Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference最新文献

Tidal volume can be estimated using the surface motions of the upper body induced by respiration. However, the precision and instrumentation of such estimation must be improved to allow widespread application. In this study, respiration induced changes in parameters that can be recorded with inertial measurement units are examined to determine tidal volumes. Based on the data of an optical motion capture system, the optimal positions of inertial measurement units (IMU) in a smart shirt for sets of 4, 5 or 6 sensors were determined. The errors observed indicate the potential to determine tidal volumes using IMUs in a smart shirt.Clinical Relevance- The measurement of respiratory volumes via a low-cost and unobtrusive smart shirt would be a major advance in clinical diagnostics. In particular, conventional methods are expensive, and uncomfortable for conscious patients if measurement is desired over an extended period. A smart-shirt based on inertial sensors would allow a comfortable measurement and could be used in many clinical scenarios - from sleep apnoea monitoring to homecare and respiratory monitoring of comatose patients.

The 12-lead ECG only has 8 independent ECG leads, which leads to diagnostic redundancy when using all 12 leads for heart arrhythmias classification. We have previously developed a deep learning (DL)-based computer-interpreted ECG (CIE) approach to identify an optimal 4-lead ECG subset for classifying heart arrhythmias. However, the clinical diagnostic criteria of cardiac arrhythmia types are often lead-specific, so this study is going to explore the selection of arrhythmia-based ECG-lead subsets rather than one general optimal ECG-lead subset, which could improve the classification performance for the CIE. The DL-based CIE model previously developed was used to learn 4 common types of heart arrhythmias (LBBB, RBBB, AF, and I-AVB) for identifying corresponding optimal ECG-lead subsets. A public dataset that splits into training (approx. 70%), validation (approx. 15%), and test (approx. 15%) sets from the PhysioNet Cardiology Challenge 2020 was used to explore the study. The results demonstrated that the DL-based CIE model identified an optimal ECG-lead subset for each arrhythmia: I, II, aVR, aVL, V1, V3, and V5 for I-AVB; I, II, aVR, and V3 for AF; I, II, aVR, aVF, V1, V3, and V4 for LBBB; and I, II, III, aVR, V1, V4, and V6 for RBBB. For each arrhythmia classification, the DL-based CIE model using the optimal ECG-lead subset significantly outperformed the model using the full 12-lead ECG set on the validation set and on the external test dataset.The results support the hypothesis that using an optimal ECG-lead subset instead of the full 12-lead ECG set can improve the classification performance of a specific arrhythmia when using the DL-based CIE approach.Clinical Relevance- Using an arrhythmia-based optimal ECG-lead subset, the classification performance of a deep-learning-based model can be achieved without loss of accuracy in comparison with the full 12-lead set (p<0.05).

An integrated circuit specified for tinnitus treatment is described. This chip, realized using a 0.18um BCD high-voltage CMOS process, is capable of generating current stimulus with any wave shape directly into the inner-ear tissue without the need for off-chip control circuitry. Used as part of a multi-chip module that can be implanted into the inner ear, this core chip contains an 8-bit digital-to-analog converter, an amplitude control block, a novel high-voltage drive and charge balance circuit, a high-voltage level shifter, an SRAM, a ROM, and an on-chip central control unit. The chip can achieve ±0.1 mV charge-balance precision.

Wearable sensors have become increasingly popular in recent years, with technological advances leading to cheaper, more widely available, and smaller devices. As a result, there has been a growing interest in applying machine learning techniques for Human Activity Recognition (HAR) in healthcare. These techniques can improve patient care and treatment by accurately detecting and analyzing various activities and behaviors. However, current approaches often require large amounts of labeled data, which can be difficult and time-consuming to obtain. In this study, we propose a new approach that uses synthetic sensor data generated by 3D engines and Generative Adversarial Networks to overcome this obstacle. We evaluate the synthetic data using several methods and compare them to real-world data, including classification results with baseline models. Our results show that synthetic data can improve the performance of deep neural networks, achieving a better F₁-score for less complex activities on a known dataset by 8.4% to 73% than state-of-the-art results. However, as we showed in a self-recorded nursing activity dataset of longer duration, this effect diminishes with more complex activities. This research highlights the potential of synthetic sensor data generated from multiple sources to overcome data scarcity in HAR.

Gastric Intestinal Metaplasia (GIM) is one of the precancerous conditions in the gastric carcinogenesis cascade and its optical diagnosis during endoscopic screening is challenging even for seasoned endoscopists. Several solutions leveraging pre-trained deep neural networks (DNNs) have been recently proposed in order to assist human diagnosis. In this paper, we present a comparative study of these architectures in a new dataset containing GIM and non-GIM Narrow-band imaging still frames. We find that the surveyed DNNs perform remarkably well on average, but still measure sizeable inter-fold variability during cross-validation. An additional ad-hoc analysis suggests that these baseline architectures may not perform equally well at all scales when diagnosing GIM.Clinical relevance- Enhanching a clinician's ability to detect and localize intestinal metaplasia can be a crucial tool for gastric cancer management policies.

This paper presents ultra-low power photoplethysmography (PPG) readout circuits. The proposed system architecture uses a current buffer between the photodiode (PD) and the transimpedance amplifier (TIA) to isolate the large parasitic capacitance of the PD leading to improves the power consumption of the TIA. A class AB topology is exploited at the output of the amplifier, which allows for increased drive capability without the use of auxiliary circuits. The maximum input current range of the TIA is 160 µA, so the large DC current of the input signal does not saturate the circuit. In the LED driver circuit, by varying the duty cycle of a pulse wave modulation (PWM) signal, the ON and OFF times of the circuits. The amplifier and LED driver are manufactured in the 130 nm TSMC CMOS process. The power consumption of the circuits with a duty cycle of 1% is 3.28 µW (at VDD = 1.2V).Clinical Relevance- Vital signs are becoming a very important research topic due to the recent prevalence of COVID-19 and other respiratory diseases. This research aims to develop and interface circuits to monitor vital signs including blood pressure, heart rate, and respiratory rate to study respiratory disease, drug safety, and efficacy.

Lumbar spinal disc herniation is a disease in which the protruding nucleus pulposus presses on the nerve due to actions that place loads on the disc, causing pain in the lower back and lower limbs. About 80% of treatments of disc herniation are conservative treatments, and although it is necessary to live with pain for a long time, there have been no studies that clearly define the relationship between pain and biomechanical parameters. In this study, we proposed a method of identifying biomechanical parameters that predict posture-related pain in patients with lumbar spinal disc herniation. The pain values were quantitatively evaluated by the Numerical Rating Scale (NRS) and the biomechanical parameters were analyzed by OpenSim. Lasso regression was performed to narrow down the biomechanical parameters that were related to pain and derive the mathematical model of the relationship. Therefore, many of the parameters of the obtained mathematical model were related to the lumbar spine and were consistent with areas that be related to lumbar spinal disc herniation.

Awake Surgery (AS) is considered the best treatment for brain tumors located in or near eloquent areas. During this intervention, Direct Electrical Stimulations (DES) are delivered by the surgeon on the patient's brain in order to obtain an accurate brain mapping of the patient. The patient is asked to perform various tasks (e.g. counting, object naming, emotion recognition) through neuropsychological tests during these stimulations. These DES may cause a reversible lesion inducing deficits on the patient which can be observed during these tasks by the medical staff. The resection is then performed or not according to the patient's response. The intraoperative deficits can take several forms and can be difficult to analyze and identify. The development of new solutions allowing the automatic detection of these deficits could be therefore essential. However, still today, no structured and organized AS dedicated database is available that could be used to train and test these algorithms. We propose a modular system allowing the synchronized multimodal acquisition of various information including physiological measurements, DES signals and parameters, and task-related data to create such database.Clinical relevance- Acquiring synchronized multimodal data during AS will allow the creation of a dedicated database that could then be used to reveal new correlations between DES and the patient's response, and to develop and test new algorithms for the automatic detection of deficits.

Within this paper, we demonstrate the feasibility of the FPGA implementation as well as the 180nm CMOS circuit design of a particular biologically plausible supervised learning algorithm (ReSuMe). Based on the Spike-Timing-Dependent Plasticity (STDP) learning phenomenon, this design proposes a fully configurable implementation of STDP learning window function to adjust the learning process for different applications, optimizing results for each use case. The CMOS implementation in 180nm technology node supplied with 1.8V shows a core area of 0.78mm² and verifies the suitability of an on-chip ReSuMe learning algorithm implementation and its capability of integration with a multitude of external and already designed structures of Spiking Neural Networks (SNNs).

A physical system to generate a PPG-mimicking signal was designed and validated using everyday low-cost components to aid in medical sensor design. The pulse waveform was created by driving a working fluid into a silicone tube and changing the pressure within it. The corresponding waveform mimics a PPG signal through an artery, is adaptable, and repeatable. The working fluid is interchangeable allowing for change of blood analyte concentrations for development and testing of PPG-based sensors. The system was validated by black ink water compared to water and air compared to water testing to confirm optical transparency of the tube. The produced PPG signal, pulse rate and pressure change were compared to that seen in subjects. Optical transparency for 660 nm - 1550 nm wavelengths of light was validated with the signal, pulse rate and total compliance matching subject data. Thus, the system can mimic arterial pulses, creating a valid PPG signal that can be detected by PPG-based sensors.Clinical Relevance- Provides a low-cost, adaptable, physical PPG signal generator for research and development of optical medical sensor technologies.