Research

Human models of stroke

Our research combines a large-scale multimodal and longitudinal assessment of stroke patients—including structural and functional MRI, high-density EEG, eye-tracking, and comprehensive behavioural evaluations. We investigate the mechanisms of damage, recovery, and outcome prediction after stroke. Specifically, we test whether the distribution of slow waves, both perilesional and remote, reflects patterns of structural connectivity damage or broader functional connectivity disruptions. We assess how these relationships predict behavioural impairment and recovery across motor and cognitive domains.

Mouse models of stroke

Led by Dr. Manuela Allegra, we study how stroke alters brain activity both locally, in areas surrounding the lesion, and across larger brain networks. We use an adult mouse model of stroke, inducing an ischemic lesion in the somatosensory cortices. Combining wide-field brain imaging (mesoscale Ca²⁺ imaging), electrophysiological recordings, and behavioral tests we aim at understanding how disrupted neuronal activity relates to recovery. By linking changes in brain activity with motor and cognitive performance, we want to elucidate how focal lesions trigger widespread network disruptions.

Specifically, we assess post-lesional slow waves from two main perspectives:

Observational

We use both non-invasive (scalp) and invasive (stereotactic implanted) electroencephalographic (EEG) recordings to characterise the presence and the spatial extent post-lesional sleep-like slow waves in neurological patients affected by focal vascular and traumatic brain injuries. We investigate whether the presence of sleep-like slow waves is predicted by anatomical and functional connectivity with respect to the lesion site and assess their role in predicting behavioural deficits and their recovery across the different stages of disease.

Causal

We test the idea that the generation and propagation of post-lesional slow waves in perilesional and remote connected cortical regions are associated with an impairment of cortico-cortical information transmission and complexity, as measured by the dynamics of the EEG response to a direct perturbation with Transcranial Magnetic Stimulation (TMS).

We assess post-lesional slow waves from two main perspectives:

Experimental Approaches

In Vitro Studies: We use cortical brain slices in desynchronized (wake-like) states to understand the percolation of slow waves, rhythmic activity associated with unconscious states. We test how different conditions (altered excitation/inhibition balance, temperature shifts, changes in potassium concentration, pH, inflammatory markers, and NMDA signaling) affect the vulnerability of the network and the spread of pathological activity.

In Vivo Mouse Models of Brain Lesion: Using radio-frequency thermal coagulation (RFTC) or traumatic brain injury (TBI), we create focal cortical lesions in mice to study how injuries alter structural and functional brain connectivity.

Computational Modeling

Our modelling team develops computational simulations of cortical networks to replicate experimental findings and explore underlying mechanisms. These in silico models allow us to test how lesions and changes in network parameters (such as excitation/inhibition balance or connectivity patterns) affect slow wave propagation and cortical synchrony.

Our computational work is structured into several lines of research:

Sleep rhythms and stroke recovery

We study how slow-wave sleep rhythms influence recovery after stroke. Using whole-brain computer models, we examine how these rhythms can sometimes reinforce maladaptive activity after injury, and how guiding them may promote recovery.

Mapping brain structure and function over time

Stroke alters both the brain’s structural connections and functional communication. We model these changes over time to understand how recovery occurs, whether through repairing damaged pathways or by recruiting alternative networks.

Brain computability

A healthy brain flexibly adapts to challenges, but stroke reduces this capacity. We use computational models to quantify this flexibility, which we call brain computability, and to assess how recovery and interventions can restore it.

Guiding brain stimulation with models

Non-invasive brain stimulation, such as transcranial electrical stimulation (tES), shows promise for stroke recovery, but decisions about where, when, and how to stimulate are still largely trial and error. Our models simulate how stimulation spreads through damaged brain networks, enabling more precise and effective strategies to restore lost functions.