A blueprint for real-time functional mapping via human intracranial recordings

Abstract

Background: The surgical treatment of patients with intractable epilepsy is preceded by a pre-surgical evaluation period during which intracranial EEG recordings are performed to identify the epileptogenic network and provide a functional map of eloquent cerebral areas that need to be spared to minimize the risk of post-operative deficits. A growing body of research based on such invasive recordings indicates that cortical oscillations at various frequencies, especially in the gamma range (40 to 150 Hz), can provide efficient markers of task-related neural network activity.

Principal findings: Here we introduce a novel real-time investigation framework for mapping human brain functions based on online visualization of the spectral power of the ongoing intracranial activity. The results obtained with the first two implanted epilepsy patients who used the proposed online system illustrate its feasibility and utility both for clinical applications, as a complementary tool to electrical stimulation for presurgical mapping purposes, and for basic research, as an exploratory tool used to detect correlations between behavior and oscillatory power modulations. Furthermore, our findings suggest a putative role for high gamma oscillations in higher-order auditory processing involved in speech and music perception.

Conclusion/significance: The proposed real-time setup is a promising tool for presurgical mapping, the investigation of functional brain dynamics, and possibly for neurofeedback training and brain computer interfaces.

Figure 1. Brain TV set-up.

(a) The online 3D bar plot displays the time-varying power modulations measured at a selected cortical site over the last 10 seconds in three frequency bands (8–30 Hz, 60–90 Hz and 110–140 Hz). The depicted power modulations represent the power for a given frequency expressed in units of standard deviation of a reference power measured during a baseline period. Baseline power is defined during a calibration phase performed at the beginning of the session while the subject was instructed to rest with eyes open. (Note that the white annotations on the figure are not part of the display). (b) The clinical setting: Patient watching a display that provides him with a visual feedback of online power variations picked up by one of his implanted electrodes.

Figure 2. Oscillatory power modulations associated with speech in the mid superior temporal sulcus.

Time-Frequency maps representing the time course of spectral power modulations at electrode t'8 (Talairach coordinates : −65, −12, 4) for frequencies up to 200 Hz in each one of the 6 experimental conditions. The frequency values range from 1 to 200 Hz and the horizontal time axis spans over 3 seconds including 2 s of auditory stimulation (black horizontal bar). Note that, power modulations are given in units of the standard deviation from baseline period (see methods) and that the normalization parameters were the same for all maps.

Figure 3. Differences in gamma band reactivity between electrodes in mid superior temporal sulcus and the primary auditory cortex.

(a,b) Bar plots of mean and standard deviation of the oscillatory power increase measured in the gamma range (60–140 Hz) during the last two seconds of auditory stimulation, for all six experimental conditions at two distinct locations: The upper plot (a) represent the modulations obtained from electrode t'8 in the mid superior temporal sulcus (Talairach coordinates : −65, −12, 4), while the lower plot (b) corresponds to site u'4 in the primary auditory cortex (Talairach coordinates : −43, −27, 11) . (c) Position of the two electrodes (t'8 and u'4) when displayed on a standard 3D cortical rendering of the Montreal Neurological Institute (MNI) single subject and (d) as seen on the patient's individual MRI.

Figure 4. Functional dissociation between regions of the mid portion of the superior temporal gyrus in response to speech and music.

(a) The general region in which each site was located is represented with green oval markers. Note that this patient had a cardiac pacemaker incompatible with MRI scanning and therefore, the Talairach coordinates of the sites could not be reconstructed with a precision higher than 1 cm along the y and z axis. (b) The spatial location of the recording sites in the coronal plane relative to the Sylvian fissure. (c) Time-Frequency maps depicting the time course of spectral power measured in left (t'4, t'6) and right (t5 and t7) sites for all frequencies up to 200 Hz during the first two seconds of auditory stimulation (black horizontal bar) in response to speech (SPEECH) and music (MUSIC). The baseline is common to the NORMAL and MUSIC conditions. (d) Bar plot display of the mean and standard deviation of the total activation measured in the high gamma (60–140 Hz) range during the first two seconds of auditory stimulation, for each one of the eight conditions of the SOUND paradigm.

Figure 5. Response to “speaker change” in the posterior part of the left superior temporal gyrus.

(a) Location of the electrode site u'8 (green oval marker) (b) Time-frequency map depicting gamma response to a sentence uttered by a new speaker (second sentence in a SP pair). (c) Bar plots illustrating gamma response during the first two seconds of auditory stimulation for the second sound sequence of each type of sound sequence pairs in this protocol. (see paradigm description in the text for more details)