The loss of voluntary muscle control, as seen in locked-in syndrome and advanced stages of amyotrophic lateral sclerosis, deprives patients of the ability to communicate while leaving cognitive functions intact. This creates a state of communicative isolation with severe repercussions on the quality of life. Brain-computer interfaces (BCIs) based on electroencephalography represent one of the most promising solutions to restore a communication channel between the central nervous system and the external environment in these patients, translating brain activity into commands without involving the neuromuscular pathway. This thesis work analyzes non-invasive BCIs based on electroencephalographic (EEG) signals for communication support (speller systems), starting from signal acquisition protocols (electrode placement according to the 10-20 standard, wet and dry electrodes, referential and bipolar montages) and subsequent processing stages (amplification, filtering, artifact removal, and extraction of temporal and spectral features). The two main paradigms are then examined separately: the P300 speller, which exploits the cognitive response to rare stimuli according to the oddball paradigm, and whose different presentation modalities (row-column, checkerboard, and region-based) are analyzed; and the SSVEP speller, which leverages the frequency response of the visual cortex to flickering stimuli, analyzed in its single-stage, hierarchical, and row-column architectures. For each paradigm, advantages, limitations, and application challenges are discussed, such as signal non-stationarity, inter-subject variability, and the tradeoff between speed and accuracy. Finally, a brief comparative analysis of both approaches and the hybrid spellers that integrate them is proposed, highlighting how the latter—by combining the robustness of the P300 with the speed of the SSVEP—achieve higher information transfer rates compared to single-paradigm systems. The work concludes with a discussion of the main future research directions: asynchronous control, adaptive personalization of stimulation parameters, and the extension of these systems to different languages and clinical populations, emphasizing that technical progress in these systems must always be accompanied by attention to the psychological well-being and dignity of the user.
La perdita del controllo muscolare volontario, come nella sindrome locked-in e nelle fasi avanzate della sclerosi laterale amiotrofica, priva il paziente della possibilità di comunicare, pur lasciando intatte le funzioni cognitive, e genera un isolamento comunicativo con gravi ripercussioni sulla qualità della vita. Le interfacce cervello-computer (BCI) basate su elettroencefalogramma rappresentano una delle soluzioni più promettenti per ripristinare in questi pazienti un canale comunicativo tra il sistema nervoso centrale e l’ambiente esterno, traducendo l'attività cerebrale in comandi senza il coinvolgimento della via neuromuscolare. Il presente lavoro di tesi analizza le BCI non invasive basate su segnali elettroencefalografici (EEG) per il supporto alla comunicazione (sistemi speller), a partire dai protocolli di acquisizione del segnale (posizionamento degli elettrodi secondo lo standard 10-20, elettrodi wet e dry, montaggi referenziali e bipolari) e dalle successive fasi di elaborazione (amplificazione, filtraggio, rimozione degli artefatti, estrazione delle caratteristiche temporali e spettrali). Vengono quindi esaminati separatamente i due paradigmi principali: lo speller P300, che sfrutta la risposta cognitiva a stimoli rari secondo il paradigma oddball, e di cui vengono analizzate le diverse modalità di presentazione (riga-colonna, a scacchiera, basata su regioni); e lo speller SSVEP, che sfrutta la risposta in frequenza della corteccia visiva a stimoli flickering, ed è analizzato nelle sue architetture a uno stadio, gerarchiche e a riga-colonna. Per ciascun paradigma si discutono vantaggi, limiti e sfide applicative, quali la non-stazionarietà del segnale, la variabilità inter-soggetto e il tradeoff tra velocità e accuratezza. Infine, viene proposta una breve analisi comparativa dei due approcci e degli speller ibridi che li integrano, evidenziando come questi ultimi, combinando la robustezza del P300 con la velocità della SSVEP, consentano di raggiungere velocità di trasferimento dell'informazione superiori rispetto ai sistemi a singolo paradigma. Il lavoro si conclude con la discussione delle principali direzioni di ricerca future: il controllo asincrono, la personalizzazione adattiva dei parametri di stimolazione e l'estensione dei sistemi a lingue e popolazioni cliniche diverse, sottolineando come il progresso tecnico di questi sistemi debba sempre accompagnarsi all'attenzione per il benessere psicologico e la dignità dell'utente.
Interfacce Cervello-Computer basate su EEG: studio dei paradigmi P300 e SSVEP per l’assistenza alla comunicazione in persone con gravi disabilità motorie
MARULLI, AGNESE
2025/2026
Abstract
The loss of voluntary muscle control, as seen in locked-in syndrome and advanced stages of amyotrophic lateral sclerosis, deprives patients of the ability to communicate while leaving cognitive functions intact. This creates a state of communicative isolation with severe repercussions on the quality of life. Brain-computer interfaces (BCIs) based on electroencephalography represent one of the most promising solutions to restore a communication channel between the central nervous system and the external environment in these patients, translating brain activity into commands without involving the neuromuscular pathway. This thesis work analyzes non-invasive BCIs based on electroencephalographic (EEG) signals for communication support (speller systems), starting from signal acquisition protocols (electrode placement according to the 10-20 standard, wet and dry electrodes, referential and bipolar montages) and subsequent processing stages (amplification, filtering, artifact removal, and extraction of temporal and spectral features). The two main paradigms are then examined separately: the P300 speller, which exploits the cognitive response to rare stimuli according to the oddball paradigm, and whose different presentation modalities (row-column, checkerboard, and region-based) are analyzed; and the SSVEP speller, which leverages the frequency response of the visual cortex to flickering stimuli, analyzed in its single-stage, hierarchical, and row-column architectures. For each paradigm, advantages, limitations, and application challenges are discussed, such as signal non-stationarity, inter-subject variability, and the tradeoff between speed and accuracy. Finally, a brief comparative analysis of both approaches and the hybrid spellers that integrate them is proposed, highlighting how the latter—by combining the robustness of the P300 with the speed of the SSVEP—achieve higher information transfer rates compared to single-paradigm systems. The work concludes with a discussion of the main future research directions: asynchronous control, adaptive personalization of stimulation parameters, and the extension of these systems to different languages and clinical populations, emphasizing that technical progress in these systems must always be accompanied by attention to the psychological well-being and dignity of the user.| File | Dimensione | Formato | |
|---|---|---|---|
|
Marulli_Agnese.pdf
accesso aperto
Dimensione
966.01 kB
Formato
Adobe PDF
|
966.01 kB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/110892