Datasets:
Helen Ryan
Helen-Ryan/
MIT-BIH-Arrhythmia
0
Downloads: 1
Diff of /mit-bih-arrhythmia-database-1.0.0/mitdbdir/intro.htm [000000] .. [b755cf]

Switch to unified view

a b/mit-bih-arrhythmia-database-1.0.0/mitdbdir/intro.htm 

  1
  
<html>





  
  

  2
  
<head>





  
  

  3
  
<title>MIT-BIH Arrhythmia Database Directory (Introduction)</title>





  
  

  4
  
</head>





  
  

  5
  
<body bgcolor="#FFFFFF">





  
  

  6
  





  
  

  7
  
<a href="mitdbdir.htm"><h1 align=center>MIT-BIH Arrhythmia Database Directory





  
  

  8
  
</h1></a>





  
  

  9
  





  
  

  10
  
<p>





  
  

  11
  
<b>Next:</b> <a href="records.htm">Records</a>





  
  

  12
  
<b>Up:</b> <a href="mitdbdir.htm#toc">Contents</a>





  
  

  13
  
<b>Previous:</b> <a href="foreword.htm">Foreword</a>





  
  

  14
  





  
  

  15
  
<a name="intro"><h1>Introduction</h1></a>





  
  

  16
  
<p>





  
  

  17
  
This introduction describes how the database records were





  
  

  18
  
obtained, and discusses the characteristics of the recorded signals.





  
  

  19
  
Following these notes are annotated ``full disclosure'' plots of the entire





  
  

  20
  
database.  These can be useful for obtaining an overall





  
  

  21
  
impression of the contents of individual records.  Following the





  
  

  22
  
``full disclosure'' plots are sample ECG strips.  These strips





  
  

  23
  
were chosen to illustrate the salient features of each record.





  
  

  24
  
Next are notes on the important features of each record.





  
  

  25
  
These notes also include background information on the subjects,





  
  

  26
  
including their medications.





  
  

  27
  
At the end of the book are tables of rhythms and annotations, which





  
  

  28
  
summarize the contents of the database.





  
  

  29
  
These tables can be helpful in finding a record with a specific





  
  

  30
  
set of characteristics.





  
  

  31
  





  
  

  32
  
<a name="selection"><h2>Selection criteria</h2>





  
  

  33
  
<p>





  
  

  34
  
The source of the ECGs included in the MIT-BIH Arrhythmia Database is a





  
  

  35
  
set of over 4000 long-term Holter recordings that were obtained





  
  

  36
  
by the Beth Israel Hospital Arrhythmia Laboratory between 1975 and 1979.





  
  

  37
  
Approximately 60% of these recordings were obtained from inpatients.





  
  

  38
  
The database contains 23 records





  
  

  39
  
(numbered from 100 to 124 inclusive with some numbers missing)





  
  

  40
  
chosen at random from this set, and 25 records





  
  

  41
  
(numbered from 200 to 234 inclusive, again with some numbers missing)





  
  

  42
  
selected from the same set to include a variety of rare but clinically





  
  

  43
  
important phenomena that would not be well-represented





  
  

  44
  
by a small random sample of Holter recordings.





  
  

  45
  
Each of the 48 records is slightly over 30 minutes long.





  
  

  46
  
<p>





  
  

  47
  
The first group is intended to serve as a representative sample of the





  
  

  48
  
variety of waveforms and artifact that an arrhythmia detector might





  
  

  49
  
encounter in routine clinical use.  A table of random numbers was used





  
  

  50
  
to select tapes, and then to select half-hour segments of them.





  
  

  51
  
Segments selected in this way were excluded only if neither of the two





  
  

  52
  
ECG signals was of adequate quality for analysis by human experts.





  
  

  53
  
<p>





  
  

  54
  
Records in the second group were chosen to





  
  

  55
  
include complex ventricular, junctional,





  
  

  56
  
and supraventricular arrhythmias and conduction abnormalities.  Several





  
  

  57
  
of these records were selected because features of the rhythm, QRS





  
  

  58
  
morphology variation, or signal quality may be expected to present





  
  

  59
  
significant difficulty to arrhythmia detectors;  these records have





  
  

  60
  
gained considerable notoriety among database users.





  
  

  61
  
<p>





  
  

  62
  
The subjects were 25 men aged 32 to 89 years, and 22 women aged 23 to 89





  
  

  63
  
years.  (Records 201 and 202 came from the same male subject.)





  
  

  64
  





  
  

  65
  
<a name="leads"><h2>ECG lead configuration</h2></a>





  
  

  66
  
<p>





  
  

  67
  
In most records, the upper signal is a modified limb lead II (MLII),





  
  

  68
  
obtained by placing the electrodes on the chest.  The lower signal is





  
  

  69
  
usually a modified lead V1 (occasionally V2 or V5, and in one instance V4);





  
  

  70
  
as for the upper signal, the electrodes are also placed on the chest.





  
  

  71
  
This configuration is routinely used by the BIH Arrhythmia Laboratory.





  
  

  72
  
Normal QRS complexes are usually prominent in the upper signal.





  
  

  73
  
The lead axis for the lower signal may be nearly orthogonal to the mean





  
  

  74
  
cardiac electrical axis, however (i.e., normal beats are usually





  
  

  75
  
biphasic and may be nearly isoelectric).





  
  

  76
  
Thus normal beats are frequently difficult to discern in the lower signal,





  
  

  77
  
although ectopic beats will often be more prominent (see, for example, record





  
  

  78
  
106).





  
  

  79
  
A notable exception is record 114, for which the signals were reversed.





  
  

  80
  
Since this happens occasionally in clinical practice, arrhythmia detectors





  
  

  81
  
should be equipped to deal with this situation.





  
  

  82
  
In records 102 and 104, it was not possible to use modified lead II because of





  
  

  83
  
surgical dressings on the patients;





  
  

  84
  
modified lead V5 was used for the upper signal in these records.





  
  

  85
  





  
  

  86
  
<a name="analog"><h2>Analog recording and playback</h2></a>





  
  

  87
  
<p>





  
  

  88
  
The original analog recordings were made using nine Del Mar Avionics





  
  

  89
  
model 445 two-channel recorders, designated <i>A</i> through <i>I</i>:





  
  

  90
  
<table border>





  
  

  91
  
<tr><th><i>Recorder</i></th><th><i>Records</i></th></tr>





  
  

  92
  
<tr><td align=center><i>A</i></td><td>102, 107, 111, 115, 121</td></tr>





  
  

  93
  
<tr><td align=center><i>B</i></td><td>212</td></tr>





  
  

  94
  
<tr><td align=center><i>C</i></td><td>203</td></tr>





  
  

  95
  
<tr><td align=center><i>D</i></td><td>118, 124, 217</td></tr>





  
  

  96
  
<tr><td align=center><i>E</i></td>





  
  

  97
  
  <td>101, 103, 106, 108, 112, 117, 119, 122, 209, 219, 220, 223, 233</td></tr>





  
  

  98
  
<tr><td align=center><i>F</i></td>





  
  

  99
  
  <td>104, 109, 123, 205, 207, 210, 215, 221</td></tr>





  
  

  100
  
<tr><td align=center><i>G</i></td>





  
  

  101
  
  <td>100, 105, 114, 116, 213, 214, 222, 228</td></tr>





  
  

  102
  
<tr><td align=center><i>H</i></td><td>113, 201, 202, 231</td></tr>





  
  

  103
  
<tr><td align=center><i>I</i></td><td>200, 230, 232, 234</td></tr>





  
  

  104
  
</table>





  
  

  105
  
<br>





  
  

  106
  
(It is not known which recorder was used for record 208.)





  
  

  107
  
<p>





  
  

  108
  
During the digitization process, the analog recordings were played back





  
  

  109
  
on a Del Mar Avionics model 660 unit.  The analog tapes used for records





  
  

  110
  
112, 115 through 124, 205, 220, 223, and 230 through 234 were played back





  
  

  111
  
and digitized at twice real time;  the rest were played back at real time





  
  

  112
  
using a specially constructed capstan for the model 660 unit.





  
  

  113
  
Skew between the two signals was found to be as great as 40





  
  

  114
  
milliseconds for some of the analog recorders.





  
  

  115
  
In addition to the fixed skew that results from extremely small differences





  
  

  116
  
in the orientations of the tape heads on the recorder and the playback unit,





  
  

  117
  
microscopic vertical wobbling of the tape, either during recording or playback,





  
  

  118
  
introduces a variable skew, which may be comparable in magnitude to the fixed





  
  

  119
  
skew.





  
  

  120
  
This problem (which also





  
  

  121
  
occurs on the AHA database) may present difficulties for certain two-channel





  
  

  122
  
analysis methods designed for real-time applications.





  
  

  123
  
<p>





  
  

  124
  
Minor tape speed variations should not pose problems for typical arrhythmia





  
  

  125
  
detectors.  It is difficult to avoid tape sticking or slippage during low-speed





  
  

  126
  
playback, and several episodes of tape slippage were noted and marked with





  
  

  127
  
comment annotations.  Wow and flutter should be studied carefully in the





  
  

  128
  
context of heart-rate variability studies, since flutter compensation





  
  

  129
  
was not possible in these recordings.  A number of frequency-domain artifacts





  
  

  130
  
have been identified and related to specific mechanical components of the





  
  

  131
  
recorders and the playback unit:





  
  

  132
  
<table border>





  
  

  133
  
<tr><th><i>Frequency (Hz)</i></th><th><i>Source</i></th></tr>





  
  

  134
  
<tr><td align=center>0.042</td><td>Recorder pressure wheel</td></tr>





  
  

  135
  
<tr><td align=center>0.083</td><td>Playback unit capstan (for twice real-time playback)</td></tr>





  
  

  136
  
<tr><td align=center>0.090</td><td>Recorder capstan</td></tr>





  
  

  137
  
<tr><td align=center>0.167</td>





  
  

  138
  
  <td>Playback unit capstan (for real-time playback)</td></tr>





  
  

  139
  
<tr><td align=center>0.18-0.10</td>





  
  

  140
  
  <td>Takeup reel (frequency decreases over time)</td></tr>





  
  

  141
  
<tr><td align=center>0.20-0.36</td>





  
  

  142
  
  <td>Supply reel (frequency increases over time)</td></tr>





  
  

  143
  
</table>





  
  

  144
  
<br>





  
  

  145
  
The most significant of these artifacts by far is the 0.167 Hz artifact on





  
  

  146
  
recordings that were played back at real time.  The next largest is the





  
  

  147
  
0.090 Hz artifact;  the 0.083 Hz artifact on recordings that were played back





  
  

  148
  
at twice real-time is of roughly the same magnitude as the 0.090 Hz artifact.





  
  

  149
  
The 0.042 Hz artifact is of much lower magnitude.  Other frequencies





  
  

  150
  
related to the drive train (at 0.42 Hz, 1.96 Hz, 9.1 Hz, and





  
  

  151
  
42 Hz) do not appear as noticeable artifacts.





  
  

  152
  
The frequencies of the last two artifacts listed in the table depend on





  
  

  153
  
how much tape is on the supply and takeup reels;  the supply reel causes





  
  

  154
  
a much more noticeable artifact than does the takeup reel.  Other





  
  

  155
  
frequency-domain artifacts generated by the supply reel appear in the





  
  

  156
  
0.10-0.18 Hz and 0.30-0.54 Hz bands.





  
  

  157
  
<p>





  
  

  158
  
Four of the 48 records (102, 104, 107, and 217) include paced beats.





  
  

  159
  
The original analog recordings do not represent the pacemaker artifacts





  
  

  160
  
with sufficient fidelity to permit them to be recognized by pulse amplitude





  
  

  161
  
(or slew rate) and duration alone, the method commonly used for real-time





  
  

  162
  
processing.  The database records reproduce the analog recordings with





  
  

  163
  
sufficient fidelity to permit use of pacemaker artifact detectors designed for





  
  

  164
  
tape analysis, however.





  
  

  165
  





  
  

  166
  
<a name="digitization"><h2>Digitization</h2></a>





  
  

  167
  
<p>





  
  

  168
  
The analog outputs of the playback unit





  
  

  169
  
were filtered to limit analog-to-digital converter (ADC)





  
  

  170
  
saturation and for anti-aliasing,





  
  

  171
  
using a passband from 0.1 to 100 Hz relative to real time, well





  
  

  172
  
beyond the lowest and highest frequencies recoverable from the recordings.





  
  

  173
  
The bandpass-filtered signals were





  
  

  174
  
digitized at 360 Hz per signal relative to real time





  
  

  175
  
using hardware constructed at the MIT Biomedical Engineering Center and





  
  

  176
  
at the BIH Biomedical Engineering Laboratory.





  
  

  177
  
The sampling frequency was chosen to facilitate implementations of 60 Hz





  
  

  178
  
(mains frequency) digital notch filters in arrhythmia detectors.





  
  

  179
  
Since the recorders were battery-powered, most of the 60 Hz noise present





  
  

  180
  
in the database arose during playback.





  
  

  181
  
In those records that were digitized at twice real time, this noise appears





  
  

  182
  
at 30 Hz (and multiples of 30 Hz) relative to real time.





  
  

  183
  
<p>





  
  

  184
  
Samples were acquired from each signal almost simultaneously (the intersignal





  
  

  185
  
sampling skew was on the order of a few microseconds).  As noted above, analog





  
  

  186
  
tape skew was several orders of magnitude larger.  The ADCs were





  
  

  187
  
unipolar, with 11-bit resolution over a &plusmn;5 mV range.  Sample values thus





  
  

  188
  
range from 0 to 2047 inclusive, with a value of 1024 corresponding to zero





  
  

  189
  
volts.





  
  

  190
  
<p>





  
  

  191
  
The 11-bit samples were originally recorded in 8-bit first difference format





  
  

  192
  
(this was necessary because of limited mass storage capacity).  Given the





  
  

  193
  
sampling frequency and the resolution of the ADC, the difference encoding





  
  

  194
  
implies a maximum recordable slew rate of &plusmn;225 mV/s.  In practice, this





  
  

  195
  
limit was exceeded by the input signals very infrequently, only during severe





  
  

  196
  
noise on a small number of records.  The effect on the quality of the recorded





  
  

  197
  
signals is totally negligible.  On this CD-ROM, the samples have been





  
  

  198
  
reconstructed from the first differences and stored as pairs of 12-bit





  
  

  199
  
amplitudes packed in triplets of consecutive bytes (for details on the storage





  
  

  200
  
format, see <a href="/physiotools/wag/signal-5.htm">signal(5)</a>).





  
  

  201
  





  
  

  202
  
<a name="annotations"><h2>Annotations</h2></a>





  
  

  203
  
<p>





  
  

  204
  
An initial set of beat labels was produced by a simple slope-sensitive





  
  

  205
  
QRS detector, which marked each detected event as a normal beat.  Two





  
  

  206
  
identical 150-foot chart recordings were printed for each 30-minute record,





  
  

  207
  
with these initial beat labels in the margin.





  
  

  208
  
For each record, the two charts were given to two cardiologists, who worked





  
  

  209
  
on them independently.  The cardiologists added additional beat labels





  
  

  210
  
where the detector missed beats, deleted false detections as necessary,





  
  

  211
  
and changed the labels for all abnormal beats.  They also added rhythm





  
  

  212
  
labels, signal quality labels, and comments.





  
  

  213
  
<p>





  
  

  214
  
The annotations were transcribed from the paper chart recordings.





  
  

  215
  
Once both sets of cardiologists' annotations for a given record





  
  

  216
  
had been transcribed and verified, they were automatically compared





  
  

  217
  
beat-by-beat, and another chart recording was printed.  This chart showed





  
  

  218
  
the cardiologists' annotations in the margin, with all discrepancies





  
  

  219
  
highlighted.  Each discrepancy was reviewed and resolved by consensus.





  
  

  220
  
The corrections were transcribed, and the annotations were then analyzed





  
  

  221
  
by an auditing program, which checked them for consistency and which





  
  

  222
  
located the ten longest and shortest R-R intervals in each record (to





  
  

  223
  
identify possible missing or falsely detected beats).





  
  

  224
  
<p>





  
  

  225
  
In early copies of the database, most beat labels were placed





  
  

  226
  
at the R-wave peak, but manually inserted labels were not always





  
  

  227
  
located precisely at the peak.





  
  

  228
  
In copies of the database made since 1983, the beat labels have been shifted





  
  

  229
  
from their original locations.  





  
  

  230
  
The ECG (usually the upper signal) was digitally bandpass-filtered to





  
  

  231
  
emphasize the QRS complexes, and each beat label was moved to the major local





  
  

  232
  
extremum, after correction for phase shift in the filter.





  
  

  233
  
A few noisy beats were manually realigned.





  
  

  234
  
This process was applied to all records except record 117 in 1983; the





  
  

  235
  
beat labels for record 117 were not realigned until March 1998, however.





  
  

  236
  
The result is that annotations generally appear at the R-wave peak, and





  
  

  237
  
are located with sufficient accuracy to make the reference annotation





  
  

  238
  
files usable for studies requiring waveform averaging and for





  
  

  239
  
heart rate variability studies (but note the comments





  
  

  240
  
with respect to analog tape wow and flutter above).





  
  

  241
  
In the annotated ECG plots produced by <tt>psfd</tt> and <tt>pschart</tt>,





  
  

  242
  
and in printed copies of this directory, each label is placed so that the





  
  

  243
  
fiducial mark for the annotation corresponds to the left edge of the label.





  
  

  244
  
<p>





  
  

  245
  
The database contains approximately 109,000 beat labels.  Sixteen





  
  

  246
  
were corrected in the first seven years after the database was released in 1980





  
  

  247
  
(in records 104, 108, 114, 203, 207, 217, and 222);  in addition,





  
  

  248
  
all of the left bundle branch block beats in record 214 were originally





  
  

  249
  
labelled as normal beats.  The rhythm labels





  
  

  250
  
have been more substantially revised and now include notations for paced





  
  

  251
  
rhythm, bigeminy, and trigeminy, which were missing in early copies.





  
  

  252
  
<p>





  
  

  253
  
In October 1998, a rhythm label in record 203 was corrected.  In





  
  

  254
  
October 2001, a seventeenth error in the beat labels was discovered





  
  

  255
  
and corrected (in record 209).  In April 2003, 26 PVC annotations in





  
  

  256
  
record 119 were manually realigned by small amounts (up to 74 ms).  In





  
  

  257
  
May 2003, an eighteenth error in the beat labels was discovered and





  
  

  258
  
corrected (in record 214).  In April 2005, many of the episodes





  
  

  259
  
previously labelled as atrial fibrillation in record 222 were





  
  

  260
  
partially or completely relabelled as atrial flutter.  In April 2008,





  
  

  261
  
three beat labels were corrected (two in record 108, and one in record





  
  

  262
  
215).  In June 2010, the 22nd and 23rd errors in the beat labels were





  
  

  263
  
found and corrected (both in record 203).  Thanks to Bob Bruce, Pat





  
  

  264
  
Hamilton, Yin Dengfeng, Roger Mark, Sebastian Vasquez, and Mariano





  
  

  265
  
Llamedo Soria for finding and reporting these errors.





  
  

  266
  





  
  

  267
  
<hr>





  
  

  268
  





  
  

  269
  
<a name="symbols"><h1>Symbols used in plots</h1></a>





  
  

  270
  





  
  

  271
  
<p>





  
  

  272
  
[An expanded and updated version of the table below can be found at





  
  

  273
  
<a href="/physiobank/annotations.shtml">





  
  

  274
  
<tt>http://www.physionet.org/physiobank/annotations.shtml</tt></a>.]





  
  

  275
  





  
  

  276
  
<p>





  
  

  277
  
<table border>





  
  

  278
  
<tr><th><i>Symbol</i></th><th><i>Meaning</i></th></tr>





  
  

  279
  
<tr><td><b>&middot;</b> <i>or</i> N</td><td>Normal beat</td></tr>





  
  

  280
  
<tr><td>L</td><td>Left bundle branch block beat</td></tr>





  
  

  281
  
<tr><td>R</td><td>Right bundle branch block beat</td></tr>





  
  

  282
  
<tr><td>A</td><td>Atrial premature beat</td></tr>





  
  

  283
  
<tr><td>a</td><td>Aberrated atrial premature beat</td></tr>





  
  

  284
  
<tr><td>J</td><td>Nodal (junctional) premature beat</td></tr>





  
  

  285
  
<tr><td>S</td><td>Supraventricular premature beat</td></tr>





  
  

  286
  
<tr><td>V</td><td>Premature ventricular contraction</td></tr>





  
  

  287
  
<tr><td>F</td><td>Fusion of ventricular and normal beat</td></tr>





  
  

  288
  
<tr><td>[</td><td>Start of ventricular flutter/fibrillation</td></tr>





  
  

  289
  
<tr><td>!</td><td>Ventricular flutter wave</td></tr>





  
  

  290
  
<tr><td>]</td><td>End of ventricular flutter/fibrillation</td></tr>





  
  

  291
  
<tr><td>e</td><td>Atrial escape beat</td></tr>





  
  

  292
  
<tr><td>j</td><td>Nodal (junctional) escape beat</td></tr>





  
  

  293
  
<tr><td>E</td><td>Ventricular escape beat</td></tr>





  
  

  294
  
<tr><td>/</td><td>Paced beat</td></tr>





  
  

  295
  
<tr><td>f</td><td>Fusion of paced and normal beat</td></tr>





  
  

  296
  
<tr><td>x</td><td>Non-conducted P-wave (blocked APB)</td></tr>





  
  

  297
  
<tr><td>Q</td><td>Unclassifiable beat</td></tr>





  
  

  298
  
<tr><td>|</td><td>Isolated QRS-like artifact</td></tr>





  
  

  299
  
<tr><td colspan=2 align=center>Rhythm annotations appear <i>below</i> the





  
  

  300
  
level used for beat annotations:</td></tr>





  
  

  301
  
<tr><td>(AB</td><td>Atrial bigeminy</td></tr>





  
  

  302
  
<tr><td>(AFIB</td><td>Atrial fibrillation</td></tr>





  
  

  303
  
<tr><td>(AFL</td><td>Atrial flutter</td></tr>





  
  

  304
  
<tr><td>(B</td><td>Ventricular bigeminy</td></tr>





  
  

  305
  
<tr><td>(BII</td><td>2&deg; heart block</td></tr>





  
  

  306
  
<tr><td>(IVR</td><td>Idioventricular rhythm</td></tr>





  
  

  307
  
<tr><td>(N</td><td>Normal sinus rhythm</td></tr>





  
  

  308
  
<tr><td>(NOD</td><td>Nodal (A-V junctional) rhythm</td></tr>





  
  

  309
  
<tr><td>(P</td><td>Paced rhythm</td></tr>





  
  

  310
  
<tr><td>(PREX</td><td>Pre-excitation (WPW)</td></tr>





  
  

  311
  
<tr><td>(SBR</td><td>Sinus bradycardia</td></tr>





  
  

  312
  
<tr><td>(SVTA</td><td>Supraventricular tachyarrhythmia</td></tr>





  
  

  313
  
<tr><td>(T</td><td>Ventricular trigeminy</td></tr>





  
  

  314
  
<tr><td>(VFL</td><td>Ventricular flutter</td></tr>





  
  

  315
  
<tr><td>(VT</td><td>Ventricular tachycardia</td></tr>





  
  

  316
  
<tr><td colspan=2 align=center>Signal quality and comment annotations appear <i>above</i>





  
  

  317
  
the level used for beat annotations:</td></tr>





  
  

  318
  
<tr><td><i>qq</i></td>





  
  

  319
  
<td>





  
  

  320
  
Signal quality change:  the first character (`c' or `n') indicates the quality





  
  

  321
  
of the upper signal (clean or noisy), and the second character indicates the





  
  

  322
  
quality of the lower signal</td></tr>





  
  

  323
  
<tr><td>U</td><td>Extreme noise or signal loss in both signals:  ECG is unreadable</td></tr>





  
  

  324
  
<tr><td>M (<i>or</i> MISSB)</td><td>Missed beat</td></tr>





  
  

  325
  
<tr><td>P (<i>or</i> PSE)</td><td>Pause</td></tr>





  
  

  326
  
<tr><td>T (<i>or</i> TS)</td><td>Tape slippage</td></tr>





  
  

  327
  
</table>





  
  

  328
  





  
  

  329
  
<HR>





  
  

  330
  
<P><ADDRESS>





  
  

  331
  
<I><A HREF="mailto:george@mit.edu">George B. Moody (<tt>george@mit.edu</tt>)</A></ADDRESS></I><BR>





  
  

  332
  
24 May 1997





  
  

  333
  
<br>





  
  

  334
  
<i>Revised 24 June 2010</i>





  
  

  335
  
</body>





  
  

  336
  
</html>