Intelligent, Ambulatory Electroencephalography (EEG) Monitoring System
- Stephen Faul
- Robert McEvoy
- Liam Marnane
The aim of the system is to advance the methods currently used in ambulatory monitoring of biomedical signals. Currently, sensors detect signals, such as EEG, ECG, activity levels, etc and save them to a hard disk or flash memory. This leads to large, power hungry devices, latency in treatment and excessive time spent analysing data by medical staff. The aims of this demonstrator are to enhance the devices by using efficient algorithms to increase battery life, reduce size and improve the features and information provided to the medical staff in order to reduce analysis time and workload.
In order for this to be realised, algorithms and constraints tested in simulation are implemented on a real-time DSP processor to be evaluated. Benchmark power and time measurements have been taken for a basic system and improvements are sought through the alteration of algorithms, signal properties and information flow. This is the point where research from around the group is integrated together to create a prototype smart, physiological signal analysis tool. A system has been designed incorporating an Analog Devices Blackfin DSP processor, Analog Devices Sigma-Delta ADC at the front end, SDRAM and Bluetooth communication. This system has the power and scope to test various implementations of EEDSP algorithms.
Block Diagram of the EEDSP Demonstrator
A block diagram of the EEDSP demonstrator is shown above. The components of the system are as follows:
1. National Instruments USB 9263 Analog Output module. This is a 16-bit Digital-to-Analog Converter (DAC), which will allow us to test our demonstrator using pre-recorded EEG files, essentially emulating the EEG signals which would come from electrodes attached to the user’s scalp. National Instruments LabVIEW is used to control the Analog Output module.
2. A simple passive RC filter, required for anti-aliasing.
3. We have chosen the Analog Devices AD7715; a 16-bit, single-channel, Sigma-Delta, low-power Analog-to-Digital Converter (ADC). The ADC provides samples to the DSP processor at 250 Samples per second (Sps), similar to commercial hospital EEG machines which sample at 256 Sps.
4. The ADC is connected to the DSP processor via a Serial Peripheral Interface (SPI). The SPI is extendible and allows further ADCs to be added at a later stage, if desired.
5. Analog Devices BF537 Blackfin DSP microprocessor. The Blackfin processor fetches the EEG samples from the ADC, and applies the seizure detection algorithms. The Blackfin is also responsible for compressing the samples and storing them securely in memory.
6. External memory is required to store all of the EEG samples. It is envisaged that the demonstrator will be able to store 24 hours’ worth of EEG (initially single-channel), which can later be copied to a PC and accessed by a healthcare professional.
7. An option exists to add an FPGA to the system, to deal with some of the more computationally intensive parts of the seizure detection algorithms. This will be employed at a later stage in the project, if required.
8. Lemos International LM048 mini Bluetooth adaptor. This adaptor is connected to the Blackfin via UART (serial port), and allows data from the demonstrator to be transmitted to a user’s (e.g. the patient’s) PC. The traditionally high power associated with Bluetooth communications is not problematic, since transmission between the demonstrator and the monitoring station will only consist of alarm messages and on-demand, infrequent status/data requests.
Photos of EEDSP Demonstrator, October 2009