RETRIEVING THE STRUCTURE OF A STOCHASTIC SEQUENCE OF TMS PULSES DRIVEN BY A CONTEXT TREE MODEL: A TMS/EEG APPROACH

• Grant number: 25/07274-6
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Start date: June 01, 2025
• End date: May 31, 2026
• Field of knowledge: Interdisciplinary Subjects
• Principal Investigator: Oswaldo Baffa Filho
• Grantee: Victor Hugo Fernandes de Moraes Faustino
• Host Institution: Faculdade de Filosofia, Ciências e Letras de Ribeirão Preto (FFCLRP). Universidade de São Paulo (USP). Ribeirão Preto , SP, Brazil
• Associated research grant: 13/07699-0 - Research, Innovation and Dissemination Center for Neuromathematics - NeuroMat, AP.CEPID

Abstract

When applied to the primary motor cortex (M1), Transcranial Magnetic Stimulation (TMS) pulses evokes electrical signals in peripheral muscles called motor evoked potentials (MEPs), which represent a measure of corticospinal excitability (CSE). It is known that the amplitude of MEPs is dependent on the intensity of stimulation and the interval between stimuli. The monitoring of electroencephalographic (EEG) activity during the application of TMS pulses (TMS-EEG) allows the acquisition of so-called TMS-evoked potentials (TEPs). TEPs are EEG signals with positive and negative deflections that last up to 300 ms after the TMS pulse and which are also modified by the intensity of the stimulation. TMS-EEG protocols can be used to detect longitudinal changes in the state of cortical circuits, being an interesting method for investigating cortical dynamics. The question is whether M1 holds a memory of a sequence of TMS pulses longer than n-1, with n corresponding to the present pulse stimulation. If so, it should be possible to recover in the MEPs and TEPs response signatures of a sequence of TMS pulses applied in the M1. This approach is inspired by the work of Duarte et al. (2019) and Hernández et al. (2021), whose subjects were subjected to a sequence of auditory stimuli while their EEG signals were recorded. They used a stochastic chain with memory of variable length to model the dependence from the past characterizing the sequence of auditory stimuli. The sequence of sounds was created inspired by the work of Rissanen (1983), who proposed that the amount of past information needed to predict the next event can vary. The relevant past information was called a context, with the set of all contexts being represented by a rooted and labeled-oriented tree. This structure, associated with a probability matrix, determines, for each context, which can be the next element and is called a Probabilistic Context Tree Model (PCTM). Duarte et al. (2019) and Hernández et al. (2021) showed that it was possible to retrieve PCTMs modeling the EEG signals. The goal of this project is to retrieve from the MEP and TEPs responses the context tree used to produce the sequence of stimuli. If so, it will indicate that the brain acquires a memory of the sequence of pulses. Twenty right-handed healthy volunteers will be recruited after the study is approved by the Research Ethics Committee of USP's Ribeirão Preto School of Philosophy, Sciences and Letters. The TMS pulse sequence will be created using a PCTM, using a ternary context tree and the matrix of transition probabilities. The symbols 0, 1 and 2 will represent different stimulus intensities. The intensities used will be 100%, 110%, and 120% of the rMT. The rMT is defined as the minimum stimulation intensity required for MEPs of 50 uV in 5 out of 10 pulses. CSE will be measured by applying single TMS pulses by means of an 8-shaped coil, poitioned above left M1, connected to a MagPro30 stimulator (MagVenture A/S, Denmark). Coil positioning will be guided using a robotic (Han's cobot) coil positioning connected to the software InVesalius Navigator developed at USP/RP. The optimal position (hot spot) for eliciting MEPs from the right first dorsal interosseous (FDI) muscle will be identified. EMG activity will be recorded from the FDI, the opponens pollicis (OP), and the flexor digitorum superficialis (FDS) via surface electrodes (CED 1902, Cambridge, UK) placed on the muscle bellies. EMG (Signal software, CED 1401, UK) and EEG (64 channels, Bittium, NeurOne, Bittium Corporation, Finland). Peak-to-peak MEP amplitudes (µV) will be measured off-line using the Signal Hunter software (Souza et al., 2015) and to investigate the effect of the sequence in the MEPs, we will use the software Goalkeeper Lab, recently developed by Passos PRC (github.com/PauloRC20/GoalkeeperLab). EEG signals will be processed based on the methodology described in Hernández et al. (2021).