How to use the wfdb.processing function in wfdb
To help you get started, we’ve selected a few wfdb examples, based on popular ways it is used in public projects.
Secure your code as it's written. Use Snyk Code to scan source code in minutes - no build needed - and fix issues immediately.
JohnDoenut / biopeaks / benchmark_resp.py View on Github 
# annotation places peak towards the middle); for non-clipped peaks, the
# agreement of manual and algorithmic annotation is within a range smaller
# than 700 msec (enable plotting to confirm)
# unilateral extend of acceptance margin centered on each algopeak; in sec
acceptance_margin = 0.7
comparitor1 = processing.compare_annotations(manupeaks1,
np.ravel(algopeaks),
int(acceptance_margin *
sfreq))
tp1 = comparitor1.tp
fp1 = comparitor1.fp
fn1 = comparitor1.fn
comparitor2 = processing.compare_annotations(manupeaks2,
np.ravel(algopeaks),
int(acceptance_margin *
sfreq))
tp2 = comparitor2.tp
fp2 = comparitor2.fp
fn2 = comparitor2.fn
# calculate two metrics for benchmarking (according to AAMI guidelines):
# 1. sensitivity: how many of the manually annotated peaks does the
# algorithm annotate as peaks (TP / TP + FN)?
# 2. precision: out of all peaks that are algorithmically annotated as
# peaks (TP + FP), how many are correct (TP)?
sensitivity1.append(float(tp1) / (tp1 + fn1))
precision1.append(float(tp1) / (tp1 + fp1))
sensitivity2.append(float(tp2) / (tp2 + fn2))
precision2.append(float(tp2) / (tp2 + fp2))# plt.scatter(algopeaks, resp[algopeaks], c='g', marker='X', s=150)
# perform benchmarking against each annotator seperately; an
# algorythmically annotated peaks is scored as true positives if it is
# within 700 msec of a manually annotated peak, this relatively large
# window was chosen to account for the clipping that occurs in many of the
# recordings; clipping results in large palteaus that make placement of
# peak somewhat arbitrary (algorithm places peak at the edges while manual
# annotation places peak towards the middle); for non-clipped peaks, the
# agreement of manual and algorithmic annotation is within a range smaller
# than 700 msec (enable plotting to confirm)
# unilateral extend of acceptance margin centered on each algopeak; in sec
acceptance_margin = 0.7
comparitor1 = processing.compare_annotations(manupeaks1,
np.ravel(algopeaks),
int(acceptance_margin *
sfreq))
tp1 = comparitor1.tp
fp1 = comparitor1.fp
fn1 = comparitor1.fn
comparitor2 = processing.compare_annotations(manupeaks2,
np.ravel(algopeaks),
int(acceptance_margin *
sfreq))
tp2 = comparitor2.tp
fp2 = comparitor2.fp
fn2 = comparitor2.fn
# calculate two metrics for benchmarking (according to AAMI guidelines):wfdb
The WFDB Python package: tools for reading, writing, and processing physiologic signals and annotations.
