Charakterisierung der hämodynamischen Antwort bei Multispektraler Intraoperativer Optischer Bildgebung von Sprachfunktionen mittels Fourier-basierter Phasenauswertung und Wavelet-Analyse

Intraoperative Optical Imaging (IOI) is a technique that is able to visualize functional areas of the cerebral cortex. Recently we have shown that the technique can be applied during awake surgery for language mapping but results with existing data analysis frameworks were too unspecific to be useful for surgical guidance. Here, we are presenting a new approach with multispectral image acquisition and in-depth analysis of recorded hemodynamic response signals to optimize the method in the future.

Data from four patients (2 male, 2 female, median age 38) undergoing tumor resection in speech eloquent regions was acquired and analyzed. Imaging was performed using a customized multispectral imaging setup, allowing the simultaneous assessment of light wavelength bands that are sensitive to oxy- and deoxyhemoglobin as well as to changes in cerebral blood volume. Patients performed speech tasks (silent speech object naming) over 9 minutes with alternating 30 second rest and stimulation trials. After IOI, language mapping with DCS was performed to identify speech eloquent areas. The resulting imaging data was analyzed with a wavelet transformation to identify and characterize the frequency content of the signals especially in those speech eloquent regions of the cortex.

The results from the wavelet analysis are revealing that besides the actual stimulation frequency component (0.0167 Hz) and vasomotor components, especially the bands directly adjacent to the stimulation frequency are also represented with a high magnitude within the measured IOI signal. The magnitude of the stimulation frequency component is peaking in all 4 patients around 6 – 7 minutes after the beginning of the stimulation paradigm. Activity (phase) maps created from frequency bands adjacent to the stimulation frequency component are in all 4 patients in good agreement to intraoperative language mapping results with DCS.

Overall, the findings are suggesting that optimization of IOI for the identification of speech eloquent areas during surgery is possible. Shifts in the measured frequency component might be a result of deviations from the actual task. Here, the silent speech protocol might be improved by haptic feedback for supervision of correct task execution. Further investigations are mandatory.