Search

Phosphorylation cycles are a common motif in biological intracellular signaling networks. A phosphorylaton cycle can be modeled as an artificial biochemical neuron, which can be considered as a variant of the artificial neurons used in neural networks. In this way the artificial neural network metaphor can be used to model and study intracellular signaling networks. The question what types of computations can occur in biological intracellular signaling networks leads to the study of the computational power of networks of artificial biochemical neurons. Here we consider the computational properties of artificial biochemical neurons, based on mass-action kinetics. We also study the computational power of feedforward networks of such neurons. As a result, we give an algebraic characterization of the functions computable by these networks.

Image-guided radiotherapy has the potential to increase success of treatment by decreasing uncertainties concerning tumour position and shape. MRI is the modality of choice when it comes to imaging for tumour delineation and characterisation, set-up correction, treatment plan adaptation, response monitoring and intra-fraction motion compensation. To boost the possibilities of image guidance by providing images with superior soft-tissue contrast during treatment, integrated diagnostic quality mri functionality with a radiotherapy linear accelerator is pursued. To this end, an integrated MRI-linac was built in cooperation with Elekta AB (Stockholm, Sweden) and Philips (Best, The Netherlands). It features a 6 MV standing wave linear accelerator supported by a circular, continuously rotating gantry around a custom 1.5 T cylindrical MRI scanner with a split gradient coil. Solutions to two technical challenges related to MRI/linac integration are investigated in this thesis. First, the use of MR images in radiotherapy can be hindered by geometrical distortions due to magnetic field inhomogeneity problems. Techniques have been developed to deal with MR image distortions, or to prevent them from occuring at all. Secondly, the MRI scanner and linac must be interfaced seamlessly to allow advanced types of radiation delivery which compensate for target motion based on feedback from MR images. A software architecture was defined and implemented that facilitates the concept of real-time on-line MRI-guidance. Based on this framework a first demonstration of MRI-guided gated and tracked radiotion delivery to a moving target was given. Further research is needed to advance the technical prototype MRI-linac to the status of a clinical prototype.