Multimodal near-infrared spectroscopy and electroencephalography recordings of the human brain

Abstract

Near-infrared light penetrates biological tissue relatively deeply (a few cm deep) compared to other light. Near infrared spectroscopy (NIRS) non-invasively assesses tissue
properties by measuring absorption and scattering. From the absorption, concentrations of chromophores can be determined, i.e. the concentrations of oxyhemoglobin ([O2Hb]), deoxyhemoglobin ([HHb]), total hemoglobin ([tHb]), water, lipids and tissue oxygen saturation
([O2Hb]/[tHb]). Since near-infrared light also penetrates the skull, NIRS can be employed to study the brain, i.e. absolute values and changes of the oxygenation and perfusion
of the brain in response to different situations such as e.g. sleep, sleep disordered breathing and sensory stimulation. The first aim of this thesis was to study brain activity in adults and neonates and to optimize the methodology. The second aim was to study the
oxygenation and perfusion of the brain during sleep and in particular changes associated with sleep disordered breathing and periodic leg movements.
Brain activity is associated with localized hemodynamic changes in blood flow by neurovascular
coupling, which leads to changes in [O2Hb] and [HHb]. This hemodynamic response can be reliably detected in groups of subjects and consequently NIRS is an excellent
tool for research. The reproducibility within an individual is still too low for clinical application.
Besides the hemodynamic response, NIRS has the potential to directly detect changes in the optical properties of neurons. However, it is unclear whether this optical
neuronal signal can be reliably measured non-invasively.
The aim of the first part of this thesis was to increase the sensitivity and repeatability of functional NIRS and to combine NIRS with electroencephalography (EEG) in healthy
adults and neonates. EEG is an established diagnostic method. In a first study, combined NIRS and EEG recordings were obtained from 15 healthy adult subjects during visual
stimulation. Each subject was measured three times on three different days. For the hemodynamic response, the results showed that reproducibility in single subjects was still
too low for clinical application. Responses were also often absent in the EEG. This suggests that the brain was not always activated. Despite low noise levels due to optimized filtering (P20 = 4.8 · 10−5 % in change of optical density), no significant optical neuronal signal was
detected.
A similar study was carried out in 15 healthy term newborns. Since the tissue of newborn
infants is more transparent and their skull is thinner in comparison to adults, a higher sensitivity for detecting hemodynamic responses and optical neuronal signals was expected. The newborn infants were visually stimulated by light flashes during sleep. During initial measurements only few hemodynamic responses were detected and sensitivity of EEG was low too. Therefore, in a first step possible reasons for the low sensitivity were investigated
in a methodological pilot study. Due to an improved stimulation device, activation was achieved in 63.3% of NIRS measurements (hemodynamic response) and in 96.3% of EEG
measurements.
To assess the reproducibility for hemodynamic responses, optical neuronal signals and EEG, a further study was conducted in 14 healthy term subjects. This is the first time that the reproducibility of NIRS and the optical neuronal signal were studied in newborns.
Again subjects underwent visual flash stimulation during sleep and combined NIRS and EEG measurements were performed twice on different days in each subject. Hemodynamic
responses were only found in both measurements in 41.7% of the infants. Stimulation was efficient, because reproducibility in EEG was high (96.3%). In two measurements
the optical neuronal signal were found. The noise level of the measurement was P20 = 9.6 · 10−5 % in change of optical density.
Another study aimed at establishing functional NIRS as neuroimaging method to assess hand motor function in the clinical setting of neurorehabilitation where traditional neuroimaging methods cannot be applied. 16 healthy right handed subjects performed five finger-tapping tasks of different complexities. It was demonstrated that significant cerebral changes were detected by NIRS in 100% of the subjects and that they are related to
increasing task complexity.
In two further studies, changes in cerebral oxygenation and perfusion were studied during sleep. Here the research focused on results within a group rather than the individual
patients.
In one of the two studies about NIRS during sleep, NIRS was applied in 19 patients with sleep disordered breathing who underwent nocturnal videopolysomnography in a sleep
laboratory. The aim was to study cerebral oxygenation and blood concentration during sleep disordered breathing. A new method of NIRS data evaluation led to the result, that
brain hypoxia only appears in cases of frequent apneas and obstructive events.

Finally, the first study to test whether NIRS can detect hemodynamic changes linked to hemispheric/cortical activity caused by periodic leg movement during sleep. Indeed,
cerebral hemodynamic changes are significantly correlated to periodic leg movement during sleep. In addition, the results showed that cerebral hemodynamic oscillations have larger amplitude when associated with changes in EEG.
In conclusion, it was shown that NIRS is a useful research tool to study hemodynamic responses to brain activity in adults and neonates, which is even able to detect task
complexity. In addition NIRS is a promising method to study brain oxygenation and blood volume in patients with sleep disorders.