This research investigated the feasibility of using time-dependent diffuse reflectance spectroscopy to differentiate
pediatric epileptic brain tissue from normal brain tissue. The optical spectroscopic technique monitored the dynamic
optical properties of the cerebral cortex that are associated with its physiological, morphological, and compositional
characteristics. Due to the transient irregular epileptic discharge activity within the epileptic brain tissue it was
hypothesized that the lesion would express abnormal dynamic optical behavior that would alter normal dynamic
behavior. Thirteen pediatric epilepsy patients and seven pediatric brain tumor patients (normal controls) were recruited
for this clinical study. Dynamic optical properties were obtained from the cortical surface intraoperatively using a timedependent
diffuse reflectance spectroscopy system. This system consisted of a fiber-optic probe, a tungsten-halogen light
source, and a spectrophotometer. It acquired diffuse reflectance spectra with a spectral range of 204 nm to 932 nm at a
rate of 33 spectra per second for approximately 12 seconds. Biopsy samples were taken from electrophysiologically
abnormal cortex and evaluated by a neuropathologist, which served as a gold standard for lesion classification. For data
analysis, spectral intensity changes of diffuse reflectance in the time domain at two different wavelengths from each
investigated site were compared. Negative correlation segment, defined by the periods where the intensity changes at the
two wavelengths were opposite in their slope polarity, were extracted. The total duration of negative correlation, referred
to as the "negative correlation time index", was calculated by integrating the negative correlation segments. The negative
correlation time indices from all investigated sites were sub-grouped according to the corresponding histological
classifications. The difference between the mean indices of two subgroups was evaluated by standard t-test. These
comparison and calculation procedures were carried out for all possible wavelength combinations between 400 nm and
800 nm with 2 nm increments. The positive group consisted of seven pathologically abnormal test sites, and the negative
group consisted of 13 normal test sites from non-epileptic tumor patients. A standard t-test showed significant difference
between negative correlation time indices from the two groups at the wavelength combinations of 700-760 nm versus
550-580 nm. An empirical discrimination algorithm based on the negative correlation time indices in this range produced
100% sensitivity and 85% specificity. Based on these results time-dependent diffuse reflectance spectroscopy with
optimized data analysis methods differentiates epileptic brain tissue from normal brain tissue adequately, therefore can
be utilized for surgical guidance, and may enhance the surgical outcome of pediatric epilepsy surgery.
|