KEYWORDS: Data modeling, Image segmentation, Fourier transforms, Deep learning, Sensors, Medical research, Machine learning, Functional near infrared spectroscopy, Digital signal processing
Functional near-infrared spectroscopy (fNIRS) presents an affordable and light-weight method to monitor the cerebral hemodynamics of the brain. However, noise and artefacts hamper the analysis of fNIRS signals. Thus, the signal quality assessment is a crucial step when planning fNIRS experiments. Currently no standardized method exists for the evaluation. Commonly used visual inspection of the signals is time consuming and prone to subjective bias. Recently use of machine learning and deep learning approaches have been applied for the fNIRS signal quality assessment, showing promising results. However, currently there are only a few experiments which have investigated the use of these approaches to evaluate fNIRS signal quality. In this human brain study, we utilized previously developed deep learning approach used for the assessment of PPG signal quality with short-time Fourier transform (STFT) to evaluate the quality of raw fNIRS signals with wavelengths 690 nm, 810 nm, 830 nm and 980 nm. The data was collected from 38 subjects with a two-channel fNIRS device, measured during breath hold protocol in sitting position. A total of 10,144 segments were extracted using a window of 10 seconds length without overlap and annotated for SQA by three independent evaluators. The segments were transformed with STFT, and further processed into 2D images. The images were used as input data for CNN deep learning network, and the output further used to classify the segments as acceptable or unacceptable. The results show high potential of using DL approach for fNIRS signal quality assessment with classification accuracy of 87.89 %.
Signal quality is crucial in any signal analysis. Typically, the reason for bad signal quality is inappropriate sensor placement which is also highly dependent on the measurement location. It is usually quite easy to get a good optical signal from finger, but not from the brain. This study aims to provide a real-time signal quality assessment method to help clinical personnel in placement of the fNIRS sensors on head to ensure good signal quality. Signal was segmented for each 10 seconds and a band-pass filter at 0.5-3 Hz was applied to isolate signal in cardiac band. Each segmented signal was subject to visual quality assessment to get bad, fair, and good labels. We used maximum to mean power ratio to generate signal quality index (SQI) score. Other methods included were skewness and kurtosis of the heart rate variability (HRV). Results showed that power ratio provides better consistency and separation among three different labels. Both skewness and kurtosis failed to separate fair and good segments. Using two threshold values, indices from power ration can be transformed into red (bad), yellow (fair), and green (good) alarm to help healthcare practitioners, who have no expertise to assess signal quality, to fix sensor placement to get good or acceptable signals.
Obtaining parameters that characterize cerebral fluid interactions in the human brain is of high interest particularly as regards studies of the brain clearance and in relation to neurodegeneration diseases (NDD). Furthermore, disturbances in sleep affecting brain clearance have been linked to NDDs like Alzheimer’s disease (AD). At present, polysomnography (PSG) is the methodological gold standard in sleep research being used in sleep labs. However, it does not provide direct information on cerebral fluid dynamics which may be an important parameter linked to brain clearance activity during sleep. We have developed functional near-infrared spectroscopy (fNIRS) based method for assessment of human cerebral fluid dynamics during sleep. It is optimized as a wearable sleep monitoring device enabling overnight sleep recordings at home without disturbing natural sleep. In this paper, we study spectral entropy (SE) of cerebral fluid dynamics during sleep study. Developed fNIRS technique measures, in addition to cerebral hemodynamics, cortical water concentration changes reflecting dynamics of the cerebrospinal fluid (CSF) volume in macroscale. Our preliminary results of overnight fNIRS sleep measurements from 10 adult subjects show that SE values fluctuate in cycle during the whole night sleep. It may indicate the transition among sleep stages.
Obtaining parameters that characterize cerebral fluid interactions in the human brain is of high interest particularly as regards studies of the glymphatic system and in relation to neurodegeneration diseases. Near-infrared spectroscopy (NIRS) based techniques commonly measure cerebral hemodynamics using a combination of wavelengths approximately between 650 nm and 950 nm, where light is to a lesser amount attenuated by water, enabling light to reach the brain. By adding a wavelength that is dominantly absorbed by water, while still penetrating below skull, we may have a possibility to measure also cortical water concentration changes, particularly dynamics of the cerebrospinal fluid (CSF) volume, which have been connected to brain’s waste removal system. In this study, we show based on in vivo human experiments that small dynamical variations in the CSF layer, between the human skull and brain cortex determined by MRI, correlate with near infrared (NIR) light intensity changes particularly above 960 nm when measured at long (< 3 cm) source-detector distance. In addition, based on the previously reported anti-correlation between total haemoglobin (HbT) and water signal fluctuations measured with NIRS, we further investigated the differences in the anti-correlations when using short (< 2 cm) and long source-detector distances. In general, at a short source-detector distance the NIRS measurement volume does not reach a depth below human skull. In consequence, our results from 12 healthy subjects show greater anti-correlation between HbT and water when using a long source-detector distance, supporting the idea that NIRS can be used to monitor also human cortical water fluctuations non-invasively.
Near-infrared spectroscopy (NIRS) based techniques in brain monitoring utilize the spectrum range approximately between 650 nm and 950 nm, where light attenuation is low enough to enable reaching the cerebral cortex of the brain. In these studies, particularly oxygenation changes in the cerebral cortex are of great interest since the concentrations of oxyhemoglobin (HbO) and deoxyhemoglobin (HbR) change due to coupling of hemodynamics to cortical neural activity. There are numerous simulation and phantom studies that show near-infrared (NIR) light can penetrate in the human head to a depth of approximately 1–2 cm, reaching the brain cortex. However, NIR propagation and light attenuation is also dependent on anatomy and size of the subject’s head. This related, we studied experimentally the effect of layer thicknesses of dura and cerebrospinal fluid (CSF), skull and skin to detected light intensity when measured in vivo from human heads with different layer thicknesses. We studied anatomy of 15 human heads in magnetic resonance imaging (MRI), particularly the thickness and morphology of the tissue layers of CSF, skull and skin. At the same time, we measured intensity and absorbance spectrum, at range of 600 nm to 1100 nm, from the forehead of these subjects when fibre detector was placed at distances of 1 cm and 3 cm from the fibre source. Our results show that each layer affects the detected NIR spectrum when layer thickness changes, particularly at 3 cm source-detector distance. However, these small spectral variations, caused by changes head anatomy, most likely do not have significant influence in quantifying cerebral hemodynamics.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.