Low Power Compression of EEG Signals Using JPEG2000 Garry Higgins, Brian Me Ginley, Martin Glavin, Edwad Jones Bioelectronics Research Cluster, NCBES National University of Irelad Galway Galway, Ireland g.higginsl@nuigalway.ie, brian.mcginley@nuigalway.ie, matin.glavin@nuigalwayje, edward.jones@nuigalwayje Abstract- This paper outlines a scheme for compressing EEG signals based on the JPEG2000 image compression algorithm. Such a scheme could be used to compress signals in an ambulatory system, where low-power operation is important to conserve battery life; therefore, a high compression ratio is desirable to reduce the amount of data that needs to be transmitted. The JPEG2000 specifcation makes use of the wavelet transform, which can be efciently implemented in embedded systems. The standard was broken down to its core components and adapted for use on EEG signals with additional compression steps added. Variations on the components were tested to maximize compression ratio (CR) while maintaining a low percentage root-mean-squared diference (PRD) and minimize power requirements. Initial tests indicate that the algorithm performs well in relation to other EEG compression methods proposed in the literature. Keord- EEG comression; Wavelets; JPEG2000 I. INTRODUCTION Recent advances in health care have seen an increased focus on at-home care and monitoring of patients. Portable devices allow patients to be monitored at home on an out patient basis, thus advancing the goal of providing ubiquitous and pervasive healthcare. This in t, relieves pressure on over-burdened hospital systems, and allows the patient remain in an environment they are comfortable in. It also allows more comprehensive monitoring with patients involved in a variety of activities in their day-to-day lives. For a device to be truly portable, there is a minimum battery life that would be required of it so that the wearer would not need to constantly remain beside a power source. It is for this reason that one of the main factors in designing a portable health care device is ensuring power consumption is at a minimum. Multichannel electroencepha l ogram (EEG) is a tool commonly used for measuring the electrical activity of the brain. The application of EEG to diagnose a variety of neurological conditions such as Epilepsy and Alzheimer's disease [1] has long been established. Diagnosis of these conditions however, ofen requires long-term monitoring of the patient's EEG activity. A portable device that monitors EEG activities at home, could allow patients remain at home in comfort ad allow the data to be processed ofine by a technician or classifcation tool. Wireless trasmission of data Digital Object Identifier: 10.410BCSTPERVASIVEHEALTH2010.8861 http://dx.doi.or/10.410BCSTPERVASIVEHEALTH2010.8861 allows for the possibility of near-constant remote monitoring of the patient by a clinical expert. Due to the nature of EEG signals however, even a short period of capture can result in large amounts of data being recorded. This can cause difculties in transmitting the data, particulaly if wireless transmission is being used to permit remote monitoring, as wireless communications can be a signifcant contributor to power consumption in a portable system [2,3]. Therefore, efective compression of the data is important to minimize the amount of information that needs to be transmitted wirelessly. A coexisting goal is that the compression itself needs to be carried out in an efcient manner so as not to unduly add to the power consumption of the device. In comparison to other measures of biomedical electrical activity, such as ECG, there has been relatively little work done in the feld of EEG compression. Of the work that has been done, most have focused on lossless compression, with only a comparative few having tested some form of lossy compression. Lossless compression maintains complete signal integrity in the decompressed signal but this limits the compression ratio (CR) that can be achieved. Lossy compression can achieve much higher CRs but results in a loss of some signal fdelity. Using a slightly lossy codec can achieve signifcantly greater compression, with minimum impact on the integrity of the signal. This work proposes a scheme for lossy EEG compression, based on the JPEG2000 image compression standard, tageted at implementation on an ambulatory device. JPEG2000 was chosen due to its use of efcient compression methods and the choice of lossless or lossy compression, thus allowing a range of trade-ofs between compression ratio, signal fdelity, and computationa l complexity. Work has already been done on low-powered implementations of this algorithm for portable hardwae [4,5], thus indicating that efcient implementation of the core elements of the algorithm is feasible. Section II of the paper describes the JPEG-2000 algorithm, in particular the wavelet transform and arithmetic coder used in the algorithm. Section III describes modifcations made to the algorithms in order to increase EEG compression efciency. Section IV details results of performance evaluation, while Section V presents conclusions.