. One drawback of voxel-level dosimetry is the requirement for sophisticated image registration and segmentation to obtain detailed information about activity distribution and the patient's liver and tumor anatomy. The objective of this study was to develop a personalized dosimetric tool that answers the needs of SIRT for hepatic tumors. In this manuscript, a quantitative multi-modal image processing framework for SIRT was developed, evaluated, and applied to improve dose prediction for treatment planning, and evaluation of the treatment. The design of our procedure enables the treatment team to practice voxel-level dosimetry in a busy clinical routine. The largest part of this study was dedicated to developing registration and segmentation techniques for reporting an accurate and comprehensive absorbed dose distribution in clinically relevant volumes of interest.
Publication year:2020
Accessibility:Closed