HomER: a review of time-series analysis methods for near-infrared spectroscopy of the brain

Abstract

Near-infrared spectroscopy (NIRS) is a noninvasive neuroimaging tool for studying evoked hemodynamic changes within the brain. By this technique, changes in the optical absorption of light are recorded over time and are used to estimate the functionally evoked changes in cerebral oxyhemoglobin and deoxyhemoglobin concentrations that result from local cerebral vascular and oxygen metabolic effects during brain activity. Over the past three decades this technology has continued to grow, and today NIRS studies have found many niche applications in the fields of psychology, physiology, and cerebral pathology. The growing popularity of this technique is in part associated with a lower cost and increased portability of NIRS equipment when compared with other imaging modalities, such as functional magnetic resonance imaging and positron emission tomography. With this increasing number of applications, new techniques for the processing, analysis, and interpretation of NIRS data are continually being developed. We review some of the time-series and functional analysis techniques that are currently used in NIRS studies, we describe the practical implementation of various signal processing techniques for removing physiological, instrumental, and motion-artifact noise from optical data, and we discuss the unique aspects of NIRS analysis in comparison with other brain imaging modalities. These methods are described within the context of the MATLAB-based graphical user interface program, HomER, which we have developed and distributed to facilitate the processing of optical functional brain data.

Figures

Fig. 1

Typical setup for a NIRS experiment. (a) Sensitivity of a NIRS measurement determined by the propagation of light emanated from a source position and recorded by a detector placed several centimeters away. (b) Sample NIRS probe used to measure the primary motor cortex [10]. (c) Absorption spectrum (extinction coefficients) for oxyhemoglobin and deoxyhemoglobin over the range of wavelengths typically used for optical imaging [11].

Fig. 2

Tomographic optical probe: sample arrangement for a tomographic-style probe [97] used for a study of visual activation. To acquire optical signals from multiple source–detector distances, a time-multiplexing scheme is used in which our system switches between three sets of laser on–off states and detector gains (lower right). In our current system, a complete cycle can be imaged at up to 3 Hz. Overlapping (tomographic) measurements provide more uniform spatial sensitivity and coverage of the underlying brain. The theoretical sensitivity profile for this probe is shown in the upper right-hand panel with contour lines at 5 dB intervals based on a semi-infinite homogeneous (μa = 0.1 cm−1 and μs′ = 10 cm−1) slab geometry.

Fig. 3

Physiological fluctuations in optical signals: physiology fluctuations are generally the dominant source of noise in NIRS measurements because of the superficial sensitivity of the technique. This figure illustrates cardiac, respiratory, and blood pressure (Mayer wave) oscillations recorded during a resting period for the subject. The data also demonstrate a motion artifact, where the probes shifted during recording and generated a large perturbation of the signal intensity.

Fig. 4

Example of motion artifact removal by principal component (PCA) filtering. The raw data shown in (a) is experimental data measured in the inferior temporal cortex in an infant presented with an object recognition task [64]. The data were dominated by several motion artifacts that resulted from the movement of the probe on the head’s surface. These motion artifacts were highly covariant across the entire probe. (b) and (c) show the filtered data after the removal of the first and second principal components. After reduction of the covariance in the data by the removal of these components, the motion artifacts were greatly reduced. The gray shading indicates the presentation of stimuli.

Fig. 5

Removing systemic physiological noise by PCA: NIRS measurements are often sensitive to systemic fluctuations arising from blood pressure changes, respiration, or the cardiac cycle. As a complication, this physiology may change during the performance of intense stimulus tasks, such as motor activity. As is shown in (a), these changes can result in a systemic response giving the appearance of global functional activation. (b) When the PCA filter described in the text is used to remove this systemic effect by reducing this covariance, the activation region is localized to the motor area. These data were previously published in [69].

Fig. 6

Linear filtering of systemic physiology based on auxiliary measurements. Here we demonstrate the use of auxiliary measurements to improve the estimate of the functional hemodynamic response. Without the removal of these signals by bandpass filtering, the calculated hemodynamic response is heavily corrupted by these fluctuations. When the cardiac cycle and blood pressure fluctuations are used as additional regression variables, the functional hemodynamic response is more clearly separated from the effects of these systemic variables. The regression of this data with the external physiological measurements allows the separation of the data into the functional and systemic contributions. The panels on the right (b)–(d) show the separated system components for the evoked response (b), the cardiac related response (c) and the blood pressure component (d) composing the raw data (a).

Fig. 7

Screen shot of the HomER program. The layout of the HomER program is based around an interactive graphical display of the NIRS probe, shown in the upper right (b). The user specifies this probe geometry within the data file imported into HomER as described in the text. By selecting source (displayed as “x”) or detector (‘o”) positions on this probe layout, the user navigates through the display of their data. The original data are shown in (a) and the average evoked response is shown in (c). The data presented are described in [134] and were recorded during a 20 s finger-tapping task. Data shown are from a single run of one of the subjects.

Fig. 8

Levels of analysis in HomER. The HomER program architecture is based on three levels of analysis and processing: a single experimental scan, a session (single subject), and group analysis. At each level, various options for processing, visualization, and data management are offered.