Validating and optimizing personalized current flow simulations across the human lifespan using in-vivo magnetic resonance current density imaging

点击翻译按钮获取中文摘要

Computer simulations have become an important tool for characterizing and optimizing the electric field distribution induced by transcranial electric stimulation (tES) in the human brain. Simulations also enable personalized stimulation approaches that control for the impact of different head and brain anatomies on the individual field distribution, and they are an integral component of the research strategy of the research unit (RU). However, as simulations estimate the electric fields based on potentially uncertain information about head anatomy and tissue conductivities, validating their accuracy is highly important. Critically, theoretical analyses and invasive electrode recordings in selected human patients undergoing surgery showed that the accuracy of field simulations can vary strongly between individuals. This generates the risk that potential associations between the estimated fields and the recorded physiological responses are obscured.In project 10 (P10) of the RU, we will for the first time apply magnetic resonance current density imaging (MRCDI) to a large group of subjects in order to systematically and non-invasively validate the electric field simulations used in the RU. The planned work will leverage our recent, comprehensive work on MR acquisition schemes, optimized tES hardware and analytical methods, which helped to mature MRCDI and make it ready for the envisioned large-scale application in humans. We will start by collecting MRCDI data of 40 healthy participants for all target regions used in the RU. We will use this data in a new Bayesian analysis framework to systematically optimize the tissue conductivities of the personalized head models. Employing a Bayesian framework will reveal the uncertainty of the estimated conductivities in a principled manner, and give insight into which conductivities benefit from the optimization by MRCDI. As second step, we aim to extend this approach towards the comparison of head models of varying anatomical complexity using Bayesian model selection. Finally, we will test the impact of anatomical features (skull thickness, CSF volume) and selected demographic variables (age, sex) on the simulation accuracy at the individual level. These results will be used for the development of an optimized head modelling pipeline with improved accuracy which will be implemented in the post-hoc analyses in projects P1-9 of the RU and also provided open source for broad use.Complementing the above work, we will streamline the MRCDI acquisition procedures that currently require expert knowledge. The improved procedures will be pilot-tested within four projects of the RU, making MRCDI ready for a broader usage. In particular, this work will prepare MRCDI for its general use across all projects in the potential second phase of the RU, where changes in skull composition and brain anatomy at old age might require further adaptations of the simulations to ensure accurate field estimates.