This work explored the potential application of functional Near-Infrared Spectroscopy (fNIRS) in clinical research by integration with MRI-derived cortical anatomy. fNIRS is a non-invasive neuroimaging technique that uses NIR-light to assess cortical brain activity. This technique is eliciting a growing interest in neuroimaging, but, to date, limitations are still hindering its applications in clinical research due to the partial lack of anatomical information and standardized protocols for data analysis and interpretation. This thesis evaluated a data-driven image reconstruction algorithm based on the functional task itself for mapping and statistically assess cortical activity from fNIRS measurements with traditional source-detector distances. The implemented image reconstruction pipeline consisted in co-registering the source-detector configuration to head surface, modelling light transport through Monte Carlo simulations into an atlas (Colin27), and finally solving the inverse problem of image reconstruction using Gauss-Markov solution with Tikhonov regularization and L-curve method. We evaluated this method over simulated data, obtaining reliable reconstructions in terms of SNR, sensitivity, and localization error. Then, this pipeline was applied on an experimental fNIRS dataset recorded on 15 healthy young adults according to a hand-grasping task, demonstrating the correct localization of the activated areas in both channels and cortical measurements. Bland-Altman analysis verified the agreement between channel-wise and averaged cortical t-values across all experimental conditions. Mapping cortical activations over atlas-based or subject-specific anatomies favors the process of interpreting fNIRS outcomes with a more reliable spatial resolution and by simultaneously exploiting the advantages of portability that this technique offers. In summary, the proposed approach represents a promising development for the functional assessment of cortical activation. Future development will require translation over different clinical populations, especially to stroke and chronic neurological diseases, functional tasks and subject-specific anatomies.
Questo lavoro ha esplorato l’applicazione della spettroscopia funzionale nel vicino infrarosso (fNIRS) nella ricerca clinica mediante l'integrazione con anatomia corticale derivata da risonanza magnetica. fNIRS è una tecnica di neuroimaging non invasiva che utilizza la luce NIR per valutare l'attività cerebrale corticale. Questa tecnica sta suscitando un crescente interesse nel neuroimaging, ma, ad oggi, i suoi limiti ne stanno ancora ostacolando l’applicazione nel campo clinico a causa della mancanza di informazioni anatomiche e di protocolli standardizzati per l'analisi e l'interpretazione dei dati. Questa tesi ha valutato un algoritmo di ricostruzione di immagine data-driven basato sul compito funzionale stesso per la mappatura e la valutazione statistica dell'attività corticale dalle tradizionali misurazioni fNIRS. La pipeline di ricostruzione dell'immagine implementata consiste nel co-registrare la configurazione sorgente-rivelatore sulla superficie della testa, modellare il trasporto della luce attraverso simulazioni Monte Carlo in un atlante (Colin27) e infine risolvere il problema inverso usando la soluzione di Gauss-Markov con la regolarizzazione di Tikhonov e il metodo della L-curve. Abbiamo valutato questo metodo su dati simulati, ottenendo ricostruzioni affidabili in termini di SNR, sensitività ed errore di localizzazione. Quindi, questa pipeline è stata applicata su dati sperimentali acquisiti su 15 giovani adulti sani durante un compito motorio della mano, dimostrando la corretta localizzazione delle aree attivate in entrambi i canali e le misurazioni corticali. L'analisi di Bland-Altman ha verificato la concordanza tra i t-value corticali medi e per canale in tutte le condizioni sperimentali. La mappatura delle attivazioni corticali su anatomie favorisce il processo di interpretazione dei risultati fNIRS con una risoluzione spaziale più affidabile, sfruttando i vantaggi di portabilità della fNIRS. L'approccio proposto rappresenta uno sviluppo promettente per la valutazione funzionale dell'attivazione corticale. Sviluppi futuri richiederanno l’applicazione su diverse popolazioni cliniche, compiti funzionali e anatomie soggetto-specifiche.
Realistic fNIRS representation on MRI derived cortical anatomy : a preliminary group study
Baldini, Benedetta
2019/2020
Abstract
This work explored the potential application of functional Near-Infrared Spectroscopy (fNIRS) in clinical research by integration with MRI-derived cortical anatomy. fNIRS is a non-invasive neuroimaging technique that uses NIR-light to assess cortical brain activity. This technique is eliciting a growing interest in neuroimaging, but, to date, limitations are still hindering its applications in clinical research due to the partial lack of anatomical information and standardized protocols for data analysis and interpretation. This thesis evaluated a data-driven image reconstruction algorithm based on the functional task itself for mapping and statistically assess cortical activity from fNIRS measurements with traditional source-detector distances. The implemented image reconstruction pipeline consisted in co-registering the source-detector configuration to head surface, modelling light transport through Monte Carlo simulations into an atlas (Colin27), and finally solving the inverse problem of image reconstruction using Gauss-Markov solution with Tikhonov regularization and L-curve method. We evaluated this method over simulated data, obtaining reliable reconstructions in terms of SNR, sensitivity, and localization error. Then, this pipeline was applied on an experimental fNIRS dataset recorded on 15 healthy young adults according to a hand-grasping task, demonstrating the correct localization of the activated areas in both channels and cortical measurements. Bland-Altman analysis verified the agreement between channel-wise and averaged cortical t-values across all experimental conditions. Mapping cortical activations over atlas-based or subject-specific anatomies favors the process of interpreting fNIRS outcomes with a more reliable spatial resolution and by simultaneously exploiting the advantages of portability that this technique offers. In summary, the proposed approach represents a promising development for the functional assessment of cortical activation. Future development will require translation over different clinical populations, especially to stroke and chronic neurological diseases, functional tasks and subject-specific anatomies.| File | Dimensione | Formato | |
|---|---|---|---|
|
2021_04_Baldini.pdf
accessibile in internet per tutti
Descrizione: Testo della tesi
Dimensione
5.88 MB
Formato
Adobe PDF
|
5.88 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/10589/175294