Download presentation

Presentation is loading. Please wait.

Published byMorgan Fincham Modified over 2 years ago

1
Movable Electrodes & Feature- Based Decoding S. Cao, Z. Nenadic, D. Meeker, R. Andersen E. Branchaud, J. Cham, J. Burdick Engineering & Applied Science Biology Get the max yield of high quality signals Extract max info from (non-optimal?) neurons Electrical Signal Feature Based Spike Decoding Feature Based LFP Decoding Reach State Reach State Prosthetic Control Signal Goals: (hardware) (software)

2
Limitations of Neuro-Probes for Chronic Recording Key Challenge: record high quality signals from many neurons for months/years Fixed positioning of implant Non-optimal (or wrong!) receptive fields. Non-optimal cell type Electrode not near cell body: Array moves in brain matrix Inflammation, Gliosis, encapsulation, …

3
Movable electrodes could: track movement due to migration improve SNR overcome implant errors find “better” neurons break through encapsulation Make the electrodes movable! (autonomously controlled) Limitations of Neuro-Probes for Chronic Recording

4
Current Research Program Outline Theory – develop probe control algorithms using computational model Model extra-cellular neuron potentials Control algorithm development guided by computational model Hardware– meso-scale test-beds Validate concept, evaluate algorithms Determine spec.s for MEMS devices Test biomechanics of movable electrodes Experiments– verify theory

5
Single Cell Extracellular Potential Simulation (adapted from Holt & Koch ’98) 3720 compartment NEURON pyramidal cell model ( adapted from Mainen & Sejnowski ‘96 ) Synaptic inputs scattered uniformly throughout dendrites. Laplace equation: Boundary condition: Since solution nearly impossible, use line source approximation (Holt & Koch ‘99) soma

6
Spatio-temporal variations of extracellular potential

7
Virtual experiment Add neural noise

8
Keep electrode in this region! Quality MetricIsolation curve How to find the maximum point of the average isolation curve when all we have are noisy observations? Peak-to-Peak Amplitude

9
Solution offered by variant of Stochastic optimization. Basis function approach Iterative Algorithm

10
Experimental Setup Microdrive in the brain Filters / Preamps X Computer with: Data Acquisition Electrode Control algorithm Move Command

11
Experimental Results (monkey Parietal Reach Region) Electrode Position Peak-to-Peak Amplitude Averaged Waveform Cell Isolation Curve Electrode Path

12
Algorithmic State Machine Initial State Spikes Detected? Move Fixed TF Spikes Detected No Spikes Detected

13
Algorithmic State Machine Initial State Spikes Detected? Move Fixed TF Spikes Detected No Spikes Detected

14
Algorithmic State Machine No Spikes Detected Spikes Detected? Move Fixed TF Spikes Detected

15
Algorithmic State Machine No Spikes Detected Spikes Detected? Move Fixed TF Spikes Detected

16
Algorithmic State Machine No Spikes Detected Spikes Detected? Move Fixed TF Spikes Detected

17
Algorithmic State Machine No Spikes Detected Spikes Detected? Move Fixed TF Spikes Detected

18
Algorithmic State Machine Spikes Detected Isolation Curve to maximize? Move Fixed Move Gradien t TF Maximize Isolation Curve

19
Algorithmic State Machine Spikes Detected Isolation Curve to maximize? Move Fixed Move Gradien t TF Maximize Isolation Curve

20
Algorithmic State Machine Spikes Detected Isolation Curve to maximize? Move Fixed Move Gradien t TF Maximize Isolation Curve

21
Algorithmic State Machine Spikes Detected Isolation Curve to maximize? Move Fixed Move Gradien t TF Maximize Isolation Curve

22
Algorithmic State Machine Maximize Isolation Curve Is Cell Isolated? Move Gradien t Do Not Move TF Maintain Isolation

23
Algorithmic State Machine Maximize Isolation Curve Is Cell Isolated? Move Gradien t Do Not Move TF Maintain Isolation

24
Algorithmic State Machine Maintain Isolation Is Cell Isolated? Move small Fixed Do Not Move TF Regain Isolation

25
Algorithmic State Machine Regain Isolation Is Cell Isolated? Move small fixed Do Not Move TF Maintain Isolation Re-isolate when signal falls below threshold

26
Movable Multi-Electrode Testbed sub-micron steps, 1cm range fits in standard chamber many adjustments can insert micro-capillary Test Multi-electrode issues Test electrode/fluid combos gather data for MEMS spec.s “Nanomotors” Chamber Acrylic Skull Dura Brain Tissue

27
Surgical Implantation Special tooling installs and positions the electrode and fluid delivery modules 5mm Motorized tooling fixture brain dura skull Implant and movable probes Stacking chamber and buss connector Implant MEMS movable probes and fluid delivery chips Fluid delivery chip: stores & delivers anti-inflammatory cytokine and neurotrophic factors Movable electrode chip: Low-heat and low-energy hydrolysis bellow actuators move electrodes Fluid injectors Peristaltic electro- lysis pumps Fluid reservoir Movable probe Low-heat hydrolysis actuators Control electronics

28
MEMS Electrolysis Actuator Concept (with Y.C. Tai) Large Force Generation Low Temperature Low Power Lockable Electrode Bellows Z-Movement Actuator Electrolysis Electrodes

29
Feature Based Bayesian Decoding Characterize receptive Fields Predict movement plan x=argmax[P(x|v)] In real time, record cell activities What features to use?

30
Decoding the Planned Reach Direction Bayesian Classifier Firing RateFeature Extraction x-Reach Direction v-Feature

31
Wavelet Packet Overview Wavelet Packet TreeHaar Wavelet Packet up to Level 14 Number of spiking in a window (firing rate) Local change of firing rate (slope in PSTH) Local oscillation in spiking train (bursting) LH LH

32
Feature Selection Goal: select the most informative wavelet bases (features) X is reach class p(v|X) is conditional probability of feature v given class X Solution: choose cost function to quantify the decodability of each feature ) Mutual Information Spike train Basis Functions

33
Wavelet Packet Tree Pruning Prune the wavelet packet tree in searching for the most informative features. Features with large mutual information Features that are orthogonal to each other Feature Template t

34
Simple Sanity Check Poisson Spike Trains with repeatable spikes at specific times Identified Features Decode Performance MFR = 25% Feature: 91%

35
Single neuron decoding comparison (PRR Neuron, left-right reach task) Mean Firing RateOptimal Feature Coef value Probability Decoding Performance 52.5% 68.0% Feature Probability Optimal Feature

36
Multiple Neuron Performance Comparison 8-direction decoding using up to PRR 41 neurons (from single electrode acute recordings) 4 neurons with no obvious MFR tuningAll 41 available neurons -red MI -blue MFR

37
Step 1. Estimate the firing rate function from the spike train ensemble Wavelet thresholding method [Donoho 1994] Firing rate function estimation using wavelet thresholding Projecting Noising Estimation Thresholding Denoising

38
Step 2: Computing the Theoretical Wavelet Packet Coefficient Distribution If the spike train process is a homogeneous Poisson … even odd Probability Coefficient Value If the spike train process is an inhomogeneous Poisson … Computational method that computes the probabilities exists

39
Example Distribution of Inhomogeneous Poisson Process (t)=10step(t)+10step(t-256) time frequency NOTE: The error on the probability P*(v) caused by the estimation error of the rate function decays exponentially with the number of spike trains in the ensemble

40
Step 3. Estimate the empirical distribution of the wavelet packet coefficients Each wavelet packet coefficient is integer valued Histogram rule estimation Coefficient Value Probability

41
Step 4: Goodness-of-fit Test between the Theoretical and Empirical Distributions Use 2 test to assess the difference between the two distributions DOF is the cardinality of the coefficient v jk If p-value > 0.95, the coefficient’s distribution deviates significantly from its Poisson counterpart If p-value < 0.95, both distributions conform

42
Results Result 1: Cyclic Poisson Process Δt Scale jj t = 32 t = 64 Δt

43
Results (II) Result 2: Brandman-Nelson Non-renewal Model [Brandman 2002] Scale jj b = 0.5 b = 0.25 As slope b decreases, the scale of renewal increases; equivalently, the process becomes more Poisson like. Spike Train Generating Process

44
Poisson Scale-Gram Characterize Poisson-ness at different scales (i.e., is rate coding appropriate?) Short time-scale non- Poisson-ness Longer time-scale non- Poisson-ness Relatively Poisson Populations of PRR neurons during virtual reach experiments (D. Meeker) Time scale Coefficient index

45
First Experimental Results (monkey Parietal Reach Region) Cell Isolation Curve Electrode Path

Similar presentations

OK

Alan L. Yuille. UCLA. Dept. Statistics and Psychology. www.stat.ucla/~yuille Neural Prosthetic: Mind Reading. STATS 19 SEM 2. 263057202. Talk 6. Neural.

Alan L. Yuille. UCLA. Dept. Statistics and Psychology. www.stat.ucla/~yuille Neural Prosthetic: Mind Reading. STATS 19 SEM 2. 263057202. Talk 6. Neural.

© 2017 SlidePlayer.com Inc.

All rights reserved.

Ads by Google

Ppt on amplitude shift keying formula Ppt on aluminium form work Ppt on statistics and probability help Download free ppt on environment Ppt on fire extinguisher types pictures Ppt on marketing strategy of coca cola in india Ppt on leadership and change management Class-9th physics ppt on motion Ppt on tata trucks south Ppt on natural numbers example