Hardware/software implementation of the EEG signal compression module for an ambulatory monitoring subsystem

Abstract

This paper presents the design and evaluation of a signal compression subsystem for an ambulatory EEG system. Ambulatory systems require low power consumption and offer a significant reduction of the patient's incommodity during the EEG signal monitori ng process. Hardware/Software co-design techniques were applied during the design of the EEG signal compression subsystem. A methodology for evaluating the efficiency of the hardware/software partitioning was applied to optimize the power consumption and to minimize the complexity of the hardware portion of the algorithm. The results show a significant reduction in the total execution time of the compression algorithm and a moderate reduction on the power consumption of the subsystem.

Wearable Electroencephalography

“…These levels are impressive, and the schemes should certainly be used where possible. However, given the results from [50] discussed previously, it is clear that the implementation of suitable algorithms at the low-power levels required will not be a trivial task.…”

“…Few EEG data compression papers provide both a data reduction figure and a power consumption figure. Avila et al [50] presents a direct cosine transform-based EEG compression algorithm but with a minimum power consumption for a hardware/software codesign implementation of 71 mW: orders of magnitude over the microwatt power budget. The question now is what compression techniques are available that can operate within this power budget, and what is their impact on the overall EEG system design?…”

Multichannel evoked neural signal compression using advanced video compression algorithm

“…This is because that the numerical range of neural signals differs from that of video signals. However, the precision of both signals is similar-8 bits [8] [9]. We transform the neural signal to 0-255 with an average of 128 by using a linear transform so that the video compression algorithm can be applied to it.…”

ElectroCorticoGraphy (ECoG) acquisition exploiting signal characteristics for reduced power

“…Prior art exists on reversible compression of electroencephalography (EEG) signals, which are similar to ECoG. Most are software oriented with no mention of power overhead [6][7]; one requires a power overhead of 2 mW per channel [8], infeasible in wireless implants; and one proposes an analog method with a theoretical overhead of 0.7 µW per channel [9] but has not actually been implemented in hardware.…”