BROWSE

ITEM VIEW & DOWNLOAD

Multi-Biosignal Sensing Interface with Direct Sleep-Stage Classification

Cited 0 times inthomson ciCited 0 times inthomson ci
Title
Multi-Biosignal Sensing Interface with Direct Sleep-Stage Classification
Author
Kim, Sung-Woo
Advisor
Kim, Jae Joon
Keywords
sleep-classification; multi-biosignal interface; rule-based decision tree; feature extraction stage; headband; wearable device
Issue Date
2020-02
Publisher
Graduate School of UNIST
Abstract
Sleep is a time of mental and physical rest in a person’s daily cycle. It is an indispensable metabolic activity that helps the body grow and boosts immunity. Therefore, sleep disorders can cause illness in the body as well as just physical condition. Among these diseases are typically included rapid eye movement (REM) sleep behavior disorder, nocturnal enuresis, sleepwalking, etc., which can cause serious injury on sleep. Sleep disorders are a common disease. According to a survey, sleep deprivation and disability affect a significant part of the world’s population. It is a disease that affects tens of millions of people around the world. In general, treatment for sleep disorders checks the quality during sleep and prescribes various sleep diseases by checking the condition of sleep. Sleep quality and sleep disease are determined by the depth and time of the sleep phase. Therefore, the analysis and classification of the sleep stage are essential. According to the manual of the American Academy of Sleep Medicine (AASM), sleep stages are divided into five stages. Various methods for sleep analysis have been developed. Polysomnography (PSG), called the golden standard, is the most reliable measurement of sleep quality in hospitals for sleep disorders, but this conventional method requires the use of various human body signals, and it is difficult to access due to the complex interface and various electrodes. It is not economical because of its infrastructure, which does not lead to direct treatment of prospective patients. In addition, the conventional interface system process is not an integrated interface system. The integrated interface system refers to the integration of the interface in the measuring and analysis process. Conventional sensing and analysis take place on the instrument measuring the patient and on the analyst’s computer. Therefore, conventional treatments are not economical and make patient self-analysis difficult. Furthermore, this makes it difficult to increase the demand for prospective patients. This paper presents a multi-biosignal sensing interface system with direct sleep-stage classification. Unlike conventional systems, this work proposes an interface system that is an integrated interface system, measuring, and analysis based on the analog circuit and system. The proposed paper configures a multi-biosignal sensing interface consisting of single-channel EEG, EMG, and 2EoG. The multi-biosignal sensing readout integrated circuit (ROIC) collects analog signals from the electrodes and extracts features from the signal. The multi-biosignal sensing ROIC has a feature extraction stage that directly extracts the characteristic of sleep stages. The analog feature extraction stage consists of the optimized circuit for three multi-biosignal extracts the feature of each stage during sleep on the waveform. The extracted signal is scored by the rule-based decision tree sleep stage proposed by the micro controller unit (MCU). The multi-biosignal sensing ROIC can analyze the sleep stage through EEG, EMG, and 2EOG, and can simultaneously analyze four channels. The multi-biosignal sensing ROIC is implemented using a compensate metal-oxide-semiconductor 0.18um process. In addition, this system implements a low-power, integrated module for portable device configuration, and from this interface makes a smart headband for prospective patients. Depending on the purpose of use, it consists of 2 type paths, including raw data recording and analog feature extraction based direct sleep classification using decision tree algorithm. Finally, sleep stage scoring can be displayed, or raw data can be sent to the personal computer interface to increase accuracy. The sleep stage was verified by comparing the OpenBCI module-based MATLAB analysis using SVM with this system, and the result shows an overall accuracy of 74% for four sleep stages.
Description
Department of Electrical Engineering
URI
Go to Link
Appears in Collections:
EE_Theses_Ph.D.
Files in This Item:
Multi-Biosignal Sensing Interface with Direct Sleep-Stage Classification.pdf Download

find_unist can give you direct access to the published full text of this article. (UNISTARs only)

Show full item record

qrcode

  • mendeley

    citeulike

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

MENU