Background
Neonatal seizures occur in around 10% of all pre-term live births, making them the most common neurological emergency in the Neonatal Intensive Care Unit (NICU), both nationally and worldwide . Although the underlying causes and subsequent effects of neonatal seizures on long-term outcome are not entirely understood, studies have shown that seizures are a risk factor for neurodevelopment sequelae such as cognitive impairment, moderate-severe brain injury, and epilepsy. However, detecting these seizures in the NICU is an ongoing clinical challenge. Although common symptoms include random eye movements, tightening of the muscles, or jerking movements of the body, these movements are often quite subtle. Thus, it is difficult to distinguish between clinically relevant and normal behavior in the neonate while relying on visual observation alone.
Currently, conventional electroencephalography or cEEG is the most commonly used method to detect seizures. In this noninvasive monitoring system, a technician connects multiple electrodes to a patient’s scalp in order to measure underlying electrical signals from the brain. A cheaper alternative to this monitoring method is amplitude integrated EEG or aEEG. Usually using only two or four EEG channels, aEEG performs data preprocessing steps such as band-pass filtering and semi-logarithmic time compression to create a simpler signal that is easier than the cEEG to analyze. Although this process of filtering makes it more difficult to detect certain types of seizures, the ease of use and interpretation often overcome this enhanced accuracy, which is why the use of aEEG in the NICU has been rising over the last few years.
In both cases, the output from the monitoring system must be continually observed and evaluated if there is hope for seizure treatment in real-time. With the conventional EEG specifically, a trained neurologist is necessary to properly analyze the electrode readings. Neonatologists, such as our client Dr. Zachary Vesoulis, would prefer to have an automatic seizure detection algorithm that could alert clinicians if and when a seizure is occurring and would remove the need for constant monitoring of EEG output.
Currently, conventional electroencephalography or cEEG is the most commonly used method to detect seizures. In this noninvasive monitoring system, a technician connects multiple electrodes to a patient’s scalp in order to measure underlying electrical signals from the brain. A cheaper alternative to this monitoring method is amplitude integrated EEG or aEEG. Usually using only two or four EEG channels, aEEG performs data preprocessing steps such as band-pass filtering and semi-logarithmic time compression to create a simpler signal that is easier than the cEEG to analyze. Although this process of filtering makes it more difficult to detect certain types of seizures, the ease of use and interpretation often overcome this enhanced accuracy, which is why the use of aEEG in the NICU has been rising over the last few years.
In both cases, the output from the monitoring system must be continually observed and evaluated if there is hope for seizure treatment in real-time. With the conventional EEG specifically, a trained neurologist is necessary to properly analyze the electrode readings. Neonatologists, such as our client Dr. Zachary Vesoulis, would prefer to have an automatic seizure detection algorithm that could alert clinicians if and when a seizure is occurring and would remove the need for constant monitoring of EEG output.
The Need
There is a need for a seizure detection algorithm that can operate with a 2 channel EEG by neonatologists in a neonatal intensive care unit (NICU) to monitor breathing impaired neonates and deliver a signal that informs physicians in the unit that a seziure is impending.
Project Scope
We propose to deliver a protoype of a system that reliably detect seizures in neonates and consists of the following components:
- Software that can detect seizure utilizing a 2-channel EEG in neonates
- Hardware that alerts physicians of an ongoing seizure with signals that are not drowned out by other noise in the NICU
- Detailed instructions, and other documentation on how to operate software and hardware