Transcript Presentation

Muscle Fatigue Interface
Final Presentation
Shay Chen, Tushar Bhushan, Roman Levitas
TA: Lydia Majure
ECE 445: Senior Design
Project #30
April 1, 2013
Introduction
• Portable electromyography device for medium
sized muscles
• Provide audio and visual feedback of muscle
fatigue under stress
• Useful in preventing overtraining, programming
workouts, physical therapy
• Incorporates engineering principles from
biomedical, controls, and signal processing
disciplines of electrical engineering
Objectives
• Acquire raw EMG signal through electrodes,
differential amplifier, and filters
• Use LabVIEW to process signal
• Acquire median EMG frequency
• Provide audio and visual feedback based on
fatigue
• Storage of results via excel
Theoretical Basis for Fatigue
Analysis
• Manifests as a decrease in tension/force of
muscle
– Lack of O2, energy stores used up, lactic acid
• Decrease in conduction velocity
– Decrease in peak twitch tensions
– Increase in contraction time
• Corresponding to decrease in firing frequency
Theoretical Basis for Fatigue
Analysis (cont'd)
• As fatigue progresses there is a shift to
lower frequencies
– Fast twitch (higher frequency) motor units drop out first
– Slow twitch (lower frequency) motor units retained
System Overview
Hardware
1) Electrodes
2) EMG Circuit
3) Data Acquisition / DSP
(Arduino / National Instruments & LabView)
4) Power Supply
5) DAQ Module
Software
1) National Instruments' LabView
Top-level Design
Electrodes
• Unilect 4560M wet gel
electrodes (38x60mm)
– Transducer that captures a
muscle’s motor unit.
• Ag/AgCl sensor material
– Difference in voltage between
two points
EMG Circuit
• Two stages: Amplification & Bandpass Filtering
EMG Circuit
(AD622 Instrumentation Amplifier)
• Easy to use
• Large range of power
supply (+/- 2.6V to+/-15V)
• Excellent CMRR
• Temperature Stability
• Rg = 50.5kW/(G-1)
• Gain of 50V/V
EMG Circuit
(LM741 Operational Amplifier)
• Excellent CMRR
• Large range of power
supply (+/- 2.6V to+/-15V)
• Active Bandpass filter
built around it
• Gain of 30V/V
Data Acquisition / DSP
(National Instruments /
LabView)
• NI-DAQ (National Instruments Data
Acquisition)
• LabView
– Graphical programming language
– Used as DSP instead of the Arduino
– Performs Fast Fourier Transform (FFT),
Finite Time Integration & Median
Frequency calculation
Frequency Domain Analysis

Transformation from the time domain to the frequency
domain
- Fast Fourier Transformation (FFT)

Removes the time between successive action potentials so
that they appear as periodic functions of time
Prefatigue
Fatigue
Median Power Frequency
Calculation




Sample data at multiples of x2
(1024 Hz)
Rectify & filter (BP or LP) raw
signal
Apply FFT
Compute Median (or mean power)
frequency
LabVIEW Code
Power Supply
• 9V Alkaline batteries
• Powers EMG
Interface
Video
Video (Link)
In case the video does not work, here's the
YouTube link:
https://www.youtube.com/watch?v=8clGWYrYyW
E
EMG Testing
• 4 Subjects
• One electrode placed on bulge of bicep,
one placed two inches lower
• 25lbs held until failure
• Four trials with three minutes rest in
between
Simulations
Simulations (cont'd)
Results
Rate of Fatigue (Slope of each trial)
Test Subject
Trial 1
(Hz/s)
Trial 2
(Hz/s)
Trial 3
(Hz/s)
Trial 4
(Hz/s)
Avg.
(Hz/s)
Slope over 4
Trials
(Hz/s^2)
Dave
-0.667
-0.450
-0.613
-0.400
-0.532
.063
Shay
-0.143
-0.133
-0.179
-0.174
-0.175
-.013
Tushar
-0.350
-0.429
-0.500
-0.619
-0.475
-.142
Roman
-0.333
-0.550
-0.800
-1.00
-0.671
-.225
Results (cont'd)



Rate of fatigue remained
in same relative range
Change in rate of fatigue
increases with
subsequent trials as
expect for 3/4 subjects
Repeatable
PCB (Eagle Cad)
-Printed Circuit Board for EMG Module
Arduino Uno
Arduino Uno:
• Atmel ATmega328
microcontroller
• 14 Digital I/O Pins
• 6 Analog Pins
• Voltage required: 5V
Microcontroller
• Used as DSP (Digital
Signal Processor)
• Takes output of the preamp stage as input
• Performs a Fourier
Transform on the timedomain signal using the
FFT arduino library
PIN Configuration
• The input is taken in by
the analog pin
• The outputs are the
digital outs to the sound
buzzer and LED’s
• An additional SD card
shield is used for the data
logger
Implementation Issues
Problems:
-Not enough processing power to perform the FFT needed
-Arduino Has 8-bit DAC, that is, it can run 2^8 (256)
instructions to “process” each sample.
-Given that the Nyquist sampling rate:
fs > 2(Bandwidth)
-Our bandwidth is about 500Hz, we require a sampling rate
greater than 1000Hz, or rather, 1000 instructions per second
Implementation Issues
Solutions:
-Use a dedicated DSP such as the ADUC7019 from Analog
Devices running on an ARM processor
-The device has a 16-bit DAC which is able to perform
2^16 (65536) instructions to “process” each sample.
-Additionally, has a assembly level optimized FFT library
that is able to process the samples with even fewer
instructions
Improvements and Future
Work




Interface with Arduino to create a portable
version
Wireless
PCB Version of the prototype board
Case + armband that is wearable
Conclusion




Designed EMG system to take real time
samples, amplify, filter, and calculate
median frequency
Used LabVIEW instead of Arduino as
originally planned
Able to detect change in median
frequency corresponding to muscle fatigue
Able to run repeatable experiments
Acknowledgement
•
•
•
•
•
Professor Carney
Professor Gentry
Lydia Majure
ECE 445 staff
Friends of the Part Shop
Thank you!
Questions?