ORT21BMI

1
ORT21BMI

• Week 5
• Principles of biomedical instrumentation

2
Clinical context

• Youre still working on doing the ERG on the
  patient with night-blindness.
• Youve made sure that the unit was safety
  inspected
• Now youre ready to set up
• It says electrode impedance limit exceeded
• Theres a low pass filter cutoff to set
• Theres a high pass filter cutoff to set
• You say stuff it and start anyway, getting a
  signal
• It looks odd
• Is it disease? Is it a set-up mistake?
• What to do???

3
Learning goals

• Being comfortable with the following concepts and
  their clinical relevance
• Input impedance
• A. What is impedance?
• B. Why should it matter to clinicians?
• AC vs DC coupled circuits
• A. Why isnt everything one or the other?
• B. How to make the choice and understand the
  implications
• Frequency-related issues
• A. Frequency of what? Decomposing things into
  their basic frequencies
• B. What does frequency response of an
  instrument mean?
• C. What is filtering and why should you care
  about it?

4
A reminder...
Patients eyes brain
Instrument
You
5
What well cover

• Techie stuff
• Some technical vocabulary is necessary if
  youre going to use clinical instrumentation,
  particularly regarding ERGs VERs, safely and
  accurately
• But dont built-in computers do all that now?

6
Why you should care

• Just because all the knobs and switches have been
  replaced by software settings, this doesnt mean
  that the instrument will set itself up for each
  individual test
• Different tests require different configurations
• Asking a piece of electronics to do your
  professional thinking for you can lead to results
  which may look lovely on the display but be wrong
  and clinically incorrect

7
(No Transcript)
8
Starting at the beginning--the input

• A common specification for any electronic
  instrument is its input impedance
• Higher is better
• But whats impedance?
• Does it help to say its the alternating current
  equivalent of resistance? That it varies with
  frequency?
• I thought so...

9
OHMS LAW

• V IR
• Where
• V Voltage
• I Current
• R Resisance

10
The capacitor and its impedance

• Because properly defining impedance requires the
  use of imaginary numbers, well dance around it
  in an approximate way, trying to use physical
  explanations
• The most clinically important circuit component
  with impedance is the capacitor

11
Meet the capacitor

• Lets try analogies again
• Recall that resistance was compared to an
  obstruction to the flow of water
• Capacitance is rather like a bucket that the
  fluid can collect in, or a room that the people
  can fill
• A bigger bucket or larger room can hold more
• Capacitors do something similar, but for electric
  charge

12
Impedance and the capacitor

• Before considering alternating current, well
  start with the simplest situation, where you have
  a DC source (a battery) and it is suddenly
  connected to a resistor and capacitor, connected
  in series
• The capacitor is simply 2 parallel conductive
  plates, close together but not touching

• What are the conditions before and after the
  switch closes?

13
Applying a DC voltage to a capacitor

• When the circuit is open, nothing happens
• Close the switch and current flows, charging up
  the plates of the capacitor
• The speed of this is determined by the resistance
  in series with the capacitor
• When the charge on the plates equals that of the
  voltage source, theres no voltage difference and
  hence no current
• The capacitor is now fully charged
• So the DC voltage led to only a transient current

14
What if the voltage isnt DC?

• In AC, the voltage swings between positive and
  negative periodically, as described by its
  frequency (eg, 50 Hz)
• What happens when you apply an AC current to a
  capacitor?
• Start considering it at the point where V 0
• How does the size (capacitance) of the capacitor
  affect it?

15
How is this expressed?

• The symbol for impedance is Z, also measured in
  ohms (?)
• In physics books, its expressed with both
  resistive and reactive components as a complex
  number
• Well just consider the magnitude of impedance,
  and only for capacitors

16
So?

• What does Z do as frequency goes up?
• Whats the impedance of a capacitor for DC?
• How does impedance vary as capacitance goes up?

17
What if you have Rs and Cs?

• Can combine impedances and resistances in the
  same way that you do resistances alone.
• Wont be doing much analytically, but well see
  situations with resistors and capacitors in
  parallel - need to understand how different
  currents could flow
• Need to?
• Currents travelling from the eye or brain through
  tissue into electrodes traverse structures whose
  electrical properties can be represented by
  combinations of Rs and Cs

18
Highs and lows of impedance

• If you are doing a clinical recording, there are
  places you want the impedance to be high and
  places you want it to be low
• Understanding these will help you generate
  accurate, clinically useful information
• Lets start by considering the impedance of
• 1) the electrodes
• 2) the recording device
• Which should be high and which should be low?

19
Heres the set-up

• Signals include the ERG and biological noise
  sources
• Each element has an impedance Z
• It also has a voltage drop V across it

Tissue electrode impedances are generally in
1000s of ohms instrument impedance is usually
1000,000 ohms
20
Points to consider

• First, consider the source
• A human retina isnt a great power source
• It cant supply much current
• You want to measure Vsignal as accurately as
  possible with your instrument
• Vmeasured Vsignal as close as possible

Velectrode
Vtissue
21
What needs to be done?

• What voltages exist in the circuit?
• Vsignal,, Vtissue,, Velectrode, Vmeasured
• It should be clear that you want Vtissue and
  Velectrode to be as low as possible if Vmeasured
  is to approximate Vsignal
• How?
• Since V IZ, to make the Vs low, make the Zs low

22
What can you control?

• You cant get more current out of the retina
• But some things are directly in your control
• Vtissue depends on Ztissue, and that you can
  sometimes influence
• If using skin electrodes, you can clean and
  lightly abrade the skin on the scalp, have
  patients wash their hair. Not much you can do
  for the cornea.
• Velectrode also depends on Zelectrode
• Use the best electrode for the task
• Be sure its clean
• If its a skin or scalp electrode, use the right
  electrode paste

23
But wont this raise the current, not reduce the
voltage?

• Not really--compare the input impedance to the
  others
• What determines the current flow is this
  impedance
• So, what should it be, high or low?
• What would Vinstrument input be if Zinstrument
  input approached infinity? Why?

24
Doing what you can...

• If youre recording an EOG or VEP, you have
  something like this
• What can you have an effect on?

25
Instrument input types

• There are two basic types of inputs that almost
  any electronic device can have--those that allow
  DC through and those that dont
• Those that do are called DC-coupled
• Those that dont are called AC-coupled
• Since AC can get through if DC, can why not
  always be sure that any signal can be processed?

26
Not every signal has a constant component

• Recall the electroretinogram
• Does it have any extended, fixed regions?
• How about an electrooculogram?
• How stable is it?
• Notice the difference?
• (note time scales)

27
But why not accommodate both?

• Think back to the discussion of electrode-tissue
  interactions
• Recall the problems of drift and mechanical
  disturbance
• Consider a large (0.5V) offset added to a tiny
  (0.5mV) ERG signal
• What would happen to the offset if you tried to
  increase the ERG components amplitude to 0.5 V?
• Do you sense an electronics problem here?

28
A problem

• If your desired signal (ERG, VEP, etc) has no
  clinically significant DC component, why process
  an artifactual DC component?
• You can avoid this by blocking DC at the input
• What electrical component weve recently met will
  keep DC from passing?
• Buy why not always keep out DC?
• Recall that EOG of a saccade?

29
How to choose?

• Heres where your knowledge is key
• The instrument doesnt know what youre doing
• It may have default settings
• They may be wrong
• Its up to you to know what the biopotential
  youre recording consists of and set up your
  apparatus appropriately

30
For example, consider the EOG

• When used to record saccades, it must reproduce
  the period of fixation after the movement
• Which sort of coupling does this the best?
• Could the wrong choice be clinically misleading?

31
What about other frequencies?

• Remember a concept that you encountered when you
  learned about contrast sensitivity
• Recall the VisTech chart in Room 317?
• Remember how the little patches of sine wave
  gratings got finer and finer?
• The contrast sensitivity function is important
  because it considers the fact that all visual
  images are composed of a wide range of spatial
  frequencies

32
The frequencies of things

• Fine detail is composed of high spatial
  frequencies large structures consist of low
  spatial frequencies
• Every visual image can be decomposed into its
  individual spatial frequency components

33
Filtering
High frequencies removed
Low frequencies removed
Whats missing in each example? Whats still
present?
34
Sharp corners mean high frequencies

• Consider the difference between square and sine
  waves (as in gratings or as graphs)
• Which shows abrupt changes?
• Which is smooth?
• How do you get from one to the other?
• It can be shown that a square wave of frequency
  f consists of a fundamental sine wave of
  frequency f, plus scaled sine waves that are odd
  harmonics of f that is, 3f, 5f, 7f, 9f
• To illustrate

35
The birth of a square wave
The more odd harmonics you add, the squarer the
result
36
SO?
37
All signals can be broken down this way

• Signals varying in time, not just space, are made
  up of different frequencies
• Think of speech
• deep vs. high-pitched voices
• Think of colour
• red vs. violet
• Think of electroretinograms
• Recall the relatively brief a-wave, the long,
  slow b-wave and the much more rapid oscillatory
  potentials

38
What does filtering do to clinical data?

• Below is the actual recording of a visual evoked
  potential the raw data are shown by the fuzzy
  dots, the low-pass filtered version by the solid
  line.
• Whats the difference?
• What does low-pass filtering mean?

39
Why removing parts of signals may be good to do

• Anything that changes the frequency content of a
  signal is a filter
• Think tinted glasses or the tone controls on your
  sound system
• Why would you want to do this clinically?
• Recall all the types of unwanted noise that can
  contaminate a physiological signal
• The more of them you can remove, the better
• Knowing what you can and cant get rid of is
  clinically important

40
Why would a VEP be noisy?

• Where does the signal come from?
• Whats between the origin of the signal and the
  recording device that could contribute noise?
• How separable is that noise from the much-desired
  signal?
• Heres where some knowledge is helpful

41
Separating the signal from noise

• You need to know two key things
• 1) the frequency content of the signal
• 2) the frequency content of the noise

42
Consider two extremes

• Recording individual muscle fibre activity (that
  is, individual spikes) with a needle electrode
• Recording the change over 20 min in the
  electrooculogram after a change in illumination
• Which of these contains mostly high frequency
  activity and which low frequency?

43
AC vs DC coupling as a special case of filtering

• When you AC-couple an instruments input, what
  are you keeping out?
• Allowing high frequencies through but not low
  frequencies is called high-pass filtering
• What, then, does low pass filtering allow through
  and what does it keep out?
• If its good, isnt more always better?

44
Lets look at some simple tests

• Here and for the next few slides, well look at
  what happens when we filter 10 msec and 1 sec
  pulses with various cutoff frequencies
• The same type of filter was used for each
• From the left, these have 50, 10 and 5 Hz cutoffs

What happens to the signals?
45
And high-pass filtering?

• Heres an example filtered with 0.1 (left) and 10
  Hz (right) cutoffs
• What happens to the original signal?

46
An example from real data

• Heres a saccade, recorded in our lab, using an
  infrared reflectance eye tracker
• Thus, theres no muscle noise, but that doesnt
  make it noise-free
• To get of what remains, why not low-pass filter
  the daylights out of it?
• What happens?

47
The results...

• Notice the good things--the noise riding on the
  raw signal is gone in both the filtered traces
• But what about the timing?
• What about the little peak on the saccade at 33.6
  sec?
• Is the heavily filtered trace any more different
  from the other two?

48
Concerns

• What if timing of a major peak is diagnostically
  crucial?
• What if little features are vital (think
  oscillatory potentials in the ERG)?
• See what too much filtering can do to you?
• And with modern technology its so easy to do!
• Youre a professional--dont let a computer do
  your thinking for you.
• Just because a device has default settings
  doesnt make them right for your investigation

49
So what do you need to know?

• To reiterate, you need to know first of all what
  frequencies compose your signal
• Only then can you decide what you can safely
  filter out
• Then you have to decide how much noise you can
  remove without harming the signal

50
Does it matter what sort of filter you use?

• What do I mean, what sort?
• In the old days, filters were all electronic
• Today, much filtering is done by computer
• Are they the same or different?
• Yesbut thats for later.