The aim of this study was to develop a method that would allow us to conduct long-term follow up studies in order to investigate dynamic cortical and subcortical network changes during epileptogenesis using EEG and fMRI.

We used glue-cement based method for the chronic implantation of MRI-coil and EEG electrodes in adult male Wistar rats. EEG electrodes (Ag-wire, loop diameter 1mm) were glued into dents drilled over somatosensory (S1) cortex at both hemispheres under isoflurane anesthesia. Stainless steel sockets of the electrodes were connected inside a plastic pedestal positioned frontal to bregma. To receive RF, oval-shaped insulated copper wire loop was fixated onto skull with bonecement and gold pin connectors protruded from the front of the implant (Fig.1). Tunable diameter range of the exchangeable loops is from 16 mm to 21 mm and prior to implantation, each coil was bench tested to verify the element tuning and matching. The external part of the receive-only coil was constructed by Rapid Biomedical GmbH, Germany. The coil is actively decoupled during transmit and the variable tune/match for adaption to different loads and loop sizes is achieved by mechanical trimmer capacitors. MRI images were acquired with a Bruker 7 Tesla PharmaScan.

For the functional MRI experiments, rats were anaesthetized with combination of Isoflurane (0.5%) and medetomidine (0.1mg/kg/h). Body temperature was maintained at 37 ± 1 °C. Oxygen saturation, heart rate, breathing rate and temperature were monitored during the entire experiment. Functional data was acquired from the coronal slices using a Spin Echo - Echo Planar Imaging (SE-EPI) sequence (FOV = 25mm, slice thickness = 1.5 mm, TR = 2000, TE = 45 ms). Anatomical images were acquired using different spin-echo based pulse sequences with T2 weighting (FOV = 50mm, slice thickness = 0.75 mm, TR = 4676.46, TE = 16.133 ms). The data were converted from Bruker format to NIfTI using Aedes. Subsequently, data were slice-timing corrected, motion-corrected, spatially smoothed, and co-registered using home-made Matlab codes, Aedes, and SPM8. Voxel time-series were averaged to Region-of-Interest (ROI) time-series. Functional connectivity between ROI time-series was calculated using Pearson correlation.

For long-term video-EEG monitoring rats were housed individually in plexiglas cages where they could move freely. Rats were connected to a Nicolet (Nervus) EEG Recording System (sampling rate 2048 Hz, low-pass filtered at 150 Hz) and Nervus magnus 32/8 Amplifier. The behavior of the animals was recorded using a video camera equipped with infrared light to allow continuous 24 h/d video-monitoring. Each EEG file was analyzed visually by scanning through the EEG recording offline. Additionally, a fast Fourier transform was used to compute the power in different frequency bands. Epileptogenesis was induced by intraperitoneal injection of pilocarpine (320mg/kg). Prior to pilocarpine, rats were injected with scopolamine (1mg/kg). Diazepam (5mg/kg) was administered 3h after pilocarpine. Behavioral scoring was done using Racine scale 1-5 (RS1-5).

Our results from animals tested for several months for the coil and EEG function, indicate that our modified screw-free implantation method was well suited for long-term follow-up studies (≤6 months, n=15) both in freely moving video-EEG settings and functional MRI without causing major susceptibility artifacts.

To test simultaneous EEG-fMRI, sensory evoked response (SEP) to stimulus (30 seconds train at 9Hz, 2mA, with 1ms single pulse duration) was tested (n=4) (Fig.2). Data shows that clean EEG signal can be acquired together with high quality fMRI data using the chronically implanted coil and electrodes.

To test the feasibility to study functional connectivity changes longitudinally using our setup with recovery anesthesia, four animals underwent invasive femoral artery cannulation to examine if ISO+MED combination was sufficient to anesthetize spontaneously breathing animals without significant alterations in blood gas values during a 60min resting-state fMRI. Arterial pH, oxygen saturation, partial pressure carbon dioxide and oxygen were within acceptable range at the end of the imaging (pH: 7.35±0.01; sO2:97.3±0.6%; pCO2: 50.75±2.7; pO2: 112±19; n=4).

To examine dynamic changes before and after an insult resulting in epileptogenesis, subset of rats (n=4) were subjected to pilocarpine treatment. Initial results indicate that our method is robust and well suited for several video-EEG and MRI data collections time-points to study e.g. EEG band-power changes prior to onset of spontaneous seizures (Fig.4).

These studies have shown that our new approach of implantable RF-coil and EEG electrodes makes possible to carry out long-term follow-up studies using video-EEG and EEG-fMRI. This would eventually gain us deeper understanding how large scale networks, as assed by resting-state fMRI, are modified during epileptogenesis or other progressive neurological conditions and what kind of neural activity contributes to this.