Titolo della tesi: Developing a Prolonged Neurophysiological Monitoring Protocol to Detect Windows of Responsiveness in Minimally Conscious State Patients: foundation for a Passive Brain-Computer Interface System
Brain-Computer Interfaces (BCIs) are systems designed to record brain activity and translate it into artificial output to replace, restore, enhance, supplement, or improve natural brain functions (Wolpaw et al., 2020). Among BCI applications, it is possible to find the replacement of communication and control for individuals with severe disabilities (Riccio et al., 2016). BCIs can be classified as either active or passive systems. An active BCI requires the user to consciously modulate brain activity or to engage in specific tasks to generate output and control an external device. In contrast, a passive BCI decodes mental and emotional states from spontaneous brain activity without necessitating the user's active participation (Zander et al., 2009).
Disorders of consciousness (DoC) are clinical conditions resulting from severe acquired brain injuries, characterized by absent or diminished vigilance and awareness of self and the environment. These conditions include coma, vegetative state/unresponsive wakefulness syndrome (VS/UWS), and minimally conscious state (MCS), existing on a continuum where patients may transition sequentially through these states.
Patients with DoC are considered potential candidates for BCI interventions, supported by substantial evidence suggesting that they may possess covert awareness - characterized by a dissociation between severely limited motor abilities and preserved cognitive functioning (Schiff, 2015). The initial efforts to establish an alternative communication channel for these patients were conducted by Monti et al. (2010), based on prior findings by Owen et al. (2006). They employed a mental imagery fMRI paradigm to assess command-following abilities in 54 patients with DoC, identifying five patients capable of intentionally modulating their mental activity. Notably, one patient diagnosed with MCS demonstrated successful communication.
Despite these preliminary findings, the effectiveness of BCIs for communication with DoC patients remains uncertain (Spüler, 2019). Various factors may hinder the ability of DoC patients to utilize a BCI for communication, including sensory deficits, cognitive impairments, and fluctuations of responsiveness. Fluctuations of responsiveness are a hallmark of MCS diagnosis, which is defined by the presence of cognitively mediated behaviors that can be reliably differentiated from reflexive behaviors, despite occurring inconsistently over time. We propose that fluctuations of responsiveness give rise to what we term “Windows of Responsiveness” (WoR), defined as temporal windows during which an MCS patient exhibits higher level of responsiveness and potential interaction with their environment with respect “No Windows of Responsiveness” (NoWoR, i.e. interval time of low responsiveness). While fluctuations have been extensively examined from a behavioral perspective (Candelieri et al., 2011; Cortese et al., 2015; Wannez et al., 2017), they have received limited attention from a neurophysiological standpoint (Piarulli et al., 2016; Sitt et al., 2014).
The objective of the present thesis is to investigate fluctuations of responsiveness from a neurophysiological perspective to identify a range of indices that may describe windows of responsiveness. This neurophysiological investigation into fluctuations of responsiveness will contribute to developing a model to characterize responsiveness and will constitute the foundation for implementing a passive BCI to automatically detect WoR.
Chapter 1 presents a systematic review of the current state of BCI applications in patients with DoC, resulted in a publication in an international journal (Galiotta et al., 2022). The review aims to: i) describe the characteristics of BCI systems based on electroencephalography (EEG) developed for DoC patients, including the control signals employed, paradigm characteristics, classification algorithms, and applications; and ii) evaluate the performance of DoC patients using BCIs. The systematic review included twenty-seven studies. It was determined that the control signals utilized for BCI operation primarily consisted of the P300 component of the event-related potential (ERP), either in isolation or in conjunction with Steady-State Visual Evoked Potentials (hybrid systems), as well as sensorimotor rhythms. Potential applications of BCI in DoC patients include assessment, communication, prognosis, and rehabilitation, with a prevalence of the assessment application. Although BCIs appear to be promising tools in managing DoC patients - especially in supporting diagnostic and prognostic evaluations - the findings remain preliminary, with no definitive conclusions drawn, particularly regarding their utility and effectiveness for communication.
Chapter 2 details the implementation and validation of a prolonged monitoring protocol aimed at detecting fluctuations of responsiveness from a neurophysiological perspective. This study specifically addresses a clinical population defined by the presence of such fluctuations, i.e. MCS patients, alongside a control group of healthy subjects. The protocol involved four hours of monitoring primarily conducted in resting state, during which multiple biosignals, including EEG, were recorded. The monitoring sessions were punctuated by two active tasks: an auditory oddball task and a motor command task. The EEG responses to these tasks were analyzed to determine patients' levels of responsiveness at various points throughout the monitoring. This chapter focuses exclusively on the oddball task. To validate the protocol, I examined the P300 ERP component in response to the oddball task, investigating whether the variability in its amplitude and latency was greater in patients than in healthy participants. My findings supported this hypothesis, revealing higher variability in both amplitude and latency among patients compared to healthy controls. These results substantiate the suitability of the P300 ERP component for detecting fluctuations of responsiveness within our protocol. Following the validation, I established a criterium based on P300 amplitude and latency to classify each monitoring moment as either WoR or NoWoR for each subject.
Chapter 3 explores biosignals in the resting state immediately preceding each task presentation to evaluate significant differences in these indices between WoR and NoWoR. This analysis focuses on EEG and electrocardiographic (ECG) signals among those recorded. I computed a series of spectral EEG indices, as well as heart rate (HR) and heart rate variability (HRV). Initial comparisons were made between patients and healthy subjects, selecting only measures that exhibited significant differences between the two groups. Subsequently, I compared the selected indices between WoR and NoWoR to identify any significant differences. Notably, the Power Ratio Index (PRI) from the EEG and the HR from the ECG were found to be significantly higher in NoWoR than WoR, while the HRV was elevated in WoR compared to NoWoR. This investigation into EEG and ECG indices aims to contribute to the development of a neurophysiological model of responsiveness, which could facilitate the implementation of a passive BCI for the detection of the presence of WoR.