In silico testing strategies for translational magnetic hyperthermia

  1. Irene Rubia Rodríguez
Supervised by:
  1. Daniel Ortega Ponce Director

Defence university: Universidad Autónoma de Madrid

Year of defence: 2022

  1. Raquel Perez Lopez Chair
  2. Erving Clayton Ximendes Secretary
  3. Silvio Dutz Committee member

Type: Thesis


This thesis is a compilation of different improvements in magnetic hyperthermia treatments safety thanks to numerical computer simulation (in silico testing). This therapy uses the heat released by magnetic nanoparticles injected into tumours when exposed to an alternating mag-netic field, which induces cancer cell apoptosis. Magnetic hyperthermia has been and is being successfully trialed to treat different types of localised tumours. In silico analysis has shortened the time needed to obtain results, while reducing the cost for producing the results. In addition, virtual tests make it possible to predict the outcome of therapy and its safety in an almost infinite variety of scenarios, thus optimising the risk-benefit ratio. For this purpose, there exists mag-netic hyperthermia treatment planning platforms based on computer simulations. This thesis presents different contributions to increase the accuracy of this planning software, as well as to increase the personalisation of therapy for each patient. These improvements are already im-plemented in the clinical trial that has derived from the European NoCanTher project. Chapter 1 is an introduction to magnetic hyperthermia and treatment planning, including a brief review of the most commonly used parameters for safety analysis. This chapter is closely related to the next chapter, chapter 2, which describes the different components used in treatment simulation, from the development of the virtual models that reproduce the clinical scenario to the mathematical models used to predict the outcome of the therapy. This chapter also includes a brief introduction to machine and deep learning, which is used to personalise therapy as much as possible by creating hyper realistic models based on the patient. Chapter 3 includes the most relevant results obtained in the development of this thesis. This chapter is divided into two sections. The first one, focused on preclinical studies, includes the analysis of the cooling of different tissues by comparing in vivo mouse models and in silico virtual mouse models. This section also investigates the mechanisms of thermoregulation of healthy and tumour tissues based on their physiology. Within this same chapter 3, the second section includes different studies with direct translation to the clinic. In this section, the first steps taken to automatically obtain a personalised model of the patient thanks to deep learning are presented. The importance of considering the blood perfusion of the tumour when elaborating this model to achieve an accurate calculation of the thermal dosage is also studied. Finally, a methodology is presented to estimate the risk associ-ated with a particular treatment scenario, thus reformulating the exclusion criteria for magnetic hyperthermia and recovering a large number of patients who may benefit from this therapy