Biology subjects

de Alteriis, G.

Publications and source records attributed to de Alteriis, G..

3 recordsLinked to original sources

bioRxiv2024

DySCo: a general framework for dynamic Functional Connectivity

1A crucial challenge in neuroscience involves characterising brain dynamics from high-dimensional brain recordings. Dynamic Functional Connectivity (dFC) is an analysis paradigm that aims to address this challenge. dFC consists of a time-varying matrix (dFC matrix) expressing how pairwise interactions across brain areas change with time. However, the main dFC approaches have been developed and applied mostly empirically, lacking a unifying theoretical framework, a general interpretation, and a common set of measures to quantify the dFC matrices properties. Moreover, the dFC field has been lacking ad-hoc algorithms to compute and process the matrices efficiently. This has prevented the field to show its full potential with high-dimensional datasets and/or real time applications. With this paper, we introduce the Dynamic Symmetric Connectivity Matrix analysis framework (DySCo), with its associated repository. DySCo is a unifying approach that allows the study of brain signals at different spatio-temporal scales, down to voxel level, that is computationally ultrafast. DySCo unifies in a single theoretical framework the most employed dFC matrices, which share a common mathematical structure. Doing so it allows: 1) A new interpretation of dFC that further justifies its use to capture the spatiotemporal patterns of data interactions in a form that is easily translatable across different imaging modalities. 2) The introduction of the the Recurrence Matrix EVD to compute and store the eigenvectors and eigenvalues of all types of dFC matrices in an efficent manner that is orders of magnitude faster than naive algorithms, and without loss of information. 3) To simply define quantities of interest for the dynamic analyses such as: the amount of connectivity (norm of a matrix) the similarity between matrices, their informational complexity. The methodology developed here is validated on both a synthetic dataset and a rest/N-back task experimental paradigm - the fMRI Human Connectome Project dataset. We demonstrate that all the measures proposed are highly sensitive to changes in brain configurations. To illustrate the computational efficiency of the DySCo toolbox, we perform the analysis at the voxel-level, a computationally very demanding task which is easily afforded by the RMEVD algorithm.

neuroscience↗

bioRxiv2023

EiDA: A lossless approach for the dynamic analysis of connectivity patterns in signals; application to resting state fMRI of a model of ageing

AO_SCPLOWBSTRACTC_SCPLOWDynamic Functional Connectivity (dFC) is the study of the dynamical patterns emerging from brain function. We introduce EiDA (Eigenvector Dynamic Analysis), a method that losslessly reduces the dimension of the instantaneous connectivity patterns of a time series to characterise dynamic Functional Connectivity (dFC). We apply EiDA to investigate the signatures of ageing on brain network dynamics in a longitudinal dataset of resting-state fMRI in ageing rats. Previous dFC approaches have relied on the concept of the instantaneous phase of signals, computing the instantaneous phase-locking matrix (iPL) and its eigenvector decomposition. In this work, we fully characterise the eigenstructure of the iPL analytically, which provides a 1000 fold speed up in dFC computations. The analytical characterization of the iPL matrix allows us to introduce two methods for its dynamic analysis. 1) Discrete EiDA identifies a discrete set of phase locking modes using k-means clustering on the decomposed iPL matrices. 2) Continuous EiDA provides a 2-dimensional "position" and "speed" embedding of the matrix; here, dFC is conceived as a continuous exploration of this 2-D space rather than assuming the existence of discrete brain states. We apply EiDA to a cohort of 48 rats that underwent functional magnetic resonance imaging (fMRI) at four stages during the course of their lifetime. Using Continuous and Discrete EiDA we found that brain phase-locking patterns become less intense and less structured with ageing. Using information theory and metastability measures derived from the properties of the iPL matrix, we see that ageing reduces the available functional repertoire postulated to be responsible for flexible cognitive functions and overt behaviours, and reduces the area explored in the embedding space.

neuroscience↗

bioRxiv2021

Euler method can outperform more complex ODE solvers in the numerical implementation of the Izhikevich artificial Spiking Neuron Model given the allocated FLOPS

The Izhikevich artificial spiking neuron model is among the most employed models in neuromorphic engineering and computational neuroscience, due to the affordable computational effort to discretize it and its biological plausibility. It has been adopted also for applications with limited computational resources in embedded systems. It is important therefore to realize a compromise between error and computational expense to solve numerically the models equations. Here we investigate the effects of discretization and we study the solver that realizes the best compromise between accuracy and computational cost, given an available amount of Floating Point Operations per Second (FLOPS). We considered three fixed-step solvers for Ordinary Differential Equations (ODE), commonly used in computational neuroscience: Euler method, the Runge-Kutta 2 method and the Runge-Kutta 4 method. To quantify the error produced by the solvers, we used the Victor Purpura spike train Distance from an ideal solution of the ODE. Counterintuitively, we found that simple methods such as Euler and Runge Kutta 2 can outperform more complex ones (i.e. Runge Kutta 4) in the numerical solution of the Izhikevich model if the same FLOPS are allocated in the comparison. Moreover, we quantified the neuron rest time (with input under threshold resulting in no output spikes) necessary for the numerical solution to converge to the ideal solution and therefore to cancel the error accumulated during the spike train; in this analysis we found that the required rest time is independent from the firing rate and the spike train duration. Our results can generalize in a straightforward manner to other spiking neuron models and provide a systematic analysis of fixed step neural ODE solvers towards an accuracy-computational cost tradeoff.

bioengineering↗