Recent advances in imaging techniques have provided the clinician with new methods to identify epileptic foci, rolandic cortex, and eloquent language areas. The best method for understanding how these imaging techniques, such as magnetoencephalography (MEG), positron emission tomography (PET), single photon emission computed tomography (SPECT), functional magnetic resonance imaging (fMRI), and optical imaging, fit together is to examine briefly their temporal and spatial resolutions.^{11,14,18,30,31,53,55,57} Accurate maps of brain function and the spread of epileptiform activity require excellent spatial resolution to identify interictal activity and the site of onset of ictal activity and to find the source of the epileptic activity and the pathways by which the seizure activity spreads.^13,47,49 Unfortunately, although each of these new methods holds significant promise, none stands alone to provide the unique answer for identifying epileptic foci and mapping functional activity. PET (Chapter 80) provides three-dimensional information regarding functional activity but has a temporal resolution of only 40 seconds and a spatial resolution of 5 to 7.5 mm², limiting its ability to identify ictal onsets and individual interictal activity.^1,38,41 SPECT has similar limitations, although newer imaging compounds have allowed identification of epileptic foci and ictal-onsets zones.^14,53 Functional MRI (Chapter 83) is rapidly advancing, and provides three-dimensional identification of rolandic cortex and has been shown to have the potential to localize eloquent regions, especially Broca area; however, identification of epileptic foci and seizure spread awaits further advancements.^2,11,14,55 MEG (Chapter 78) has the fastest temporal resolution, which is necessary to identify epileptic foci, but is somewhat limited by its overall spatial resolution and the methods necessary to identify rolandic cortex and eloquent regions.^15,18 Optical imaging is one of the latest imaging techniques to be used on human cortex with potential advantages of being the only technique used for intraoperative localization of epileptic foci, rolandic cortex, and eloquent language regions.^26,38,44 This chapter focuses on the use of optical imaging in identifying epileptic foci and the spread of seizure activity in human patients undergoing epilepsy surgery.

The foundational work that allowed optical imaging to be used in the operating room to identify speech and sensory cortical regions began more than 50 years ago. Using the nerve trunk of the crayfish, Hill and Keynes²⁷ found optical changes in the nerve trunk that correlated with stimulation of the nerve. The “intrinsic signal” changes (without any voltage-sensitive dyes) in the nerve trunk recorded by a photocell showed an increase in the transmission of light during stimulation and a more prolonged undershoot following the stimulation. The potential uses of optical recordings have been advanced during the last 20 years by the work of a small group of investigators including, but not limited to, Waggoner, Davila, Salzberg, Grinvald, Ross, and Orbach.^{7,8,9,10,16,19,36,41} These investigators have made vital contributions to the techniques of optical imaging by screening and developing optical probes of membrane potential (voltage-sensitive dyes) as well as by pioneering different technologies for their use.

Blasdel and Salama⁵ expanded on these techniques by using a television camera to obtain greater spatial resolution (120 × 100) than had been previously possible with the standard photodiode arrays (24 × 24) by previous investigators. Blasdel went on to use optical imaging and voltage-sensitive dyes to visualize functional domains of visual cortex in nonhuman primates such as ocular dominance columns and orientation preferences.^3,4 Frostig et al., Grinvald et al., and Ts’o et al.^16,19,54 were later able to identify similar functional regions in the primate visual cortex without voltage-sensitive dyes by using the intrinsic signal changes in the optical reflectants from the cortical surface. These advances led to the first human epileptiform and functional images of language and sensory cortex obtained by Haglund et al.²⁶ There are a vast number of questions that can be addressed with optical imaging techniques; however, this chapter focuses on the basics of the technique and how the technique can be used to identify epileptic foci and follow the spread of seizure activity across neocortex. The use of optical imaging to localize rolandic cortex of cortical regions involved in higher cognitive functions such as language, memory, or phrase processing is not covered in this chapter and has been reviewed by Pouratian et al.³⁸ and Suh et al.⁵⁰

Much of the fundamental work on optical imaging concerns the correlation of optical changes observed with voltage-sensitive dyes and electrophysiologic changes.^3,4,5 The voltage-sensitive dyes were known to bind to cell membranes and their absorption properties change when the cell membrane voltage changes. The work on the invertebrates established that the optical changes observed with voltage-sensitive dyes were tightly correlated with action potential depolarization recorded by the means of intracellular electrodes. Correlation of optical changes to functional regions in cortex was developed from work on the rat somatosensory cortex with the identification of individual barrels using single-unit recordings and optical images.^32,34 Blasdel and Salama added to the correlative information on optical imaging and single unit recordings by demonstrating that in the intact nonhuman primate visual cortex, the peak optical changes for the specific orientations correlate with peak optical imaging changes
for that specific orientation stimulus. Anatomic studies also confirm that these optical changes correlated with anatomic boundaries.^3,4,5

The understanding of the mechanism underlying the intrinsic signals still awaits more detailed studies, but some information has been presented in the last several years. Initially, Frostig et al.¹⁶ demonstrated in a monkey visual cortex that ocular dominance orientation preferences that correlated with anatomic substrates could be found using the intrinsic signal. These studies have shown that the intrinsic signal changes can be used for mapping functional activity during visual-evoked changes.^{12,46,47,48,54,56} Using an in vitro slice preparation, MacVicar and Hochman³³ have shown that there was a broad range of wavelengths over which the intrinsic signal changes allow more light transmission through the slice when the Schaffer collaterals are stimulated in the hippocampal slice. They went on to show that the intrinsic signal depends on a Na-K-2Cl cotransport mechanism, which is blocked by furosemide. Because the intrinsic signal is abolished by the Na-K-2Cl cotransport inhibitor, the mechanism underlying the intrinsic signal in the bloodless brain slice preparation appears to be a consequence of cellular swelling. However, Frostig et al.¹⁶ have shown that the intrinsic signal also depends on blood volume and oxygen delivery changes. More recent studies from Haglund and Hochman^23,24 demonstrated that in the human neocortex, the changes in the intrinsic signal could be either due to blood volume or blood oxygenation changes, depending on the wavelength used. Using a wavelength at the isobestic point for hemoglobin (535 nm), changes in the negative direction (the tissue getting darker) are localized to pial arterioles that are dilating during functional or epileptic activity. In contrast, imaging done at a wavelength where the oxy-deoxy-hemoglobin absorption curves are maximally separated shows blood oxygenation changes restricted mainly to the draining veins. These blood oxygenation changes are more likely to go along with the BOLD (blood oxygen level–dependent) signal that is part of the fMRI imaging paradigms.^29,31,37,40 Using this technique, Haglund and Hochman were able to demonstrate spontaneous ictal activity in the human cortex and more recently interictal activity in the human cortex that correlated with intrinsic signal changes. The most localized changes appear to go along with the blood volume at the isobestic point for hemoglobin (green light, 535 nm).