Gun-Ki Lee

Design of filter to reject motion artifact of pulse oximetry

Abstract The performance of the pulse oximeter, an instrument which non-invasively measures oxygen saturation in blood, is influenced by a patients respiration or movement. Generally, the frequency band of the motion artifact signal generated from the patients movement is overlapped with that of the patients pulse wave which is measured by the received light sensor. Accordingly, it is difficult to filter the motion artifact of the pulse wave by using a filter which has a cut-off frequency that is fixed. In this study, the motion artifact is removed using the filter bank and the matched filter. Compared with traditional adaptive filter methods, the ratio variation is 50% lower than that of the moving average filter, allowing more stable measurement of oxygen saturation despite the patients movement.

A study on lung nodule detection using neural networks

In this study, the authors developed a method for disease detection using an artificial neural network and digital image processing of a chest radiograph. In a conventional physical examination radiologists check the chest image projected on a viewing box by a magnifying glass and determine what the disease is. The detection of disease on X-ray fluoroscopy images is tedious and time-consuming for humans. This lowers the efficiency for chest diagnosis as many mistakes by the radiologist are caused because of the need to detect micropathology from a film of small size. So, the authors propose a method to quickly find out what the object on a chest radiograph is. This method comprises the functions of image sampling, median filter, neural network image equalizer and neural network pattern recognition. The authors confirm that this method has improved the problems of conventional methods.

A Study on the Evaluation of Irradiance and Radiance risk for LED Luminaires

Electrical safety, optical properties, and electromagnetic safety evaluation of LED applications are under way, but there is insufficient evaluation on the effect of radiance and irradiance on the human body. The visible region of the human eye is 400～700nm. In this study, we classify the LED luminaires in the 300～700nm blue light region as exemption group and risk group 1. Exemption groups have no effect on the human body, but when the risk rating is assessed as risk 1 or risk 2, the effects on the human body may be particularly dangerous because of the possible weakening or loss of vision and the rapid acceleration of age-related macular degeneration. The results of this study will be used to inform about the necessity of risk assessment of LED luminaires products that are about to be released to the market, and to create a system that can protect public health and public safety.

Study on the Detection the Section with the Motion Artifacts of the PPG Signal

Abstract: The biomedical information (heart rate, HRV, SpO 2 etc) from the Photoplethysmography (PPG) signal sec-tion can be estimated without the motion artifacts if the PPG signal section in the motion artifacts is detected accu-rately. But the PPG signal is easily exposed to the motion artifacts by the decrease of the medical instrument sizeand applying as portable or wearable. Besides, it is difficult to completely eliminate the motion artifacts from the PPGsignal without distortion and ensure reliability as well. In this paper, the method was suggested to determine themotion artifacts or not on the PPG signal of the section divided into intervals of constant length. By comparing thespectrum of each section, it can be determined whether the motion artifacts are or not after obtaining the spectrumof each section by the Goertzel algorithm. Moreover, an amount of computation while maintaining a high detectionrate can be reduced by using the Goertzel algorithm.Key words: Photoplethysmography (PPG), Goertzel algorithm, Motion artifacts

A design of adaptive neural filter banks with filter neuron

In this study, we propose that the new filter bank that is an adaptive filter bank using neural networks in time domain. Also, we propose a filter neuron as band pass filter (BPF) with hamming window, the structure and algorithm for filter banks. The performance of neural filter banks is shown from two examples. It shows its characteristics such as the simple structure and a higher speed processing compared to traditional methods (filter banks in frequency domain, etc.). In many applications, the proposed method will provide a higher performance to feature detection of signals in time domain.