In this section
@phdthesis{Giannari:2024,
author = {Giannari, A},
title = {Modelling of brain neuronal networks and therapy design for neurodegenerative diseases via nonlinear control},
url = {https://profiles.imperial.ac.uk/anastasia.giannari14},
year = {2024}
}
TY - THES
AB - This thesis introduces a novel computational framework for modelling heterogeneous neuronal networks as well as for developing optimal virtual treatment regimens via the application of nonlinear control. The modularity, scalability and adaptability of the proposed framework are attributed to the segregation of the neuron and network dynamics via a control-inspired feedback structure. This allows for the independent manipulation of the connectivity matrices that define the network structure and size. The involvement of biophysically realistic variables and parameters make it ideal for the study of healthy networks as well as ones that are compromised by neurodegeneration under the possible influence of pharmaceutical substances. We focus on providing scalable one-dimensional and two-dimensional realistic lateral inhibition models, which perform contrast enhancement and edge detection on visual stimuli, with the latter being responsible for the perception of optical illusions by the human eye retina. The abnormal perception of the illusions is an early symptom of diabetic retinopathy, a neurodegenerative disease that attacks the synaptic couplings within the lateral inhibition network. Based on this, we produce diabetic lateral inhibition models that fail to perceive the optical illusions by altering the parameters that are impaired due to the pathophysiology of diabetic retinopathy. We consider the healthy and diabetic images of optical illusions as computational phenotypes and the error between them is used to design an adaptive terminal error iterative learning controller to find the optimal drug amount sufficient to recover the functionality of the network and therefore to restore the perception of the optical illusions. The virtual drug acts upon the defective model parameter that is considered an effective therapeutic target. To the extent of our knowledge, this is the first instance a realistically measurable output is utilised as feedback to a nonlinear contro
AU - Giannari,A
PY - 2024///
TI - Modelling of brain neuronal networks and therapy design for neurodegenerative diseases via nonlinear control
UR - https://profiles.imperial.ac.uk/anastasia.giannari14
UR - http://hdl.handle.net/10044/1/113893
ER -