Functional connectivity (FC) studies have historically relied on static measures over periods of minutes. Recent works have focused on transient patterns of brain activity in dynamic FC (dFC) analyses. Yet several open questions remain regarding the correct time sampling of dFC and the possible intermittent emergence of patterns. To address this issue, we applied a non-stationary seed-based dynamic method, named Co-activation patterns (CAP) analysis aimed at investigating the dFC of the sensory-motor network (SMN) extracted by the independent component analysis (ICA), thus combining the two approaches. This study was conducted at the CADiTeR Research unit, situated within the Don Carlo Gnocchi Foundation – IRCCS Centro Santa Maria Nascente, located in Milan. The subject dataset utilized in this analysis was composed of 42 subjects, including 24 Parkinson’s disease (PD) patients (13 males, 11 females; mean age 69 ± 4.18 years) and 18 age-matched healthy control (HC) subjects (7 males, 11 females; mean age 59.3 ± 5.57 years). MRI data were acquired according to a multi-echo protocol by a 3T Siemens Prisma scanner (Siemens, Erlangen, Germany) equipped with a 64-channel head coil. Data analysis was performed using the TbCaps toolbox and the FMRIB’s Software Library (FSL). Group ICA analysis including PD and HC subjects gave 15 resting state networks (RSN) consistent with the literature, including the addressed SMN. However, the dual regression analysis extracting individual patterns from the group ICA provided no significant contrast between PD and HC. CAP analysis based on the SMN signal extracted for each subject by the dual regression was tuned setting the coactivation threshold to the 85th percentile (i.e., 15% of total duration), which already captured the average pattern. A root mean square (RMS) analysis confirmed that major activity was concentrated above the 85th percentile. The dynamic CAP analysis was conducted, and k-means clustering (k=3) was used to determine transient activation patterns of the SMN. The occurrence frequency of different dynamic CAP brain states and the state transitions within each CAP were then compared between the HC and PD groups. Dynamic brain configurations characterized by co-activation of the SMN with the visual network, ventral stream, and task-positive networks appeared less frequently and less resilient in PD compared to HC. However, the co-activation of the SMN with the salience, cerebellum, and auditory network showed more frequent and resilient in PD compared to HC. This study highlights the utility of time-varying approaches for studying altered SMN function in prevalent neurodevelopmental disorders. We hypothesize that changes in SMN dynamics in PD could be a mechanism for the unstable behaviors frequently seen in individuals with this disorder. In conclusion, the proposed protocol merging ICA and CAPs was shown to be sufficiently robust for the effective evaluation of alterations occurring in PD. The proposed ICA and CAP combination, possibly based on other RSNs, might provide a better understanding of the brain’s pathophysiological mechanisms by capturing the disruption of connectivity strength dynamics in several neurodegenerative diseases.
Gli studi sulla connettività funzionale (CF) si sono storicamente basati su misure statiche. Studi recenti hanno focalizzato l’attenzione sullo studio di attività transitorie dell’attività cerebrale proponendo lo studio della CF dinamica (CFd). Tuttavia rimangono ancora diverse questioni aperte riguardo al corretto campionamento temporale della CFd e al possibile emergere di pattern intermittenti. Per affrontare questo problema, in questo studio abbiamo applicato un modello “seed-based” di connetività dinamica non stazionaria, chiamato analisi dei pattern di co-attivazione (CAP), finalizzata a studiare la CFd della rete (“network”) sensori-motoria (SMN) estratta dall'analisi delle componenti indipendenti (ICA), combinando così i due approcci. Questo studio è stato condotto presso l'unità di ricerca CADiTeR, situata all'interno della Fondazione Don Carlo Gnocchi - IRCCS Centro Santa Maria Nascente, con sede a Milano. Il set di dati utilizzato in questa analisi era di 42 soggetti, tra cui 24 pazienti con malattia di Parkinson (PD) (13 maschi, 11 femmine; età media 69 ± 4,18 anni) e 18 soggetti di controllo sani (HC) (7 maschi e 11 femmine; età media 59,3 ± 5,57 anni). I dati di RM funzionale sono stati acquisiti su scanner 3T Siemens Prisma (Siemens, Erlangen, Germania) dotato di una bobina a 64 canali. L'analisi dei dati è stata effettuata utilizzando il toolbox TbCaps e la Software Library (FSL) di FMRIB. Dall'analisi dell'ICA di gruppo, comprendente sia i soggetti PD che HC, sono state derivate e classificate 15 “resting-state networks” (RSN) coerenti con quelle descritte in letteratura, inclusa la network sensori-motoria (SMN) di interesse. A seguito di analisi standard comprensiva di regressione duale e confronto tra networks derivate dai due gruppi non sono emerse differenze significative tra HC e PD. L'analisi dei CAP basata sul segnale estratto per ogni soggetto dalla SMN tramite la regressione duale è stata condotta impostando la soglia di coattivazione 85° percentile (considerando cioè il 15% della durata totale dell’acquisizione). Un'analisi del valore quadratico medio (RMS) ha consentito di confermare che l'attività principale era concentrata sopra la soglia scelta (85° percentile). È stata condotta l'analisi dinamica per la scelta ottimale del numero di cluster con dimensionalità k=3 al fine di determinare i modelli di co-attivazione transitoria riferiti alla SMN. La frequenza di occorrenza dei diversi stati cerebrali dinamici, CAP, e le transizioni di stato all'interno di ogni CAP sono stati poi confrontati tra i gruppi HC e PD. Le configurazioni cerebrali dinamiche caratterizzate dalla co-attivazione del SMN con la rete visiva, il ‘ventral stream’ e le ‘task positive network’ sono apparse con frequenza minore e minor resilienza nel gruppo dei PD rispetto al gruppo dei HC. Tuttavia, la co-attivazione del SMN con la ‘salient network’, il cervelletto e la network uditiva hanno mostrato invece maggiore frequenza e resilienza nel gruppo dei PD rispetto al gruppo dei HC. Questo studio evidenzia l'utilità degli approcci dinamici per lo studio delle alterazioni di funzionalità della SMN disturbi neurologici quali la malattia di Parkinson. Ipotizziamo che i cambiamenti osservati nella dinamica della SMN nei PD potrebbero essere legati alla sintomatologia osservata in individui affetti da tale disturbo. In conclusione, il protocollo proposto che unisce gli approcci di analisi ICA e CAP è stato dimostrato sufficientemente robusto per una efficace valutazione delle alterazioni che si verificano nel PD, in accordo a quanto già osservato in letteratura. La combinazione proposta di ICA e CAP, applicata ad altre RSNs, dimostratasi utile nel catturare le alterazioni di connettività funzionale dinamica, potrebbe fornire una migliore comprensione dei meccanismi patofisiologici dell’encefalo in diverse malattie neurodegenerative.
Resting state contemporaneous activation patterns in the sensory motor network in Parkinson's disease
SRAJ, YEHYA
2023/2024
Abstract
Functional connectivity (FC) studies have historically relied on static measures over periods of minutes. Recent works have focused on transient patterns of brain activity in dynamic FC (dFC) analyses. Yet several open questions remain regarding the correct time sampling of dFC and the possible intermittent emergence of patterns. To address this issue, we applied a non-stationary seed-based dynamic method, named Co-activation patterns (CAP) analysis aimed at investigating the dFC of the sensory-motor network (SMN) extracted by the independent component analysis (ICA), thus combining the two approaches. This study was conducted at the CADiTeR Research unit, situated within the Don Carlo Gnocchi Foundation – IRCCS Centro Santa Maria Nascente, located in Milan. The subject dataset utilized in this analysis was composed of 42 subjects, including 24 Parkinson’s disease (PD) patients (13 males, 11 females; mean age 69 ± 4.18 years) and 18 age-matched healthy control (HC) subjects (7 males, 11 females; mean age 59.3 ± 5.57 years). MRI data were acquired according to a multi-echo protocol by a 3T Siemens Prisma scanner (Siemens, Erlangen, Germany) equipped with a 64-channel head coil. Data analysis was performed using the TbCaps toolbox and the FMRIB’s Software Library (FSL). Group ICA analysis including PD and HC subjects gave 15 resting state networks (RSN) consistent with the literature, including the addressed SMN. However, the dual regression analysis extracting individual patterns from the group ICA provided no significant contrast between PD and HC. CAP analysis based on the SMN signal extracted for each subject by the dual regression was tuned setting the coactivation threshold to the 85th percentile (i.e., 15% of total duration), which already captured the average pattern. A root mean square (RMS) analysis confirmed that major activity was concentrated above the 85th percentile. The dynamic CAP analysis was conducted, and k-means clustering (k=3) was used to determine transient activation patterns of the SMN. The occurrence frequency of different dynamic CAP brain states and the state transitions within each CAP were then compared between the HC and PD groups. Dynamic brain configurations characterized by co-activation of the SMN with the visual network, ventral stream, and task-positive networks appeared less frequently and less resilient in PD compared to HC. However, the co-activation of the SMN with the salience, cerebellum, and auditory network showed more frequent and resilient in PD compared to HC. This study highlights the utility of time-varying approaches for studying altered SMN function in prevalent neurodevelopmental disorders. We hypothesize that changes in SMN dynamics in PD could be a mechanism for the unstable behaviors frequently seen in individuals with this disorder. In conclusion, the proposed protocol merging ICA and CAPs was shown to be sufficiently robust for the effective evaluation of alterations occurring in PD. The proposed ICA and CAP combination, possibly based on other RSNs, might provide a better understanding of the brain’s pathophysiological mechanisms by capturing the disruption of connectivity strength dynamics in several neurodegenerative diseases.| File | Dimensione | Formato | |
|---|---|---|---|
|
2024_04_Yehya_Sraj_10811285_Executive Summary.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Master's Tesi Executive Summary_Yehya_Sraj
Dimensione
532.98 kB
Formato
Adobe PDF
|
532.98 kB | Adobe PDF | Visualizza/Apri |
|
2024_04_Yehya_Sraj_10811285_Tesi.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Master's Tesi _Yehya_Sraj
Dimensione
4.28 MB
Formato
Adobe PDF
|
4.28 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/218418