IEEE Biomedical Circuits and Systems Conference : healthcare technology : [proceedings]. IEEE Biomedical Circuits and Systems Conference最新文献

Electrical capacitance tomography (ECT) is a non-optical imaging technique in which a map of the interior permittivity of a volume is estimated by making capacitance measurements at its boundary and solving an inverse problem. While previous ECT demonstrations have often been at centimeter scales, ECT is not limited to macroscopic systems. In this paper, we demonstrate ECT imaging of polymer microspheres and bacterial biofilms using a CMOS microelectrode array, achieving spatial resolution of 10 microns. Additionally, we propose a deep learning architecture and an improved multi-objective training scheme for reconstructing out-of-plane permittivity maps from the sensor measurements. Experimental results show that the proposed approach is able to resolve microscopic 3-D structures, achieving 91.5% prediction accuracy on the microsphere dataset and 82.7% on the biofilm dataset, including an average of 4.6% improvement over baseline computational methods.

In this paper we present spatio-temporally controlled electrochemical stimulation of aqueous samples using an integrated CMOS microelectrode array with 131,072 pixels. We demonstrate programmable gold electrodeposition in arbitrary spatial patterns, controllable electrolysis to produce microscale hydrogen bubbles, and spatially targeted electrochemical pH modulation. Dense spatially-addressable electrochemical stimulation is important for a wide range of bioelectronics applications.

Brain computer interfaces (BCIs) provide clinical benefits including partial restoration of lost motor control, vision, speech, and hearing. A fundamental limitation of existing BCIs is their inability to span several areas (> cm²) of the cortex with fine (<100 μm) resolution. One challenge of scaling neural interfaces is output wiring and connector sizes as each channel must be independently routed out of the brain. Time division multiplexing (TDM) overcomes this by enabling several channels to share the same output wire at the cost of added noise. This work leverages a 130-nm CMOS process and transfer printing to design and simulate a 384-channel actively multiplexed array, which minimizes noise by adding front end filtering and amplification to every electrode site (pixel). The pixels are 50 μm × 50 μm and enable recording of all 384 channels at 30 kHz with a gain of 22.3 dB, noise of 9.57 μV rms, bandwidth of 0.1 Hz - 10 kHz, while only consuming 0.63 μW/channel. This work can be applied broadly across neural interfaces to create high channel-count arrays and ultimately improve BCIs.

Feature selection, or dimensionality reduction, has become a standard step in reducing large-scale neural datasets into usable signals for brain-machine interface and neurofeedback decoders. Current techniques in fMRI data reduce the number of voxels (features) by performing statistics on individual voxels or using traditional techniques that utilize linear combinations of features (e.g., principal component analysis (PCA)). However, these methods often do not account for the cross-correlations found across voxels and do not sufficiently reduce the feature space to support efficient real-time feedback. To overcome these limitations, we propose using factor analysis on fMRI data. This technique has become increasingly popular for extracting a minimal number of latent features to explain high-dimensional data in non-human primates (NHPs). Here, we demonstrate these methods in both NHP and human data. In NHP subjects (n=2), we reduced the number of features to an average of 26.86% and 14.86% of the total feature space to build our multinomial classifier. In one NHP subject, the average accuracy of classifying eight target locations over 64 sessions was 62.43% (+/-6.19%) compared to a PCA-based classifier with 60.26% (+/-6.02%). In healthy fMRI subjects, we reduced the feature space to an average of 0.33% of the initial space. Group average (n=5) accuracy of FA-based category classification was 74.33% (+/- 4.91%) compared to a PCA-based classifier with 68.42% (+/-4.79%). FA-based classifiers can maintain the performance fidelity observed with PCA-based decoders. Importantly, FA-based methods allow researchers to address specific hypotheses about how underlying neural activity relates to behavior.

Imaging of spinal cord microvasculature holds great potential in directing critical care management of spinal cord injury (SCI). Traditionally, contrast agents are preferred for imaging of the spinal cord vasculature, which is disadvantageous for long-term monitoring of injury. Here, we present FlowMorph, an algorithm that uses mathematical morphology techniques to segment non-contrast Doppler-based videos of rat spinal cord. Using the segmentation, it measures single-vessel parameters such as flow velocity, rate, and radius, with visible cardiac cycles in individual vessels showcasing the spatiotemporal resolution. The segmentation outlines vessels well with little extraneous labeling, and outlines are smooth through time. Radius measurements of perforating vessels are similar to what is seen in the literature through other methods. Verification of the algorithm through comparison to manual measurement and in vitro microphantom standards highlights points of future improvement. This method will be vital for future work studying the vascular effects of SCI and can be adopted to other species as well.

In extracellular neural electrophysiology, individual spikes have to be assigned to their cell of origin in a procedure called "spike sorting". Spike sorting is an unsupervised problem, since no ground-truth information is generally available. Here, we focus on improving spike sorting performance, particularly during periods of high synchronous activity or so-called "bursting". Bursting entails systematic changes in spike shapes and amplitudes and remains a challenge for current spike sorting schemes. We use realistic simulated bursting recordings of high-density micro-electrode arrays (HD-MEAs) and we present a fully automated algorithm based on template matching with a focus on recovering missed spikes during bursts. To compare and benchmark spike-sorting performance after applying our method, we used ground-truth information of simulated recordings. We show that our approach can be effective in improving spike sorting performance during bursting. Further validation with experimental recordings is necessary.

Neural interfaces with increasing channel counts require a scalable means of testing. While multiplexed potentiostats have long been the solution to this problem, most have been dedicated to one specific probe design or potentiostat, limited in the electrochemical techniques available, inordinately expensive, or they support multiplexing of too few channels. We present the design of an automated multiplexed potentiostat system that addresses these limitations-it is easily generalizable to any probe and potentiostat, supports any electrochemical technique available with the potentiostat, is low-cost, and can readily be expanded to hundreds of channels with support for multiple simultaneous potentiostats. This paper discusses the design philosophy and architecture of our 512-channel, 4-potentiostat system before demonstrating functionality with electrochemical impedance spectroscopy data, cyclic voltammetry curves, and an example of electrochemical surface modification, all on functional implantable microelectrode arrays currently being used for in vivo electrophysiological studies. Finally, we discuss the limitations to some sensitive or high-frequency impedance measurements due to reactive parasitics.

This paper presents a multi-antenna external reader system that enables orientation insensitive communication with implantable medical devices (IMDs) for wireless biomedical applications. The proposed system consists of a circular array with six loop antennas. The antenna placement and orientations are determined by analyzing the near-field magnetic field variations of the loop antenna. The proposed system is first simulated using HFSS electromagnetic simulation software. Our simulations show that the received power at the proposed external reader with six antennas only varies about 5 dB for any given orientation of the implanted antenna, which is highly significant compared to the 20-35 dB variation with a single external antenna. Here, we select the antenna which provides the largest coupling between the IMD to receive/transmit signals. A prototype of the proposed multi-antenna external reader is then implemented using custom-designed PCBs that interconnect loop antennas, transceiver ICs, and commercially-available circuit components. A custom PCB with a miniaturized loop antenna is used to emulate an implantable device. Based on measurement results, the received power in the external reader only varies about 3 dB when the miniaturized antenna rotates with respect to the x-axis. These measurements show good agreement with the simulated reader.

Multi-access networking with miniaturized wireless implantable devices can enable and advance closed-loop medical applications to deliver precise diagnosis and treatment. Using ultrasound (US) for wireless implant systems is an advantageous approach as US can beamform with high spatial resolution to efficiently power and address multiple implants in the network. To demonstrate these capabilities, we use wirelessly powered mm-sized implants with bidirectional communication links; uplink data communication measurements are performed using time, spatial, and frequency-division multiplexing schemes in tissue phantom. A 32-channel linear transmitter array and an external receiver are used as a base station to network with two implants that are placed 6.5 cm deep and spaced less than 1 cm apart. Successful wireless powering and uplink data communication around 100 kbps with a measured bit error rate below 10^-4 are demonstrated for all three networking schemes, validating the multi-access networking feasibility of US wireless implant systems.

Adaptive deep brain stimulation (aDBS) is an emerging method to alleviate the side effects and improve the efficacy of conventional open-loop stimulation for movement disorders. However, current adaptive DBS techniques are primarily based on single-feature thresholding, precluding an optimized delivery of stimulation for precise control of motor symptoms. Here, we propose to use a machine learning approach for resting-state tremor detection from local field potentials (LFPs) recorded from subthalamic nucleus (STN) in 12 Parkinson's patients. We compare the performance of state-of-the-art classifiers and LFP-based biomarkers for tremor detection, showing that the high-frequency oscillations and Hjorth parameters achieve a high discriminative performance. In addition, using Kalman filtering in the feature space, we show that the tremor detection performance significantly improves (F_(1,15)=32.16, p<0.0001). The proposed method holds great promise for efficient on-demand delivery of stimulation in Parkinson's disease.