Linear aspects of changes in deoxygenated hemoglobin concentration and cytochrome oxidase oxidation during brain activation.

We used near infrared spectroscopy (NIRS) to investigate the vascular and metabolic response to brain activation in human primary and adjacent secondary visual cortex. NIRS is able to measure concentration changes in deoxygenated hemoglobin ([deoxy-Hb]) (which mainly contribute to the blood oxygenation level-dependent (BOLD) signal in functional magnetic resonance imaging (fMRI)) as well as concentration changes of oxygenated hemoglobin ([oxy-Hb]) and corpuscular blood volume ([total-Hb] = [oxy-Hb] + [deoxy-Hb]) and changes in the redox status of the cytochrome c oxidase ([Cyt-Ox]), a putative parameter for cellular oxygenation. A sound understanding of the transfer functions between stimulus parameters, neuronal activity, and vascular/metabolic parameters is important for interpretation of data acquired with indirect neuroimaging techniques like fMRI, especially in event-related design studies. In the present study we tested whether the vascular/metabolic response to stimulation can be described as a linear and time invariant system. Since linearity is a property attributed to systems that satisfy the scaling and superposition properties, as a first simple test, superposition of the responses obtained from short duration visual stimuli was used to predict the responses of longer duration stimuli. Our results showed that the predictions of [deoxy-Hb] and [Cyt-Ox] responses to stimuli of 6- to 24-s duration were satisfactory whereas predictions of [oxy-Hb] and [total-Hb] were insufficient. In a second step, a calculated convolution function of an assumed impulse response function and the stimulus function was fitted with the measured [deoxy-Hb] and [Cyt-Ox] curves to obtain amplitude, time delay, and time constant parameters. We show that predictions of cellular and vascular oxygenation responses to visual stimulation are good for 6- to 24-s stimuli duration under the assumption of a linear transfer characteristic. This model is not valid for corpuscular volume changes which affect mainly the [oxy-Hb] response. Noninvasive NIRS is shown to be a suitable method to get more direct information about neuronal-activity-associated changes in cerebral parameters which are partly reflected in BOLD signal but are not fully understood yet. Copyright 2001 Academic Press.