1
Victor Babes UNIVERSITY OF MEDICINE AND
PHARMACY TIMISOARA

• DEPARTMENT OF
• MEDICAL INFORMATICS AND BIOPHYSICS
• Medical Informatics Division
• www.medinfo.umft.ro/dim
• 2007 / 2008

2
BIOLOGICAL SIGNALS (I)AQUISITION.
FILTERINGPERIODICAL SIGNALS PROCESSING

• COURSE 7

3
1. BIOLOGICAL SIGNALS AQUISITION
4

• 1.1. DEFINITION
• TIME EVOLUTION OF A BIOLOGICAL VARIABLE
• GENERAL SCHEME OF BIOSIGNAL ANALYSIS

5

• 1.2. CLASSIFICATION
• a) ON THEIR NATURE
• ELECTRICAL (ECG, EEG, EMG etc)
• NON-ELECTRICAL (pressure, concentration etc)
• b) ON EVOLUTION
• PERIODICAL (ECG)
• NON-PERIODICAL (EEG)

6

• 1.3. AQUISITION SYSTEMS
• ELECTRICAL SIGNALS electrodes
• NON-ELECTRICAL transducers
• (pH, pressure etc)

7
1.4. ANALOG - DIGITAL CONVERSION

• a) SAMPLING
• DISCRETIZATION ON OX AXIS (time)
• SAMPLING PERIOD Te (s)
• Time interval between two successive readings
• SAMPLING FREQUENCY fe (Hz)
• Number of readings in time unit (nr./sec)
• fe 1 / Te (1)

8
Example recorded signal
9
SAMPLING
10
SAMPLING
11
SAMPLING
12
SAMPLING
13
Example recorded signal
14

• a) SAMPLING THEOREM (Shannon)
• fe gt 2 . Fmax (2)
• Sampling frequency should be at least twice the
  maximal frequency of the signal
• NYQUIST FREQUENCY 2.Fmax (Hz)

15
Good sampling
16

• b) QUANTISING
• Discretization of OY axis (amplitude)
• INTERVAL BETWEEN VMAX AND VMIN IS DIVIDED INTO
  N AMPLITUDE STEPS
• THE WIDTH OF A STEP (Quantum)
• D V (V Max - V min ) / N (3)
• Relation of N with n number of bits used by ADC
  to express a reading
• N 2 n (4)

17
Quantising
18
1.5. A - D CONVERTERS

• MAXIMAL SAMPLING FREQUENCY (10 kHz - 1 MHz)
• NUMBER OF BITS (8 - 16)
• INPUT RANGE (-10/10 V, -0.1/0.1 v)
• NUMBER OF CHANNELS (MULTIPLEXING)

19
1.4. FREQUENTIAL ANALYSIS

• SIGNAL REPRESENTATION
• TEMPORAL
• Ampl f (time)
• FREQUENTIAL (spectrum)
• Ampl f (freq)

20

• b) FILTER ANALYSIS
• BAND - PASS FILTERS ( d, q, a, b )
• WAVES PROPORTIONS - mingographs
• c) FOURIER ANALYSIS
• Definition SIGNAL DECOMPOSITION INTO FREQUENCIAL
  COMPONENTS
• Domain 0 - 30 Hz
• Types of spectra
• AMPLITUDE
• POWER (proportional to A2)

21
Exemplu semnal sinusoidal de 1 Hertz si spectrul
sau
22
Semnal de 2 Hz
23

• c) SPECTRAL RESOLUTION
• DEFINITION distance between two neighbour points
  in the spectrum
• RELATION WITH EPOCH LENGTH (recorded signal
  duration, in seconds)
• D f 1 / D T (5)
• d) TIME CONSTANT
• e) TESTS FOR SIGNALS
• STATIONARITY, NORMALITY AND TREND TESTS

24
Exemple - problem

• We record an EMG signal using a 10 bit ADC,
  with a sampling frequency of 500 Hz, recording
  epochs of 2 seconds. The input signal has
  values between 0 and 100 mV. Calculate
• Sampling period (in ms)
• Maximal frequency in the spectrum
• Spectral resolution
• Number of amplitude steps
• Reading precision (quantum value, how many mV
  correspond to 1 bit)

25
2. FILTERING
26

• 2.1. DEFINITION removing or diminishing the
  perturbations
• 2.2. NOISE CLASSIFICATION (perturbations)
• a) PERIODICAL (pink noise low frequencies)
• b) NON-PERIODICAL (white noise)
• 2.3. SIGNAL / NOISE RATIO (SNR, decibels dB)
• 2.4. FILTERING MODES
• ELECTRONIC FILTER (before ADC)
• NUMERIC FILTER (after ADC)

27
2.5. TYPES OF FILTERS
28
3. PROCESSING PERIODICAL SIGNALS
ELECTROCARDIOGRAPHIC SIGNAL (ECG)
29
(No Transcript)
30
(No Transcript)
31
3.2. ECG PROCESSING - PHASES
32

• b) ARTIFACT ELIMINATION
• ZERO LINE
• SMOOTHING
• c) QRS TYPIFICATION
• d) ST - T TYPIFICATION
• ST SEGMENT AMPLITUDE
• (in coronary diseases)
• e) P - WAVE DETECTION
• VERY SMALL AMPLITUDE

33

• 3.3. ECG ANALYSIS
• RYTHM
• INTERVALS
• AMPLITUDES
• SLOPES
• 3.4. OTHER ANALYSES
• VECTOCARDIOGRAMS
• CARDIAC MAPPING
• LATE POTENTIALS, ARRHYTMIAS

34
- e n d -