Biomechanical Instrumentation Considerations in Data Acquisition ÉCOLE DES SCIENCES DE L ACTIVITÉ PHYSIQUE SCHOOL OF HUMAN KINETICS

Transcription

1 Biomechanical Instrumentation Considerations in Data Acquisition

2 Data Acquisition in Biomechanics Why??? Describe and Understand a Phenomena Test a Theory Evaluate a condition/situation Data Acquisition provides information that is used in making decisions.

3 The Goal!!! Good Decision Accuracy in Data Acquisition

4 Levels of Data Acquisition Visual Observation and Human Interpretation Limited Information Processing Capacity Subjectivity in interpretation Instrumented Observation and Human Interpretation Un-limited information processing capacity Decreased subjectivity of interpretation Instrumented Observation and Interpretation Un-limited information processing capacity and Objectivity of interpretation * Lack of spontaneity and creativity

5 Factors that Maximize Accuracy in Data Acquisition Selection of the correct measurement technique Use of established techniques Attention to appropriate sensitivity levels Calibration Standardization of protocols Adequate preparation (ie training, pilot testing, etc.)

6 Factors that Maximize Accuracy in Data Acquisition Attention to the Details Good Decisions

7 Biomechanical Data What s it like? Continuous Wide range of Amplitudes Variability of Duration Wide range of Frequencies

8 Data are Continuous ROM Degrees EMG Gait Cycle EMG (uv)

9 Wide Range of Amplitudes Ground Reaction Forces Hundreds of Newton EMG Millionths of a volt

10 Wide Range of Amplitudes Measurement Range Frequency, Hz Method Blood flow 1 to 300 ml/s 0 to 20 Electromagnetic or ultrasonic Blood pressure 0 to 400 mmhg 0 to 50 Cuff or strain gage Cardiac output 4 to 25 L/min 0 to 20 Fick, dye dilution Electrocardiography 0.5 to 4 mv 0.05 to 150 Skin electrodes Electroencephalography 5 to 300 µ V 0.5 to 150 Scalp electrodes Electromyography 0.1 to 5 mv 0 to Needle electrodes Electroretinography 0 to 900 µ V 0 to 50 Contact lens electrodes ph 3 to 13 ph units 0 to 1 ph electrode pco 2 40 to 100 mmhg 0 to 2 pco 2 electrode po 2 30 to 100 mmhg 0 to 2 po 2 electrode Pneumotachography 0 to 600 L/min 0 to 40 Pneumotachometer Respiratory rate 2 to 50 breaths/min 0.1 to 10 Impedance Temperature 32 to 40 C 0 to 0.1 Thermistor

11 Wide Variability of Duration Continuous Motion Studies - hours Reaction Time Studies - msec

12 Wide Range of Frequencies ROM in Walking 2 to 4 Hz Foot Impact Shock 200 to 300 Hz EMG > 2000 Hz

13 How Do We Acquire Biomechanical Data?? Video/Cine Force Plates Electromyography Pressure Plates Accelerometers Force Transducers Electrogoniometers Etc.

14 How do we Record the Data?? Old technology (yuk) Chart Recorders Oscilloscopes Tape Recorders New Technology Computers Data loggers

15 The Problem!!! Instruments produce continuous data (Analog Data) Computers like discrete data (Digital Data)

16 The Problems!!! Amplitude 5 mv -5 mv Amplitude 1 V -1 V Time Time Dynamic Range (a) An input signal which exceeds the dynamic range. (b) The resulting amplified signal is saturated at ±1 V.

17 The problems!!! Amplitude 5 mv Time Dynamic Range -5 mv Amplitude Time Dc offset (a) An input signal without dc offset. (b) An input signal with dc offset.

18 The Solution The Analog to Digital (A/D) Converter Changes the in-coming (analog) signal to (digital) information that can be processed by the computer

19 Principles of A/D Conversion An analog signal (typically a voltage) is measured at periodic intervals. At each interval the voltage is given a numerical value that represents the amplitude of the voltage

20 Principles of A/D Conversion The Analog values that represent the signal are then stored, as an array of numbers, for processing

21 Data Sampling and Data Treatment

22 Data Sampling and Data Treatment Issues Transferring Analog Signals to a Digital Computer Time and Frequency Domain Analysis Determining Optimal Sampling Rates Prevention and Treatment of Noisy Data Data Normalization

23 The Analog to Digital (A/D) Converter Analog Signals GRF ROM EMG

24 The Analog to Digital (A/D) Converter Changes the in-coming analog signal to digits (numerical information) that can be processed by the computer

25 Principles of A/D Conversion An analog signal (typically a voltage) is measured at periodic intervals. At each interval the voltage is given a numerical value that represents the amplitude of the voltage

26 Features of the A/D Converter Channels - 4, 8 16, 32, 64 Gain - 2, 4 8, 10 (typical) Input Range - variable (+-10 Volts) Sampling Rate Low 1000 Hz to High 500 khz Resolution 8 Bit 256 units 12 Bit 4096 units 16 Bit units D/A Capacity

27 Time and Frequency Domain Analysis Time Domain Frequency Domain Mv Mv Time (seconds) Frequency (hz)

28 Time and Frequency Domain Analysis Time Domain Represents change in signal Amplitude relative to change in Time Frequency Domain Represents change in signal Amplitude relative to the Rate of Change in Amplitude Time Domain Fourier Transform (FFT) Frequency Domain

29 Frequency Domain Analysis Examples

30 Determining Optimal Sampling Rates Sampling Rate: The rate at which periodic measurements of a signal are made. Units are samples per second or Hz Examples An EMG signal being sampled at 1000 Hz A video picture being sampled at 60 Hz How Fast Do We Need to Sample the Data?

31 Considerations in Selecting a Sampling Rate Frequency Characteristics of the Signal the rate at which the amplitude of the signal changes Examples: Rapidly Changing Signals Slowly Changing Signals -

32 Considerations in Selecting a Sampling Rate Frequency Characteristics of the Signal - the Nyquist Sampling Theory Speed of Signal Processing and Data Analysis Depends on: What s needed Computer Processing Speed Amount of Data Requisite Processing

33 Considerations in Selecting a Sampling Rate Frequency Characteristics of the Signal - the Nyquist Sampling Theory Speed of Signal Processing and Data Analysis Storage Capacity of the System Number of Channels Simultaneously Sampled Capacity (speed and channels) of A/D system

34 Typical Sampling Rates for Biomechanical Data Force Platform - 10 Hz (balance) to 1000 Hz (running, jumping, etc.) EMG Hz to 2000 Hz Video - 15 fps to 500 fps

35 Determining Optimal Sampling Rates Theoretical Determine the frequency characteristics of the signal to be sampled Apply the Nyquist Theory ( i.e. At least 2 x the highest frequency in the signal) Practical Copy what someone else has done!!!

36 Determining Optimal Sampling Rates What Happens if we Sample too slow Aliasing Error (introduces frequencies into the data that aren t actually there Sample Too Fast Generates excess data

37 Sampling Rate Examples

38 Prevention and Treatment of Noisy Data A BIG Problem!!!

39 Noisy Data Noisy EMG Signal Not Noisy (clean)

40 Minimizing the Effect of Noisy Data Control sources of noise before contamination eliminate sources of noise Vibration Radiant electrical energy Movement artifact (cable movement) Filter data after contamination with appropriate hardware and/or software filters

41 Filtering Data The Goal Extracting the Noise without Changing the Signal

42 Digitial Filtering Based on the Frequency characteristics of the data A mathematical process that selectively eliminates that part of the data that is caused by noise Based on the assumption that the noise occurs at frequencies that are different from those of the actual signal

43 Digital Filtering Raw Signal (signal + noise) Filtered Signal

44 Digital Filtering Examples

45 Noise Reduction Other Techniques Smoothing Moving Window Curve Fitting Cubic Spline Root Mean Square *All of the above are effective but less specific *May also be used to simplify complex waveforms to enhance analysis

46 Noise Reduction - other Examples

47 Data Normalization The Goal To convert the data from one base unit to an alternative base unit 1. To enhance ease of interpretation 2. To establish a common base so that averaging across subjects/conditions is possible Examples The mean level of muscle activity in the biceps during the arm curl was 80 mv. The mean level of muscle activity in the biceps during the arm curl was 98% of a maximum voluntary contraction

48 Data Normalization The Goal To convert the data from one base unit to an alternative base unit 1. To enhance ease of interpretation 2. To establish a common base so that averaging across subjects/conditions is possible Examples The force on impact with the ground was equal to 1100 Newtons The force on impact with the ground was equal to 1.5 bodyweights

49 Data Normalization Other types of data normalization Normalizing time by the duration of a cycle Ex. Expressing gait events relative to a gait cycle ie. 20% of the gait cycle Normalizing O 2 consumption by expressing it as a function of body mass and/or time Ex. Ml/Kg/Min

50 THE END