Direct, intraoperative observation of ~0.1 Hz hemodynamic oscillations in awake human cortex: implications for fMRI

Abstract

An almost sinusoidal, large amplitude ~0.1 Hz oscillation in cortical hemodynamics has been repeatedly observed in species ranging from mice to humans. However, the occurrence of 'slow sinusoidal hemodynamic oscillations' (SSHOs) in human functional magnetic resonance imaging (fMRI) studies is rarely noted or considered. As a result, little investigation into the cause of SSHOs has been undertaken, and their potential to confound fMRI analysis, as well as their possible value as a functional biomarker has been largely overlooked. Here, we report direct observation of large-amplitude, sinusoidal ~0.1 Hz hemodynamic oscillations in the cortex of an awake human undergoing surgical resection of a brain tumor. Intraoperative multispectral optical intrinsic signal imaging (MS-OISI) revealed that SSHOs were spatially localized to distinct regions of the cortex, exhibited wave-like propagation, and involved oscillations in the diameter of specific pial arterioles, indicating that the effect was not the result of systemic blood pressure oscillations. fMRI data collected from the same subject 4 days prior to surgery demonstrates that ~0.1 Hz oscillations in the BOLD signal can be detected around the same region. Intraoperative optical imaging data from a patient undergoing epilepsy surgery, in whom sinusoidal oscillations were not observed, is shown for comparison. This direct observation of the '0.1 Hz wave' in the awake human brain, using both intraoperative imaging and pre-operative fMRI, confirms that SSHOs occur in the human brain, and can be detected by fMRI. We discuss the possible physiological basis of this oscillation and its potential link to brain pathologies, highlighting its relevance to resting-state fMRI and its potential as a novel target for functional diagnosis and delineation of neurological disease.

Keywords: 0.1Hz oscillation; Cerebral hemodynamics; Intraoperative optical imaging; Resting state; Vasomotion; fMRI.

Figure 1. Direct observation of the 0.1 Hz hemodynamic oscillations in the awake human brain

(A) Grayscale image of subject 1 craniotomy under 530 nm illumination with 4 regions of interest (ROIs) indicated. (B) 4 corresponding time courses showing ΔHbT, ΔHbO and ΔHbR dynamics for each ROI indicated in A. (C) Power spectrum (abs(FFT)) of ΔHbT time courses shown in B. Green arrow indicates SSHO at ∼0.1 Hz. (D) Log-log abs(FFT) plot over wider frequency range showing otherwise 1/f shape. Heart-rate (2 Hz) is indicated by the thick blue arrow, breathing (0.4 Hz) with a thin blue arrow and the SSHO with a green arrow. (E) Fourier image of the field of view at 0.1 Hz, delineating 3 distinct vascular networks, outlined with white dashed ovals (stars indicate artifacts from specular reflections). (F) Fourier image at 0.15 Hz, away from the 0.1 Hz peak in the power spectrum, showing no delineated regions.

Figure 2. Non-sinusoidal low frequency hemodynamics in a second human subject

(A) Grayscale image of subject 2 craniotomy under 530 nm illumination with 4 ROIs indicated. (B) 4 corresponding time courses showing ΔHbT, ΔHbO and ΔHbR dynamics for each ROI indicated in A. (C) Power spectrum (abs(FFT)) of HbT time courses shown in B with 0.17 Hz breathing rate indicated by thin blue arrow. (D) Log-log abs(FFT) plot of wider frequency range with thicker blue arrows indicating heart rate (1 Hz) and its harmonic (2 Hz). (E) Fourier image of the field of view at 0.1 Hz, and (F) a Fourier image at 0.15 Hz. Stars indicate artifacts from specular reflections. No distinct vascular networks are apparent in either image. A movie of ΔHbT dynamics in subject 2 is shown in supplemental movie M2.

Figure 3. Properties of the ∼0.1 Hz oscillation in the human brain

(A) Pearson correlation coefficient image using the ΔHbT time course from ROI 3 in Figure 1. (B,C) Magnification and corresponding comparison of the correlation image with raw green channel grayscale image of the region indicated by the white dashed box in A. Note the large vascular network delineated by the correlation image extends under the large vein (black arrow). (D) Change in vessel diameter relating to change in ΔHbT over vessel indicated by white arrow in C. Light brown indicates raw vessel diameter time course, dark brown is the raw vessel diameter time course low pass filtered at 0.2 Hz. (E) Wave-like propagation of the 0.1 Hz oscillation depicted ασ a ΔHbT kymograph image of time courses extracted along the long dashed yellow ROI in A and in yellow in the inset gray scale image below (F). Red arrows mark pixels corresponding to blood vessels. Black arrow indicates the direction of a single wave's propagation. Equivalent data from two more trials is shown in supplementary figures S1 and S2. White arrow indicates a point of merge. (G) Using cross-correlation analysis, time lags between ∼0.1 Hz oscillating time courses extracted from 4 ROIs shown on the grayscale image to the left demonstrate directional propagation of the SSHO wave.

Figure 4. Observation of 0.1 Hz oscillation in pre-operative fMRI BOLD signal in subject 1

(A) Optical field of view and (B) correspondingly oriented field of view on the MRI outlined by a white dashed region showing 2 ROIs. (C) Additional deeper horizontal section showing an ROI over the left hand region in the right motor cortex. Sagittal view showing the location of the horizontal sections is displayed on the top right for B and C. A = anterior, P = posterior, R = right. (D) BOLD signal responses to left hand task in ROI 1. (E,F) BOLD signal time course from ROI 2 and ROI 3. Black crosses mark measured BOLD time-points. A spline interpolated trace is also shown in each case. (G) Power spectra (abs(FFT)) showing a ∼0.1 Hz peak in ROI 2 (indicated by the green arrow). Peak frequency corresponding to the left hand task is indicated by the brown arrow. (inset) Power spectra (abs(FFT)) from (G) plotted on a log-log scale, showing 1/f behavior. (H,I,J) Saggital, coronal and horizontal views of subject 1 respectively, showing the location and extent of the tumor.