Jaakko Virtanen

Comparison of principal and independent component analysis in removing extracerebral interference from near-infrared spectroscopy signals

Near-infrared spectroscopy (NIRS) is a method for noninvasive estimation of cerebral hemodynamic changes. Principal component analysis (PCA) and independent component analysis (ICA) can be used for decomposing a set of signals to underlying components. Our objective is to determine whether PCA or ICA is more efficient in identifying and removing scalp blood flow interference from multichannel NIRS signals. Concentration changes of oxygenated (HbO(2)) and deoxygenated (HbR) hemoglobin are measured on the forehead with multichannel NIRS during hyper- and hypocapnia. PCA and ICA are used separately to identify and remove signal contribution from extracerebral tissue, and the resulting estimates of cerebral responses are compared to the expected cerebral responses. Both methods were able to reduce extracerebral contribution to the signals, but PCA typically performs equal to or better than ICA. The improvement in 3-cm signal quality achieved with both methods is comparable to increasing the source-detector separation from 3 to 5 cm. Especially PCA appears to be well suited for use in NIRS applications where the cerebral activation is diffuse, such as monitoring of global cerebral oxygenation and hemodynamics. Performance differences between PCA and ICA could be attributed primarily to different criteria for identifying the surface effect.

Accelerometer-based method for correcting signal baseline changes caused by motion artifacts in medical near-infrared spectroscopy

In medical near-infrared spectroscopy (NIRS), movements of the subject often cause large step changes in the baselines of the measured light attenuation signals. This prevents comparison of hemoglobin concentration levels before and after movement. We present an accelerometer-based motion artifact removal (ABAMAR) algorithm for correcting such baseline motion artifacts (BMAs). ABAMAR can be easily adapted to various long-term monitoring applications of NIRS. We applied ABAMAR to NIRS data collected in 23 all-night sleep measurements and containing BMAs from involuntary movements during sleep. For reference, three NIRS researchers independently identified BMAs from the data. To determine whether the use of an accelerometer improves BMA detection accuracy, we compared ABAMAR to motion detection based on peaks in the moving standard deviation (SD) of NIRS data. The number of BMAs identified by ABAMAR was similar to the number detected by the humans, and 79% of the artifacts identified by ABAMAR were confirmed by at least two humans. While the moving SD of NIRS data could also be used for motion detection, on average 2 out of the 10 largest SD peaks in NIRS data each night occurred without the presence of movement. Thus, using an accelerometer improves BMA detection accuracy in NIRS.

Spontaneous Hemodynamic Oscillations during Human Sleep and Sleep Stage Transitions Characterized with Near-Infrared Spectroscopy

Understanding the interaction between the nervous system and cerebral vasculature is fundamental to forming a complete picture of the neurophysiology of sleep and its role in maintaining physiological homeostasis. However, the intrinsic hemodynamics of slow-wave sleep (SWS) are still poorly known. We carried out 30 all-night sleep measurements with combined near-infrared spectroscopy (NIRS) and polysomnography to investigate spontaneous hemodynamic behavior in SWS compared to light (LS) and rapid-eye-movement sleep (REM). In particular, we concentrated on slow oscillations (3–150 mHz) in oxy- and deoxyhemoglobin concentrations, heart rate, arterial oxygen saturation, and the pulsation amplitude of the photoplethysmographic signal. We also analyzed the behavior of these variables during sleep stage transitions. The results indicate that slow spontaneous cortical and systemic hemodynamic activity is reduced in SWS compared to LS, REM, and wakefulness. This behavior may be explained by neuronal synchronization observed in electrophysiological studies of SWS and a reduction in autonomic nervous system activity. Also, sleep stage transitions are asymmetric, so that the SWS-to-LS and LS-to-REM transitions, which are associated with an increase in the complexity of cortical electrophysiological activity, are characterized by more dramatic hemodynamic changes than the opposite transitions. Thus, it appears that while the onset of SWS and termination of REM occur only as gradual processes over time, the termination of SWS and onset of REM may be triggered more abruptly by a particular physiological event or condition. The results suggest that scalp hemodynamic changes should be considered alongside cortical hemodynamic changes in NIRS sleep studies to assess the interaction between the autonomic and central nervous systems.

Impaired cerebral vasoreactivity may cause cerebral blood volume dip following obstructive sleep apnea termination

Near-infrared spectroscopy (NIRS) is a non-invasive technique for estimating cortical concentration changes of oxy(Δ[HbO2]), deoxy(Δ[HbR]), and total (Δ[HbT]=Δ[HbO2] +Δ[HbR]) hemoglobin [1, 2]. Cortical Δ[HbT] is commonly used as an indicator of cerebral blood volume (CBV) changes. Obstructive sleep apnea (OSA) is characterized by apneas (pause in breathing lasting over 10 s) or hypopneas (reduced respiratory air flow lasting over 10 s, accompanied by blood oxygen desaturation of at least 4% or EEG arousal) during sleep. The resulting oxygen and sleep deprivation can lead to severe health problems ranging from fatigue to coronary artery disease and stroke [3]. In a recent study, transcranial Doppler sonography (TCD) was used to measure cerebral blood flow velocity (CBFV) during sleep in OSA patients [4]. The study concluded that cerebral vasoreactivity decreases during OSA sleep, and apnea termination is followed by a drop in CBFV. Several studies indicate that apnea-induced changes in CBFV should be associated with parallel changes in cerebral blood flow (CBF) and consequently also in CBV [3, 5, 6]. However, in a recent NIRS study, CBV was found to stay relatively constant after apnea termination [2]. Here, we show repeatable NIRS results from a single OSA subject that agree with the TCD results and suggest a hemodynamic response pattern not previously reported in NIRS OSA studies.

Cyclic alternating pattern is associated with cerebral hemodynamic variation: A near-infrared spectroscopy study of sleep in healthy humans

The cyclic alternating pattern (CAP), that is, cyclic variation of brain activity within non-REM sleep stages, is related to sleep instability and preservation, as well as consolidation of learning. Unlike the well-known electrical activity of CAP, its cerebral hemodynamic counterpart has not been assessed in healthy subjects so far. We recorded scalp and cortical hemodynamics with near-infrared spectroscopy on the forehead and systemic hemodynamics (heart rate and amplitude of the photoplethysmograph) with a finger pulse oximeter during 23 nights in 11 subjects. Electrical CAP activity was recorded with a polysomnogram. CAP was related to changes in scalp, cortical, and systemic hemodynamic signals that resembled the ones seen in arousal. Due to their repetitive nature, CAP sequences manifested as low- and very-low-frequency oscillations in the hemodynamic signals. The subtype A3+B showed the strongest hemodynamic changes. A transient hypoxia occurred during CAP cycles, suggesting that an increased CAP rate, especially with the subtype A3+B, which may result from diseases or fragmented sleep, might have an adverse effect on the cerebral vasculature.

Properties of end-expiratory breath hold responses measured with near-infrared spectroscopy

Near-infrared spectroscopy (NIRS) can be used to assess the cerebrovascular response to breath hold. We measured eight healthy subjects during voluntary end-expiratory breath hold to study inter- and intraindividual variability of the deoxy- (HbR) and oxyhemoglobin (HbO2) response curves for the scalp and cerebral cortex. Although cortical [HbO2] behaves qualitatively similarly in all subjects, there is large inter- and intraindividual variability, and in the case of [HbR] also qualitative variability. However, the linearity of [HbO2] increase during the breath hold has encouraging measurement repeatability, and it may even indicate an individuals CO2 tolerance. This result may help understand why breath hold duration varies between subjects more than the total [HbO2] increase during breath hold.

Slow spontaneous hemodynamic oscillations during sleep measured with near-infrared spectroscopy

Spontaneous cerebral hemodynamic oscillations below 100 mHz reflect the level of cerebral activity, modulate hemodynamic responses to tasks and stimuli, and may aid in detecting various pathologies of the brain. Near-infrared spectroscopy (NIRS) is ideally suited for both measuring spontaneous hemodynamic oscillations and monitoring sleep, but little research has been performed to combine these two applications. We analyzed 30 all-night NIRS–electroencephalography (EEG) sleep recordings to investigate spontaneous hemodynamic activity relative to sleep stages determined by polysomnography.Signal power of hemodynamic oscillations in the low-frequency (LF, 40–150 mHz) and very-low-frequency (VLF, 3–40 mHz) bands decreased in slow-wave sleep (SWS) compared to light sleep (LS) and rapid-eye-movement (REM) sleep. No statistically significant (p < 0.05) differences in oscillation power between LS and REM were observed. However, the period of VLF oscillations around 8 mHz increased in REM sleep in line with earlier studies with other modalities. These results increase our knowledge of the physiology of sleep, complement EEG data, and demonstrate the applicability of NIRS to studying spontaneous hemodynamic fluctuations during sleep.

Sleep apnea termination decreases cerebral blood volume: a near-infrared spectroscopy case study

Medical near-infrared spectroscopy (NIRS) can be used to estimate cerebral haemodynamic changes non-invasively. Sleep apnea is a common sleep disorder where repetitive pauses in breathing decrease the quality of sleep and exposes the individual to various health problems. We have measured oxygenated and deoxygenated haemoglobin concentration changes during apneic events in sleep from the forehead of one subject using NIRS and used principal component analysis to extract extracerebral and cortical haemodynamic changes from NIRS signals. Comparison of NIRS signals with EEG, bioimpedance, and pulse oximetry data suggests that termination of apnea leads to decreases in cerebral blood volume and flow that may be related to neurological arousal via neurovascular coupling.

Principal component analysis and LMS filtering in removing surface effects from near-infrared spectroscopy signals

Signal processing methods can be used to remove surface effects from near-infrared-spectroscopic cerebral signals. In this study we successfully tested two methods: principal component analysis and least-mean-squares filtering.