Underatanding digital data acquisition in medical diagnostics by rbt28073

VIEWS: 4 PAGES: 16

									    Understanding digital
       data acquisition
    in medical diagnostics
       advantages, risks,
      technical approaches




           Lutz Issler
         Martin Baumann
1
                 What‘s it about?


    Same topics as in our introductory seminar:
    Part 1: Basics and some theory.
    Part 2: Using filters.
    Part 3: Real data are real problems.
    Finally...




2
               Part 1:
      Basics and some theory.




    Bit by bit - some terms.
    Analogue vs. digital.




3
           Bit by bit - some terms.


    What are quantisation steps?
       ex.: 12 bit resolution  212 = 4096 steps
       quantisation error = 1/(212) = 0.024% of
       max-value
    What is a sampling-rate?
       1 kHz acquisition rate  1000 samples per
       second
       every 1 ms a new value
       faster events may remain invisible


4
          Analogue vs. digital: Example.



                               1.322
                               1.412
                               1.911
                               2.332
                               2.627
                               2.872
                               2.972
                               3.099
                               3.499
                               3.881
                               4.007
                               3.502
                               2.391
                               1.000
                               0.480
                               0.479
                               0.653
                               1.304
                               1.600
                               2.099
                               2.126
                               1.982
                               1.578
                               0.910
                               0.344
                               0.107
                               0.019
                               0.124
                               0.299
                               0.505 mV
                                Digital
                               Multimeter




    Oscilloscope
                                            Start!
5
         Analogue vs. digital: Facts.


    Transient signals can be followed easier:
        slow signal changes
        rapid signal changes
    “Exact” multimeter values are not that exact:
        pretend to be accurate
        “jumping” digits
    Possibility of further processing of data records




6
                 Part 2:
              Using filters.


    Fourier, not furious.
    Filter - When and why?
    Filter - Universal solution?




7
    Fourier, not furious.
             match?
             match!   Fourier-Algorithm: build
                           any signal from an
                       addition of sine-waves.




              add
8
           Filter - When and why?


    Real signal = ideal signal plus noise.
    Task: remove noise and keep ideal signal.
    There are many different kinds of noise, so
        there is no all-purpose filter algorithm!




9
          Filter - Universal solution?


     No, non, nein, non est.
         (almost) every filter will change
         signal characteristics
         one must not trust the filter result in
         every case
     Yes, oui, ja, sic est.
          better to filter and know the possible errors
          than not to filter and keep guessing
          one must trust own knowledge and
          experience

10
                Part 3:
     Real data are real problems.




     True lies.
     Storage and restoration.




11
                     True lies.


     Manipulation is possible:
         inserting / deleting data
         points
         selecting display range
     Display of data is possible
     in various ways:
          can be used to support or

         to hide information
         may guide the
         interpretation
12
           Storage and restoration.


     Acquired data are just - data:
         store, compress, calculate, combine, ...
         attention: 1hr 10kHz 10dig  >340MByte
     Store where?
         paper, (hard)disk, (RW)CD, tape, ...
         how long do media / technologies survive?
     Store what?
         raw data, analysed data



13
                       Finally...




     Why did I tell you all this?




14
                        Why?


     Students should learn good skills in ...
         ... data acquisition,
         ... data calculation,
         ... data interpretation.
     Students must learn ...
         ... how to detect and solve
         data acquisition errors,
         ... a reasonable distrust in any data,
         ... an effective presentation of data.


15
           Thank you for listening!


     Contact:
         Lutz Issler, Martin Baumann
         Institute of Physiology
         RWTH, 52057 Aachen
         phone:      +49/241/80-88823
         fax:        +49/241/8888-434
         (issler|mbaumann)@physiology.rwth-
         aachen.de
     I look forward to answering your questions.

16

								
To top