Sastra Kusuma Wijaya

Sleep stages classification using shallow classifiers

A person with sleep disorder such as apnea will stop breathing for a while during sleep. If frequently occurs, sleep disorder is dangerous for health. An early step for diagnosing apnea is by classifying the sleep stages during sleep. This study explores some shallow classifiers and their feasibility applied to sleep data. Recently, a sleep stages classification system that use deep unsupervised features learning representations have been proposed [9]. In our view, an adequate study on this problem using shallow classifiers still need to be investigated. This study, using some of the data on [9], focuses on evaluating some shallow classifier to the sleep stages classification problem. This study evaluates five classifiers: SVM, Neural Network, Classification Tree, k-Nearest Neighborhood (k-NN), and Naive Bayes. Experiment result shows that neural network gives best performance for sleep stage classification problem. Compared to the SVM (the 2-nd rank of accuracy on S000 data), the neural network is also more efficient than SVM in term of computational time and memory requirement.

Ischemic stroke identification based on EEG and EOG using ID convolutional neural network and batch normalization

In 2015, stroke was the number one cause of death in Indonesia. The majority type of stroke is ischemic. The standard tool for diagnosing stroke is CT-Scan. For developing countries like Indonesia, the availability of CT-Scan is very limited and still relatively expensive. Because of the availability, another device that potential to diagnose stroke in Indonesia is EEG. Ischemic stroke occurs because of obstruction that can make the cerebral blood flow (CBF) on a person with stroke has become lower than CBF on a normal person (control) so that the EEG signal have a deceleration. On this study, we perform the ability of ID Convolutional Neural Network (1DCNN) to construct classification model that can distinguish the EEG and EOG stroke data from EEG and EOG control data. To accelerate training process our model we use Batch Normalization. Involving 62 person data object and from leave one out the scenario with five times repetition of measurement we obtain the average of accuracy 0.86 (F-Score 0.861) only at 200 epoch. This result is better than all over shallow and popular classifiers as the comparator (the best result of accuracy 0.69 and F-Score 0.72). The feature used in our study were only 24 handcrafted feature with simple feature extraction process.

Design and development of electrical impedance tomography system with 32 electrodes and microcontroller

Electrical Impedance Tomography (EIT) is an imaging method that is able to estimate electrical impedance distribution inside an object. This EIT system is developed by using 32 electrodes and microcontroller based module. From a pair of electrodes, sinusoidal current of 3 mA is injected and the voltage differences between other pairs of electrodes are measured. Voltage measurement data are then sent to MATLAB and EIDORS software; the data are used to reconstruct two dimensions image. The system can detect and determine the position of a phantom in the tank. The object’s position is accurately reconstructed and determined with the average shifting of 0.69 cm but object’s area cannot be accurately reconstructed. The object’s image is more accurately reconstructed when the object is located near to electrodes, has a larger size, and when the current injected to the system has a frequency of 100 kHz or 200kHz.Electrical Impedance Tomography (EIT) is an imaging method that is able to estimate electrical impedance distribution inside an object. This EIT system is developed by using 32 electrodes and microcontroller based module. From a pair of electrodes, sinusoidal current of 3 mA is injected and the voltage differences between other pairs of electrodes are measured. Voltage measurement data are then sent to MATLAB and EIDORS software; the data are used to reconstruct two dimensions image. The system can detect and determine the position of a phantom in the tank. The object’s position is accurately reconstructed and determined with the average shifting of 0.69 cm but object’s area cannot be accurately reconstructed. The object’s image is more accurately reconstructed when the object is located near to electrodes, has a larger size, and when the current injected to the system has a frequency of 100 kHz or 200kHz.

Automatic detection of ischemic stroke based on scaling exponent electroencephalogram using extreme learning machine

Stroke is one of cerebrovascular diseases caused by the obstruction of blood flow to the brain. Stroke becomes the leading cause of death in Indonesia and the second in the world. Stroke also causes of the disability. Ischemic stroke accounts for most of all stroke cases. Obstruction of blood flow can cause tissue damage which results the electrical changes in the brain that can be observed through the electroencephalogram (EEG). In this study, we presented the results of automatic detection of ischemic stroke and normal subjects based on the scaling exponent EEG obtained through detrended fluctuation analysis (DFA) using extreme learning machine (ELM) as the classifier. The signal processing was performed with 18 channels of EEG in the range of 0-30 Hz. Scaling exponents of the subjects were used as the input for ELM to classify the ischemic stroke. The performance of detection was observed by the value of accuracy, sensitivity and specificity. The result showed, performance of the proposed method to classify the ischemic stroke was 84 % for accuracy, 82 % for sensitivity and 87 % for specificity with 120 hidden neurons and sine as the activation function of ELM.

Capacitance-digital and impedance converter as electrical tomography measurement system for biological tissue

This research aims to explore the feasibility of capacitance-digital converter and impedance converter for measurement module in electrical capacitance tomography (ECT) system. ECT sensor used was a cylindrical sensor having 8 electrodes. Absolute capacitance measurement system based on Sigma Delta Capacitance-to-Digital-Converter AD7746 has been shown to produce measurement with high resolution. Whereas, capacitance measurement with wide range of frequency is possible using Impedance Converter AD5933. Comparison of measurement accuracy by both AD7746 and AD5933 with reference of LCR meter was evaluated. Biological matters represented in water and oil were treated as object reconstructed into image using linear back projection (LBP) algorithm.

Data acquisition instrument for EEG based on embedded system

An electroencephalogram (EEG) is a device for measuring and recording the electrical activity of brain. The EEG data of signal can be used as a source of analysis for human brain function. The purpose of this study was to design a portable multichannel EEG based on embedded system and ADS1299. The ADS1299 is an analog front-end to be used as an Analog to Digital Converter (ADC) to convert analog signal of electrical activity of brain, a filter of electrical signal to reduce the noise on low-frequency band and a data communication to the microcontroller. The system has been tested to capture brain signal within a range of 1-20 Hz using the NETECH EEG simulator 330. The developed system was relatively high accuracy of more than 82.5%. The EEG Instrument has been successfully implemented to acquire the brain signal activity using a PC (Personal Computer) connection for displaying the recorded data. The final result of data acquisition has been processed using OpenBCI GUI (Graphical User Interface) based thro...

Electroencephalogram analysis with extreme learning machine as a supporting tool for classifying acute ischemic stroke severity

Stroke is one of the highest causes of death in adults and disability in Indonesia, even in the world. Therefore, it is necessary to diagnose stroke in the early stage and give accurate prognosis assessment to improve stroke management. This study tried to automatically classify AIS severity based on EEG signals by using digital signal processing such as Wavelet transform and feedforward type of neural network with ELM algorithm. In this study, Delta Alpha Ratio (DAR), (Delta+Theta)/(Alpha+Beta) Ratio (DTABR) and Brain Symmetry Index (BSI)s value were used as the ELM input feature score, which were obtained by using Wavelet transformation (Daubechies 4) and Welchs method to classify the acute ischemic stroke severity which refers to the National Institutes of Health Stroke Scale (NIHSS). It had shown that the performance of system test accuracy, the sensitivity and specificity were above 72%. These results were useful for classifying AIS based on EEG signals.

Decision support system prototype on obstetrics ultrasonography for primary service physicians

Introduction. The National Social Security System (SJSN) prioritizes primary service as a spearhead to assist Primary Care Physicians to make medical decisions. The purpose of this research is to develop computer software that will assist primary care physicians in the fields of Obstetrics Ultrasonography, related to referral decision-making abilities. Methods. A quasi-experimental post-test only design without a control group. The stages of the research process: Systems Analysis and Design, Prototyping and Testing by Lecture, Students, Programmers and Doctors. Results. From Analysis and Systems design document, has been produced prototype of software, and a test run has been proven successful Decision Support System software helping doctors develop diagnosis and specialty referrals. Conclusion: The Decision Support System software can be used in Obstetrics Ultrasonography by Primary Care Physicians to provide aid in their diagnosis and referrals. Before it is used, it is recommended for trainings and application tests.