Abstract
Personalized health monitoring of neural signals usually results in a very large dataset, the processing and transmission of which require considerable energy, storage, and processing time. We present bioinspired electroceptive compressive sensing (BeCoS) as an approach for minimizing these penalties. It is a lightweight and reliable approach for the compression and transmission of neural signals inspired by active electroceptive sensing used by weakly electric fish. It uses a signature signal and a sensed pseudo-sparse differential signal to transmit and reconstruct the signals remotely. We have used EEG datasets to compare BeCoS with the block sparse Bayesian learning-bound optimization (BSBL-BO) technique—A popular compressive sensing technique used for low-energy wireless telemonitoring of EEG signals. We achieved average coherence, latency, compression ratio, and estimated per-epoch power values that were 35.38%, 62.85%, 53.26%, and 13 mW better than BSBL-BO, respectively, while structural similarity was only 6.295% worse. However, the original and reconstructed signals remain visually similar. BeCoS senses the signals as a derivative of a predefined signature signal resulting in a pseudo-sparse signal that significantly improves the efficiency of the monitoring process. The results show that BeCoS is a promising approach for the health monitoring of neural signals.
Highlights
There is a steady rise in the number of patients with neurological disorders, and it has become the leading cause of deaths and disability-adjusted life-years [1]
personalized health monitoring (PHM) of neural signals has grown in popularity, in part, due to the rising number of people with neurological disorders and the dwindling budget for healthcare
The sensing, compression, and reconstruction of neural signals incur a heavy penalty in terms of metrics such as energy, memory, and processing time, affecting the system’s capacity for lightweight processing
Summary
There is a steady rise in the number of patients with neurological disorders, and it has become the leading cause of deaths and disability-adjusted life-years [1]. EEG signals are usually monitored with multichannel systems that may have up to 256 channels This can result in a very large dataset that takes up significant storage space and consumes considerable power for the sensing, processing, and transmission phases. The BSBL-BO and BeCoS techniques are compared with respect to coherence, latency, compression ratio, power, and structural similarity. The coherence compared the frequency content of the original and transmitted signals and BeCoS showed an improvement of 35.38% when compared to BSBL-BO. The compression ratio showed the capacity of both techniques to sparsify the EEG signal, while the power metric analyzed the energy consumption. BeCoS showed an improvement of 53.26% for the compression ratio, and it utilized 13 mW less in processing each epoch of EEG data.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have