Abstracts

Rationale: Imaging of intrinsic optical signals (IIOS) involves measuring the optical changes in the cortex elicited by neuronal activity. At certain wavelengths, these optical responses can be associated with changes in blood volume within the pial arteriole network (at 535 nm) or blood oxygen content within the venous network (at 660 nm). Intraoperatively acquired IIOS data is typically contaminated with large noise artifacts from respiration, heartbeat, and other sources. Further, the reliability of the IIOS signals at various wavelengths in their responses to cortical activity in the human brain has not been carefully investigated. To address these issues, we have developed a dynamic linear modeling (DLM) analysis of IIOS data that allows for removal of noise artifacts and statistical inference on the activity-evoked optical signals. Methods: The cortices of subjects undergoing neurosurgical treatment for intractable epilepsy were illuminated with either 535 nm or 660 nm light. Images were acquired with a cooled 16-bit digital CCD camera. Sequences of images were integrated over 100 ms or 200 ms intervals and stored on hard disk for offline analysis. In order to visually represent the data for qualitative analysis, “difference-images” were generated by subtracting a randomly chosen pre-stimulation image (i.e., “control-image”) from subsequent images. For quantitative analysis, time series of pixel values from within selected regions of interest were generated and analyzed using a DLM incorporating a polynomial component for the linear trend and cyclic components for respiration and heartbeat. Results: We first investigated the response of the cortex to electrical stimuli below the afterdischarge threshold. The intra-subject blood volume responses (535 nm) showed remarkably similar spatial and temporal patterns of responses from trial-to-trial. However the blood oxygenation responses (660 nm) showed variability, especially at higher stimulation currents. We next studied the optical changes elicited by spontaneously occurring ictal discharges in an awake patient. The blood volume changes where highly localized with a constant spatial pattern over time. In contrast, the blood oxygenation changes were highly variable in time, spreading at times to areas distant from the site of ictal discharges. Conclusions: DLM is found to be a useful method for removing unwanted artifacts and facilitating quantitative analysis of intraoperative IIOS data. This analysis showed that the same electrical stimuli elicited similar patterns of blood volume changes from trial-to-trial in the same subject, but that the patterns of blood oxygenation changes show variability, especially at larger stimulation currents. These differences between the variability of the blood volume and oxygenation maps are particularly apparent during ictal discharges. These results suggest that at the spatial resolution provided by IIOS, measurements of blood volume changes at 535 nm and their analysis with DLM may be a reliable means for intraoperative cortical mapping.