Pietro Avanzini 1 , Fausto Caruana 2 , Gaetano Cantalupo 3 , Francesco Cardinale 4 , Alessio Moscato 4 , Roberto Mai 4 , Roberto Callieco 5 , Ivana Sartori 4 , Michele Terzaghi 5 1 Department of Neuroscience, University of Parma, Italy; 2 Brain Center for Social and Motor Cognition, Italian Institute of Technology, Italy; 3 Child Neuropsychiatry Unit, University of Parma, Italy; 4 “Claudio Munari” Center for Epilepsy Surgery, Ospedale Niguarda-Ca’ Granda, Italy; 5 Department of Public Health and Neurological Sciences, University of Pavia, Italy High-frequency reactivity to somato-sensory stimulation in the human thalamus: an intracerebral EEG study Introduction Thalamus is a deep brain structure subdivided in nuclei, each with speciic aferences and projections to cortical areas. The ventral postero-lateral nucleus ǻVPLǼ is the thalamic region mostly connected with the somatosensory cortex. Many segmentation procedures of the thalamus has been implemented in neuroimaging, mostly according to the probabilistic connectivity with cortical regions. Such an approach was recently validated by a joint iEEG and DTI study ǻElias, ŘŖŗŘǼ. However, while VPL is reported to be somatotopically organized by monkey studies ǻJones, ŗşŞŘǼ, no such evidence is given by human electrophysio- logical studies. Only a DTI study reported ǻHong, ŘŖŗŗǼ VPL to be somatotopi- cally arranged in the anteroposterior direction. Since studies by Cracco & Cracco ǻŗşŝŜǼ, high frequency oscillations ǻHFOǼ up to ŜŖŖ Hz were recorded from scalp in humans after electrical median nerve stimulation. These HFOs appeared both preceding and following the scalp NŘŖ, and many authors hypothesized that the early-HFOs ǻbefore NŘŖǼ are subcortical in origin, while the late-HFOs are generated in the somatosensory cortex ǻYamada et al., ŗşŞŞǼ. However, up to now, direct electrophysiological ev- idences are lacking, being available only various interference approaches and modeling in support to this subcorti- -cortical distinction ǻOzaki et al., ŘŖŗŘǼ. Materials and Methods Patients  Ŝ patients admited to the Niguarda Hospital, Milan, for presurgical evaluation of drug resistant epilepsy.  MRI gave negative indings in all patients, but Pś. No motor or sensory deicit could be found in all patients. Implantation of electrodes  Intracerebral electrodes were stereotactically placed ǻCossu et al, ŘŖŗŘǼ for clinical purposes.  Semi-rigid electrodes ǻDixi Medical, Besançon, FranceǼDZ ø Ŗ.Ş mm, from ŗŖ to ŗŞ leads of Ř-mm length, ŗ.ś mm apart. Somatosensory evoked potentials methodology.  Transcutaneous stimulation ǻNeuropack Mŗ- Nihon Kohden Corporation, JapanǼ for both tibial and median nerve,  Stimulation parametersDZ intensity ŗŖ% over the motor thresholdDz pulse width Ŗ.Ř msDz Ř Hz rate, ŗŖŖ repetitions.  EEG was acquired ǻsampling rate ŜŘśŖ HzǼ from all leads of each electrode exploring thalamus ǻneutral referenceǼ. Neuroimaging  Pre-implantation diagnostic MRI ǻTŗW-řD, ŗ.śT Achieva, Philips Medical Systems, Best, NetherlandsǼ.  DTIDZ difusion-weighted single shot spin-echo EPI sequence, voxel size of ŘxŘxŘ.Ŝ mm ř , Ŝś non collinear gradients.  Real intracranial electrodes position was assessed by cone-beam CT scanner ǻO-Arm, Medtronic Inc. MN, USAǼ. Selection of VPL contacts  Thalamic volume was deined according to FreeSurfer segmentation ǻhtpDZ//surfer.nmr.mgh.harvard.eduǼ.  Thalamo-cortical pathway for the hand and the leg were determined using the lemniscus medialis as seed region and the cortical representation of upper and lower limb in the postcentral gyrus as targets.  The leads within thalamic volume and closest to thalamo-cortical iber were identiied. Data Analysis  SegmentationDZ from –ŗŖ up to śŖ ms with respect to the stimulation onset. Prestimulus ǻŗŖ msǼ was used as baseline.  Somatosensory evoked potentials ǻSEPsǼDZ separate average for Ř series ǻŗŖŖ rep eachǼ were calculated ǻhighpass ŗŖ HzǼ.  The leads showing the earliest reactivity were identiied.  Time-frequencyDZ complex Morlet wavelet was computed for frequencies from řŖŖ Hz up to ŗśŖŖ Hz, returning both power ǻTF mapsǼ and inter-trial coherence ǻITC mapsǼ.  High frequency oscillations ǻHFOǼDZ only the high frequency ǻhighpass ilter řŖŖ HzǼ component of SEPs was ploted. Conclusions  We conirmed through the combination of neuroimaging and electrophysiological techniques the somatotopical organiza- tion within the thalamic VPL nucleusDz in particular, the foot representation appears to be more dorsal and caudal with re- spect to the hand one. Results are also compatible with a larger hand representation as compared to the foot one.  We characterized the VPL reactivity to both median and tibial nerve stimulation, describing the evoked activity in both time and frequency domain, highlighting the occurrence of HFO ǻup to ŗŘŖŖ HzǼ.  We demonstrated that, while the early HFO ǻoccurring before the scalp NŘŖǼ recorded from the primary somatosensory cortex relect the aferent volley from thalamus, the late HFO, widely described in scalp EEG or MEG recordings ǻOzaki, ŘŖŗŗǼ, relect an intracortical activity, with no involvement of the subcortical centers. MRI-DTI Tibial Nerve Stimulation MRI-DTI Median Nerve Stimulation Somatotopical Organization of VPL: Median nerve SEPs Median nerve stimulation gave a highly reproducible complex potential made up in all the subjects of a negative component followed by a broad second positive delection. On the descending and subsequent ascending branches of these waves, multiple high frequency oscillations ǻHFOǼ are clearly visible. Latencies of the irst negative delection and of the Nŗś peak are reported in table aside. Tibial nerve SEPs Tibial nerve SEPs were recorded only from PŚ, Pś and PŜ. No responses were detectable in Pŗ, PŘ and Př. SEPs consisted of a triphasic potential made up of two negative components separated by a positive one. HFO are visible on both descending and ascending branches of the potentials. Latencies of the irst negative peak result to be later than the median nerve SEPs. Yellow-highlighted leads show the earliest response ǻirst negative delectionǼ and match exactly the ones identiied by DTI imaging. Median nerve SEPs was visible on a greater number of leads with respect to the tibial nerve ones. This can be due to a larger representation of the hand within the VPL, and/or to the diferent orientation of the thalamo-cortical pathway of the upper and lower limb at their exit point from thalamus. When both median and tibial nerve SEPs were available, a clear segregation of the two responses was evident across adjacent leads. The Ŝ leads identiied for the median nerve SEPs were coregistered with the Schaltenbrand-Wahren atlas for the basal ganglia, in order to compare their position. PŘ and Př resulted to have the more rostral lead ǻnear VIMǼ, while PŚ, Pś and PŜ exhibited a more posterior location. Single trial analysis In both SEPs and HFO igures, the upper panel shows each trial as row ǻcolour coded voltageǼ. The lower panel depicts two separate averages of the irst and the second series of trials collected in the same acquisition. Note the repeatability of each single component of the median nerve SEPs ǻNŗś, PŘŖDz Patients Ř and řǼ and tibial nerve SEPs ǻNřŖ, PřśDz Patients Ś and śǼ. The HFO images, relative to the same data set after a highpass iltering at řŖŖ Hz, demonstrate that even the high fre- quency oscillations are characterized by a stereotyped behaviour with a strong phase reseting. The TF and ITC maps allowed to distinguish the pure power increase efect from the phase reseting one. In the TF maps, the power into the frequency range [řŖŖ-ŗśŖŖ Hz] was compared at each time point versus the baseline. According to a bootstrap statistic ǻp<Ŗ.ŖŖŗǼ, only time-frequency points with a signiicant power increase or decrease were depicted over the green background. In the same way, ITC maps shows only time-frequency points where the phase of the oscillations is signiicant with respect to the prestimulus. Blue and red colours indicate a positive and a negative phase, respectively. These measurements demonstrate that the ascending sensory volley evokes a stereotyped rapid succession of micro-waves ǻduration ŗ-ř msǼ with constant laten- cies and phase, possibly relecting the diferent timing of sensory input arriving to thalamic nuclei. First Negative deflection [msec] Negative Peak [msec] P1 12,85 14,00 P2 14,60 17,10 P3 11,90 12,90 P4 13,30 15,85 P5 12,00 13,95 P6 12,60 15,40 MNS First Negative deflection [msec] Negative Peak [msec] P4 26,50 30,70 P5 20,90 28,00 P6 28,10 31,70 TNS ITC HFO TF SEPs P1: the propagation of HFO through the thalamo-cortical pathway In patient Pŗ three leads belonging to diferent electrodes explored the VPL, one point alongside the thalamo -cortical pathway and the prima- ry somatosensory area. For the irst time, we were able to electrophisiologically describe the propagation of the somatosensory input ǻmedian nerve stimulationǼ from thalamus up to somatosensory cortex in humans. The VPL lead shows once again the behaviour described for median nerve SEPs. A strong HFO set ǻŞŖŖ -ŗŖŖŖ HzǼ becomes clearly visible supe- rimposed to the irst negative peak ǻŗś msǼ. The lead recording from the thalamo-cortical pathway shows a more simple shape ǻstrong negative peak at about ŗŝ.ś msǼ and an enhance- ment of HFO between ŗś and ŘŖ ms. The cortical lead exhibits a very complex SEPs, with many negative and positive peaks. Two clear sets of HFO are visibleDZ the irst before the 20 ms, with slightly higher frequency and lower amplitude, the later after the P20 peak, lower frequency and higher amplitude. As this last HFO set has no correlate in the deep recording leads, it must be the expression of an intracortical activity triggered only once the stimulus reaches the cortex. 111.16