More tools in WFDB

WFDB is a set of functions that include a lot of aspects in signal treatment, specially in the field of Cardiology. As an example, we are going to show how the package can be used to detect QRS complex and how the result can be compared with the original annotations of the program. It is an example included by George B. Moody and we are going to reproduce the commands from the Scilab prompt. As in the previous chapters, we assume that we are in `/home/j/physionet', that WFDB and Scilab are installed and that the files have been copied in the directory.

We are trying to use the QRS detector (command `sqrs'). We wil apply it to the file `100.dat'. We begin by exploring the contents of the directory:

-->unix_g('ls')
 ans  =
 
!100.atr   !
!          !
!100.dat   !
!          !
!100.hea   !
!          !
!100s.atr  !
!          !
!100s.dat  !
!          !
!100s.hea  !

Then we type

-->unix ('sqrs -r 100')
 ans  =
 
    0.

A warning in the terminal where Scilab was initiated indicates that we could consider re-sampling at 250 Hz and a new file called `100.qrs' appears

-->unix_g('ls')        
 ans  =
 
!100.atr   !
!          !
!100.dat   !
!          !
!100.hea   !
!          !
!100.qrs   !
!          !
!100s.atr  !
!          !
!100s.dat  !
!          !
!100s.hea  !

Now we are going to see the content of `100.qrs'. It is an annotation file and can be seen with `rdann'

 
-->unix_x('rdann -r 100 -a qrs')

 
    0:00.194       70     N    0    0    0
    0:01.005      362     N    0    0    0
    0:01.822      656     N    0    0    0
    0:02.608      939     N    0    0    0
    0:03.400     1224     N    0    0    0
    0:04.188     1508     N    0    0    0...

We would like to compare the result with `100.atr'. We have a function to compare both files

-->unix_x('bxb -r 100 -a atr qrs')

We get

Beat-by-beat comparison results for record 100
Reference annotator: atr
     Test annotator: qrs

               Algorithm
        n    v    f    q    o    x
   _______________________________
 N | 1900    0    0    0    1    0
 V |    1    0    0    0    0    0
 F |    0    0    0    0    0    0
 Q |    0    0    0    0    0    0
 O |    0    0    0    0
 X |    0    0    0    0

           QRS sensitivity:  99.95% (1901/1902)
 QRS positive predictivity: 100.00% (1901/1901)
           VEB sensitivity:   0.00% (0/1)
 VEB positive predictivity:      -  (0/0)
   VEB false positive rate:  0.000% (0/1900)


  Beats missed in shutdown:   0.00% (0/1902)
      N missed in shutdown:   0.00% (0/1901)
      V missed in shutdown:   0.00% (0/1)
      F missed in shutdown:      -  (0/0)
       Total shutdown time:     0 seconds

Finally we are going to encode a signal in MIT format. Instead of storing real signals, we will store three columns of random values in the range -5000 to 5000. We store the data as if they would have been sampled at 100 Hz

-->rand('seed',0)                                
 
-->x = rand(10,3)*10000 - 5000                   
 x  =
 
! - 2886.7513    608.48606  - 1923.9093 !
!   2560.4385    1623.5694    4329.6162 !
! - 4997.7887    2263.5068  - 2853.9921 !
! - 1696.7291  - 3014.8562  - 1873.58   !
!   1653.811     442.57316  - 1383.639  !
!   1283.9179  - 2679.2521  - 2077.7334 !
!   3497.4524  - 2687.7628    664.24882 !
!   1857.3102  - 2835.3674  - 173.52803 !
!   3782.1648    3833.8878  - 1678.2811 !
! - 4316.2596    1525.1349    935.0947  !

-->fprintfMat('dummy',x)          

-->unix('cat dummy| wrsamp -o test -F 100 0 1 2')
 ans  =
 
    0.

Now we are going to see the result

-->unix_x('rdsamp -r test')   
 
-->rand('seed',0)             
 
-->fprintfMat('dummy',x)                         
 
-->unix('cat dummy| wrsamp -o test -F 100 0 1 2')
 ans  =
 
    0.  
 
-->unix_x('rdsamp -r test')

And the result is

     0	-2886	  608	-1923
     1	 2560	 1623	 4329
     2	-4997	 2263	-2853
     3	-1696	-3014	-1873
     4	 1653	  442	-1383
     5	 1283	-2679	-2077
     6	 3497	-2687	  664
     7	 1857	-2835	 -173
     8	 3782	 3833	-1678
     9	-4316	 1525	  935

If we see the header

-->unix_x('cat test.header')

then, the result is:

test 3 100 10
test.dat 16 200 12 0 -2886 737 0 column 0
test.dat 16 200 12 0 608 -921 0 column 1
test.dat 16 200 12 0 -1923 -6032 0 column 2

We have closed the circle. We can send data from Scilab to WFDB or from WFDB to Scilab. So we can choose the tool more adapted to our work. These chapters are only a little introduction to the properties of WFDB and they are centered in the connection with Scilab. But WFDB has many more functions that are worth to explore.