Virtual brain simulations reveal network-specific parameters in neurodegenerative dementias
Creators
- Monteverdi Anita1
- Palesi Fulvia2
- Schirner Michael3
- Argentino Francesca
- Merante Mariateresa
- Redolfi Alberto4
- Conca Francesca1
- Mazzocchi Laura2
- Cappa Stefano5
- Costa Alfredo6
- Pichiecchio Anna6
- Farina Lisa1
- Cotta Ramusino Matteo1
- Jirsa Viktor7
- Ritter Petra3
- Gandini Wheeler-Kingshott Claudia8
- D'Angelo Egidio6
- 1. IRCCS Mondino Foundation, Pavia (Italy)
- 2. University of Pavia, Pavia (Italy)
- 3. Charité – Universitätsmedizin Berlin, Berlin (Germany)
- 4. IRCCS Centro San Giovanni di Dio Fatebenefratelli, Brescia (Italy)
- 5. Scuola Universitaria Superiore IUSS di Pavia, Pavia, (Italy); IRCCS Mondino Foundation, Pavia (Italy)
- 6. University of Pavia, Pavia (Italy); IRCCS Mondino Foundation, Pavia (Italy)
- 7. Institut de Neurosciences des Systèmes, Marseille (France)
- 8. University of Pavia, Pavia (Italy); IRCCS Mondino Foundation, Pavia (Italy); University College London, London (United Kingdom)
Description
Neural circuit alterations, although laying at the core of brain physiopathology, are usually hard to unveil in living subjects. Virtual brain modelling (TVB), by leveraging structural and functional MRI to simulate brain dynamics, can yield mesoscopic parameters of connectivity and synaptic transmission including excitatory and inhibitory coupling and recurrent excitation. In this work, we used TVB to simulate resting-state network dynamics in Alzheimer disease (AD) and Frontotemporal Dementia (FTD) patients and compared them to healthy controls (HC). The simulated parameter patterns differed between AD and FTD networks. Individual subjects, even when belonging to the same group (e.g., AD, FTD, or HC), presented subtle differences in network parameter patterns that significantly correlated with their own neuropsychological, clinical, and pharmacological profiles.
This database includes structural and functional connectivity matrices estimated from tractography and rs-fMRI time-series of each subject analyzed (10 HC, 16 AD, 7 FTD). An ad-hoc grey matter (GM) parcellation atlas has been created combining 93 cerebral (including cortical/subcortical structures) and 31 cerebellar labels. Each GM parcellation is reported as a node in the connectivity matrices. Two types of SC matrices are reported: a distance matrix containing the length of tracts connecting each pair of nodes and a weight matrix in which connections strengths (number of streamlines) are normalized by the maximum value per each subject. The time-course of BOLD signals has been extracted for each node. To perform brain dynamics simulations in multiple functional networks a subset of nodes and connections need to be extracted from whole-brain SC and FC matrices and used as an input for TVB.