wfdb
wfdb copied to clipboard
Published
20 hours ago •
bemoody
bemoody
Incorrect test measurements annotations count in mxm
Dear George
Using -f flag in mxm I noticed some invalid output in the count of test measurements annotations variable ntmeas
Sincerely Yuriy
[0:00:01.025] 369 = 0 0 0 71.76
[0:00:01.872] 674 = 0 0 0 70.82
[0:00:02.722] 980 = 0 0 0 70.59
[0:00:03.566] 1284 = 0 0 0 71.05
[0:00:04.408] 1587 = 0 0 0 71.29
[0:00:05.219] 1879 = 0 0 0 73.97
[0:00:06.041] 2175 = 0 0 0 72.97
[0:00:06.875] 2475 = 0 0 0 72.00
[0:00:07.683] 2766 = 0 0 0 74.23
[0:00:08.483] 3054 = 0 0 0 75.00
[0:00:09.344] 3364 = 0 0 0 69.68
[0:00:10.205] 3674 = 0 0 0 69.68
[0:00:11.050] 3978 = 0 0 0 71.05
[0:00:11.894] 4282 = 0 0 0 71.05
[0:00:12.722] 4580 = 0 0 0 72.48
[0:00:13.563] 4883 = 0 0 0 71.29
[0:00:14.363] 5171 = 0 0 0 75.00
[0:00:15.177] 5464 = 0 0 0 73.72
measurements annotation saved in htr annotator
$ mxm -r 118e00 -a htr htr -f 0 -t 10
Normalized RMS error: 0.0000% (n = 13)
Mean reference measurement: 66.3877
using -f 5 to start from 5th second results in the same n = 13 and incorrect mean reference measurement
$ mxm -r 118e00 -a htr htr -f 5 -t 10
Normalized RMS error: 0.0000% (n = 13)
Mean reference measurement: 39.0408
It seems that despite the use of -f ntmeas is counted from the signal start resulting in altered RMS and mean reference measurement.
$ mxm -r 118e00 -a htr htr
Normalized RMS error: 0.0000% (n = 2255)
Mean reference measurement: 65.2973
default -f is 5:00
$ mxm -r 118e00 -a htr htr -f 0
Normalized RMS error: 0.0000% (n = 2255)
Mean reference measurement: 76.8367
starting from -f 0 gives exactly the same n = 2255 and different mean reference measurement
as mxm is used for ANSI/AAMI-standard measurement-by-measurement annotation file comparator for FDA related tests using default mxm settings gives increased RMS performance and lower mean reference?