Biomedical Engineering Reference
In-Depth Information
Table 2.7 Coefficient of variation figures (in 10
-3
)
Patient file ID
in Physionet
Lead
Amplitude features
Wave interval features
QRS
amplitude
T-height
P-height
RR
QRS
QT
P-width
T-width
P117/
s0292
(N)
II
0.734
0.225
0.412
0.963
0.099
0.145
0.252
0.096
aVF
1.106
0.197
0.328
0.978
0.384
0.627
0.489
0.224
V2
1.141
1.522
0.027
0.970
0.014
0.013
0.967
0.012
V5
1.782
0.309
0.384
0.956
0.032
0.124
0.064
0.057
P105/
s0303(N)
II
7.626
6.475
0.155
1.329
0.245
4.726
0.105
1.798
aVF
6.036
2.403
0.196
1.336
0.191
38.10
0.119
3.423
V2
2.459
0.280
0.221
1.348
0.045
4.802
0.404
1.387
V5
4.870
0.315
0.149
1.328
0.028
0.045
0.181
0.129
P246/
s0472(N)
II
10.76
7.269
1.294
0.064
0.191
34.83
0.065
0.622
aVF
0.506
3.161
0.764
0.065
0.506
31.63
0.014
0.702
V2
27.08
10.72
0.337
0.067
0.335
3.921
0.332
1.255
V5
23.58
3.161
0.764
0.065
0.506
31.63
0.014
0.702
P266/
s0502(N)
II
0.865
0.157
0.045
0.075
0.222
0.144
0.045
0.159
aVF
0.750
0.146
0.200
0.074
0.439
0.243
0.044
0.108
V2
0.183
1.106
0.063
0.068
0.060
0.051
0.918
0.022
V5
1.970
0.255
0.250
0.077
0.169
0.158
0.250
0.071
P277/
s0527(N)
II
1.002
0.058
0.257
3.199
0.083
0.173
0
0.052
aVF
2.110
4.570
0.116
3.194
0.238
18.39
0.034
5.887
V2
1.012
0.867
0.115
3.283
1.012
0.062
0.487
0.072
V5
0.449
0.542
0.115
3.199
0.104
0.091
1.127
0
P093/
s0378
(MI -
Inf)
II
29.67
0.817
24.00
0.515
0.015
55.60
0.198
0.600
aVF
3.567
2.208
7.7259.588
1.854
0.053
12.45
0.730
1.301
V3
0.664
2.546
0.291
0.543
0.180
0.209
0.507
0.013
V5
1.146
2.987
0.081
0.529
0.053
0.068
0.267
0.118
(Narration: N: Normal; MI: Myocardial Infarction; Inf: Inferior)
The developed algorithm is tested with 240 single-lead ECG data from ptb-db
database. Among the eight different features, the lowest COV is obtained with P-
wave width with a value of 0.302 9 10
-3
and highest with T-wave height, with a
value of 7.87 9 10
-3
. Considering all 12 leads in ptb-db files, an average value of
2.42 9 10
-3
is obtained.
2.5 Conclusion
In this chapter, the basic techniques of computerized ECG analysis techniques are
reviewed. R-peak detection is considered as one of the important criteria for
accurate feature extraction. The performance evaluation parameters for R peak are
also described. Algorithm steps for simple time-plane morphology-based feature
extraction developed by us are elaborated in detail.
Search WWH ::
Custom Search