RT-NET: real-time reconstruction of neural activity using high-density electroencephalography

AbstractMyoelectric hand prostheses offer a way for upper-limb amputees to recover gesture and prehensile abilities to ease rehabilitation and daily life activities. However, studies with prosthesis users found that a lack of intuitiveness and ease-of-use in the human-machine control interface are among the main driving factors in the low user acceptance of these devices. This paper proposes a highly intuitive, responsive and reliable real-time myoelectric hand prosthesis control strategy with an emphasis on the demonstration and report of real-time evaluation metrics. The presented solution leverages surface high-density electromyography (HD-EMG) and a convolutional neural network (CNN) to adapt itself to each unique user and his/her specific voluntary muscle contraction patterns. Furthermore, a transfer learning approach is presented to drastically reduce the training time and allow for easy installation and calibration processes. The CNN-based gesture recognition system was evaluated in real-time with a group of 12 able-bodied users. A real-time test for 6 classes/grip modes resulted in mean and median positive predictive values (PPV) of 93.43% and 100%, respectively. Each gesture state is instantly accessible from any other state, with no mode switching required for increased responsiveness and natural seamless control. The system is able to output a correct prediction within less than 116 ms latency. 100% PPV has been attained in many trials and is realistically achievable consistently with user practice and/or employing a thresholded majority vote inference. Using transfer learning, these results are achievable after a sensor installation, data recording and network training/fine-tuning routine taking less than 10 min to complete, a reduction of 89.4% in the setup time of the traditional, non-transfer learning approach.

Download Full-text

Modeling pedestrian emotion in high-density cities using visual exposure and machine learning: Tracking real-time physiology and psychology in Hong Kong

Building and Environment ◽

10.1016/j.buildenv.2021.108273 ◽

2021 ◽

Vol 205 ◽

pp. 108273

Author(s):

Luyao Xiang ◽

Meng Cai ◽

Chao Ren ◽

Edward Ng

Keyword(s):

Machine Learning ◽

Hong Kong ◽

Real Time ◽

High Density ◽

Visual Exposure

Download Full-text

Toxicogenomics screening of small molecules using high-density, nanocapillary real-time PCR

International Journal of Molecular Medicine ◽

10.3892/ijmm_00000102 ◽

1998 ◽

Author(s):

Puskás

Keyword(s):

Small Molecules ◽

Real Time ◽

Real Time Pcr ◽

High Density

Download Full-text

Aggressive, Tense or Shy? Identifying Personality Traits from Crowd Videos

Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence ◽

10.24963/ijcai.2017/17 ◽

2017 ◽

Cited By ~ 9

Author(s):

Aniket Bera ◽

Tanmay Randhavane ◽

Dinesh Manocha

Keyword(s):

Personality Traits ◽

Real Time ◽

Personality Trait ◽

User Study ◽

Movement Patterns ◽

Time Algorithm ◽

High Density ◽

Motion Model ◽

Classification Criterion ◽

Trait Theory

We present a real-time algorithm to automatically classify the behavior or personality of a pedestrian based on his or her movements in a crowd video. Our classification criterion is based on Personality Trait theory. We present a statistical scheme that dynamically learns the behavior of every pedestrian and computes its motion model. This model is combined with global crowd characteristics to compute the movement patterns and motion dynamics and use them for crowd prediction. Our learning scheme is general and we highlight its performance in identifying the personality of different pedestrians in low and high density crowd videos. We also evaluate the accuracy by comparing the results with a user study.

Download Full-text

Multi-shanks SiNAPS Active Pixel Sensor CMOS probe: 1024 simultaneously recording channels for high-density intracortical brain mapping

10.1101/749911 ◽

2019 ◽

Author(s):

Fabio Boi ◽

Nikolas Perentos ◽

Aziliz Lecomte ◽

Gerrit Schwesig ◽

Stefano Zordan ◽

...

Keyword(s):

Neural Activity ◽

Brain Activity ◽

Brain Area ◽

Large Population ◽

Brain Structures ◽

High Density ◽

Single Neurons ◽

Electrode Arrays ◽

Unit Recording ◽

Recording Channels

AbstractThe advent of implantable active dense CMOS neural probes opened a new era for electrophysiology in neuroscience. These single shank electrode arrays, and the emerging tailored analysis tools, provide for the first time to neuroscientists the neurotechnology means to spatiotemporally resolve the activity of hundreds of different single-neurons in multiple vertically aligned brain structures. However, while these unprecedented experimental capabilities to study columnar brain properties are a big leap forward in neuroscience, there is the need to spatially distribute electrodes also horizontally. Closely spacing and consistently placing in well-defined geometrical arrangement multiple isolated single-shank probes is methodologically and economically impractical. Here, we present the first high-density CMOS neural probe with multiple shanks integrating thousand’s of closely spaced and simultaneously recording microelectrodes to map neural activity across 2D lattice. Taking advantage from the high-modularity of our electrode-pixels-based SiNAPS technology, we realized a four shanks active dense probe with 256 electrode-pixels/shank and a pitch of 28 µm, for a total of 1024 simultaneously recording channels. The achieved performances allow for full-band, whole-array read-outs at 25 kHz/channel, show a measured input referred noise in the action potential band (300-7000 Hz) of 6.5 ± 2.1µVRMS, and a power consumption <6 µW/electrode-pixel. Preliminary recordings in awake behaving mice demonstrated the capability of multi-shanks SiNAPS probes to simultaneously record neural activity (both LFPs and spikes) from a brain area >6 mm2, spanning cortical, hippocampal and thalamic regions. High-density 2D array enables combining large population unit recording across distributed networks with precise intra- and interlaminar/nuclear mapping of the oscillatory dynamics. These results pave the way to a new generation of high-density and extremely compact multi-shanks CMOS-probes with tunable layouts for electrophysiological mapping of brain activity at the single-neurons resolution.

Download Full-text

A novel device for real-time measurement and manipulation of licking behavior in head-fixed mice

Journal of Neurophysiology ◽

10.1152/jn.00500.2018 ◽

2018 ◽

Vol 120 (6) ◽

pp. 2975-2987 ◽

Cited By ~ 1

Author(s):

Brice Williams ◽

Anderson Speed ◽

Bilal Haider

Keyword(s):

Real Time ◽

Neural Activity ◽

Closed Loop ◽

Neural Circuits ◽

Neural Circuit ◽

Visual Behavior ◽

Sensory Stimuli ◽

Neural Recordings ◽

Control Signals ◽

Sensory Motor

The mouse has become an influential model system for investigating the mammalian nervous system. Technologies in mice enable recording and manipulation of neural circuits during tasks where they respond to sensory stimuli by licking for liquid rewards. Precise monitoring of licking during these tasks provides an accessible metric of sensory-motor processing, particularly when combined with simultaneous neural recordings. There are several challenges in designing and implementing lick detectors during head-fixed neurophysiological experiments in mice. First, mice are small, and licking behaviors are easily perturbed or biased by large sensors. Second, neural recordings during licking are highly sensitive to electrical contact artifacts. Third, submillisecond lick detection latencies are required to generate control signals that manipulate neural activity at appropriate time scales. Here we designed, characterized, and implemented a contactless dual-port device that precisely measures directional licking in head-fixed mice performing visual behavior. We first determined the optimal characteristics of our detector through design iteration and then quantified device performance under ideal conditions. We then tested performance during head-fixed mouse behavior with simultaneous neural recordings in vivo. We finally demonstrate our device’s ability to detect directional licks and generate appropriate control signals in real time to rapidly suppress licking behavior via closed-loop inhibition of neural activity. Our dual-port detector is cost effective and easily replicable, and it should enable a wide variety of applications probing the neural circuit basis of sensory perception, motor action, and learning in normal and transgenic mouse models. NEW & NOTEWORTHY Mice readily learn tasks in which they respond to sensory cues by licking for liquid rewards; tasks that involve multiple licking responses allow study of neural circuits underlying decision making and sensory-motor integration. Here we design, characterize, and implement a novel dual-port lick detector that precisely measures directional licking in head-fixed mice performing visual behavior, enabling simultaneous neural recording and closed-loop manipulation of licking.

Download Full-text