Search

openaire: EC/H2020/810377/EU//ConnectToBrain Therapeutic non-invasive transcranial brain stimulation with previous treatment protocols showed at best moderate effect sizes and large interindividual variability with a substantial proportion of non-responders. A currently intensively discussed approach to address these problems is individualized closed-loop stimulation. ConnectToBrain is a synergy project funded by the European Research Council to develop noninvasive closed-loop therapeutic stimulation of network disorders of the human brain. It consists of three main pillars: (1) development of a multichannel transcranial magnetic stimulation (mTMS) coil array that covers nearly all of the cerebral cortex and enables highly precise electronic control of location, direction, intensity and timing of the induced electrical fields, (2) development of real-time analysis of activity and connectivity in brain networks using electroencephalography (EEG) for instantaneous spatial and temporal control of stimulation (brain state-dependent, closed-loop stimulation)and adaptive optimization of treatment effects by machine learning and (3) translation of these neurotechnological innovations into physiological and clinical studies.

Es wird eine für die Realzeit-Analyse geeignete SW-Architektur vorgestellt, die eine weitgehend automatisierte, echtzeitfähige Implementierung graphischer Spezifikationen ermöglicht. Es werden zustands- und signalflussorientierte Laufzeitmodelle berücksichtigt sowie deren Abbildung auf zyklische und ereignisgesteuerte SW-Teile unterstützt. Alle Berechnungswünsche im System werden auf Ereignisse abgebildet. Ihre statischen Prioritäten dienen zur Laufzeit nicht nur zum Sortieren der Warteschlangen, sondern werden auch bei der Verarbeitung dem jeweiligen Server-Prozess vererbt. Um ein analysierbares Zeitverhalten der SW-Architektur zu erreichen, wurden unterschiedliche Prioritäts-Vererbungsstrategien (Basic-Priority-Inheritency, Preemption-Threshold) entwickelt und im unterlagerten Laufzeitsystem umgesetzt. Um einen durchgängigen modellbasierten Entwurfsprozess zu erreichen, wurden die Konzepte für das Werkzeug Simulink/Stateflow umgesetzt. A software architecture for the widely automated implementation of graphical models is presented. The combination of state- and signal-based runtime models are considered as well as the mapping onto cyclic and event driven software parts is supported. All computational requests in the system are mapped onto events. Their static priorities are used during runtime to sort the event queues and are inherited by their respective server threads. To achieve a predictable behavior for a worst-case response time analysis of the overall sw-architecture different priority inheritance strategies (basic priority inheritance, preemption-threshold) were applied and realized in the underlying runtime system. The presented concepts were realized within the tools Simulink/Stateflow to provide an integrated development process.