• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.56 no.11
• /
• pp.1998-2003
• /
• 2007

A flexible but implantable polyimide-based neural implant was fabricated for reliable and stable long-term monitoring of neural activities from brain. The developed neural implant provides 3-dimensional (3D) $3{\times}3$ structure, avoids any hand handling, and makes the insertion more efficient and reliable. Any film curvature caused by residual stress was not observed in the electrode. The 3D flexible polyimide electrode penetrated a dense gel whose stiffness is close to live brain tissue, because a ${\sim}1{\mu}m$ thick nickel was electroplated along the edge of the shank in order to improve the stiffness. The recording sites were positioned at both side of the shank to increase the probability of recording neural signals from a target volume of tissue. Impedance remained stable over 72 hours because of extremely low moisture uptake in the polyimide dielectric layers. At electrical recording test in vitro, the fabricated electrode showed excellent recording performance, suggesting that this electrode has the potential for great recording from neuron firing and long-term implant performance.

• Imaging Science in Dentistry
• /
• v.52 no.2
• /
• pp.219-224
• /
• 2022

Purpose: This study aimed to evaluate the performance of transfer learning in a deep convolutional neural network for classifying implant fixtures. Materials and Methods: Periapical radiographs of implant fixtures obtained using the Superline (Dentium Co. Ltd., Seoul, Korea), TS III(Osstem Implant Co. Ltd., Seoul, Korea), and Bone Level Implant(Institut Straumann AG, Basel, Switzerland) systems were selected from patients who underwent dental implant treatment. All 355 implant fixtures comprised the total dataset and were annotated with the name of the system. The total dataset was split into a training dataset and a test dataset at a ratio of 8 to 2, respectively. YOLOv3 (You Only Look Once version 3, available at https://pjreddie.com/darknet/yolo/), a deep convolutional neural network that has been pretrained with a large image dataset of objects, was used to train the model to classify fixtures in periapical images, in a process called transfer learning. This network was trained with the training dataset for 100, 200, and 300 epochs. Using the test dataset, the performance of the network was evaluated in terms of sensitivity, specificity, and accuracy. Results: When YOLOv3 was trained for 200 epochs, the sensitivity, specificity, accuracy, and confidence score were the highest for all systems, with overall results of 94.4%, 97.9%, 96.7%, and 0.75, respectively. The network showed the best performance in classifying Bone Level Implant fixtures, with 100.0% sensitivity, specificity, and accuracy. Conclusion: Through transfer learning, high performance could be achieved with YOLOv3, even using a small amount of data.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.49 no.12
• /
• pp.249-255
• /
• 2012

A cochlear implant system uses charge-balanced biphasic pulses that are known to reduce tissue damage than monophasic pulses. In this study, we investigated effect of pulse pattern on neural responses using a computer model, based on the Hodgkin-Huxley equation. Electric pulse phase, pulse duration, and phase gap have been systematically varied to characterize auditory nerve responses. The results show that neural responses, dynamic range and threshold are represented as a function of stimulus patterns and duration. The results could greatly extend to develop more efficient cochlear implant stimulation.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.53 no.10
• /
• pp.34-39
• /
• 2016

A low-voltage, low-power analog front-end IC for neural recording implant devices is presented. The proposed IC consists of a low-noise neural amplifier and a programmable active bandpass filter to process neural signals residing in the band of 1 Hz to 5 kHz. The neural amplifier is based on a source-degenerated folded-cascode operational transconductance amplifier (OTA) for good noise performance while the following bandpass filter utilizes a low-power current-mirror based OTA with programmable high-pass cutoff frequencies from 1 Hz to 300 Hz and low-pass cutoff frequencies from 300 Hz to 8 kHz. The total recording analog front-end provides 53.1 dB of voltage gain, $4.68{\mu}Vrms$ of integrated input referred noise within 1 Hz to 10 kHz, and noise efficiency factor of 3.67. The IC is designed using $18-{\mu}m$ CMOS process and consumes a total of $3.2{\mu}W$ at 1-V supply voltage. The layout area of the IC is $0.19 mm^2$.

• JSTS:Journal of Semiconductor Technology and Science
• /
• v.17 no.1
• /
• pp.35-41
• /
• 2017

We propose a novel multi-photodiode array (MPDA) based retinal implant IC with on/off stimulation strategy for a visual prosthesis with improved spatial resolution. An active pixel sensor combined with a comparator enables generation of biphasic current pulses when light intensity meets a threshold condition. The threshold is tuned by changing the discharging time of the active pixel sensor for various light intensity environments. A prototype of the 30-channel retinal implant IC was fabricated with a unit pixel area of $0.021mm^2$, and the stimulus level up to $354{\mu}A$ was measured with the threshold ranging from 400 lx to 13120 lx.

• Journal of Biomedical Engineering Research
• /
• v.34 no.2
• /
• pp.47-54
• /
• 2013

Electrical stimulation is delivered to auditory nerve (AN) through the electrodes in cochlear implant system. Neurogram is a spectrogram that includes information of neural response to electrical stimulation. We hypothesized that the similarity between a neurogram and an input-sound spectrogram could show how well a cochlear implant system works. In this study, we evaluated electrical stimulus configuration of CIS strategy using the computational model. The computational model includes stochastic property and anatomical features of cat auditory nerve fiber. To evaluate similarity between a neurogram and an input-sound spectrogram, we calculated Structural Similarity Index (SSIM). The results show that the dynamic range and the stimulation rate per channel influenced SSIM. Finally, we suggested the optimal configuration within the given stimulus CIS. We expect that the results and the evaluating procedure could be employed to improve the performance of a cochlear implant system.

• Journal of Biomedical Engineering Research
• /
• v.11 no.1
• /
• pp.83-88
• /
• 1990

A New speech processing algorithm using neural networks is proposed. We transform input data into frequency domain and process them by neural networks of 22 output neurons which have Bark scale on the ground that the Bark scale is similiar with that of the characteristics of human cochlea. An utilized neural network is multilayer perceptron, and the characteristics of cochlea have it trained by error back propagation learning algorithm. The trained neural networks suffices functions of human cochlea including the effects of automatic gain control, compression and equalization. Simulation results show that the proposed speech processing algorithm has good performance in automatic gain control, compression and equalization.

• Journal of Biomedical Engineering Research
• /
• v.29 no.6
• /
• pp.415-430
• /
• 2008

The technologies in medical electronic implant(MEI) devices are developing rapidly, and already, there are various kinds of the MEI devices in the current medical equipments market. Recently, the global market scale of MEI devices have been increased about 13% year by year, and the import amount of MEI devices in Korea is increasing rapidly. In the near future, the demands of MEI devices will be magnificently increasing by the continuous development of the biomedical electronics devices which coupled with neural, brain and other organs will bring us to tremendous effects, such as providing new therapeutic solutions to patients, extension and saving human life, and an important clue of medical development. However, the investment of the research and the activity of developments in this field are still very weak in the Korea. Consequently, this paper introduces about the research trends of MEI devices, and technological problems those must be solved, and then concludes with the suggestions in order to be the leading country in this field.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2012.05a
• /
• pp.1197-1198
• /
• 2012

• Journal of rehabilitation welfare engineering & assistive technology
• /
• v.8 no.1
• /
• pp.9-17
• /
• 2014

Electrical compound action potential (ECAP) can be recorded on cochlear implant. This study will investigate stimulation and recording to enhance the efficacy of ECAP. 34 articles was used. We analyzed pulse and stimulating condition, artifact suppression, recording condition. The cathod-leading biphasic pulse was used with as short as possible pulse width and inter phase gap for the efficacy of neural firing, stable threshold and preventing neural degeneration. Around C-level was stimulated to apical, middle and basal turn of cochlea. Artifact was eliminated by forward-masking, template-subtraction technique. For clearer waveform, we need to change distance between stimulating and recording electrode, the gain of amplification, number of average.