Abstracts

Rationale: Intracranial monitoring with depth electrodes allows mapping of seizure onset zones and associated networks in patients with medication refractory epilepsy. Given the lack of model systems with high-fidelity to human brain disease, research recordings from the human brain have proven pivotal in understanding normal and pathologic neurophysiology, leading to new therapies. Biosignal amplifiers sensitive to electromagnetic fields prevalent in the hospital setting complicate acquisition of low noise recording from the human brain. Here we document sources of artifact and signal digitization considerations affecting acquisition of continuous human brain recordings. We share our lessons learned to provide a roadmap for research groups beginning to record intracranial EEG data in the Epilepsy Monitoring Unit.



Methods: As a part of Institutional Review Board approved study, we obtained patient informed consent to acquire intracranial EEG using a research grade biosignal amplifier sampling at 4800Hz parallel to the clinical data stream. Our group assembled a research cart with a power supply, 256-channel biosignal amplifier, switching box, neurostimulator, and data review monitor. The intracranial EEG signal was split from a Passive Electrode Connector Box and routed to both the clinical and research amplifiers. The headboxes were mounted to an IV pole at the patient bedside to minimize distance between the recording electrode and amplifier. Line noise was calculated as the power at 60Hz and compared following each experimental change.


Results: The research cart installation involved collaboration with the device manufacturer, biomedical engineering, nursing, EEG technician, and epileptologists. Patient safety concerns including provider emergent access to the bedside were considered. The obtained research and clinical recordings were systematically evaluated to screen for sources of artifact after each modification of the data acquisition cart. The placement of the recording equipment on the cart was changed to avoid crossing AC and DC signals, a source of noise that affected banks of channels independently from each other due to the physical orientation of the equipment. We designed the mounting for the headboxes to ensure that the clinical team could clean and move the tether point of the patient safely. We identified major sources of noise within the patient environment that could be kept away from the recording equipment to reduce noise contamination, especially during research recordings, leading to a large reduction of line noise. Finally, we worked with the biosignal amplifier vendor to address issues of signal dropout due to overloading recording laptop memory and computing power.


Conclusions: We report a research cart and equipment spatial configuration optimized to mitigate signal artifact while maintaining a low-profile at the bedside. A systematic visual and quantitative approach to evaluate for sources of artifact is critical to identify non-uniform effects across recording channels and time. Vigilance for these sources of artifact and data loss are critical to ensure a representative dataset of underlying neurophysiology.


Funding: Neurosurgery research and education foundation