def setup_module_load(request):
"""
Fixture to setup module. Performs check for presence of test files,
creates initial batch object.
"""
path = os.path.join(os.path.dirname(os.path.realpath(__file__)), 'data')
files = [
"A00001.hea", "A00001.mat", "A00002.hea", "A00002.mat", "A00004.hea",
"A00004.mat", "A00005.hea", "A00005.mat", "A00008.hea", "A00008.mat",
"A00013.hea", "A00013.mat", "REFERENCE.csv"
]
# TODO: make better test for presence of files .hea and
# REFERENCE.csv
if np.all([os.path.isfile(os.path.join(path, file)) for file in files]):
ind = bf.FilesIndex(path=os.path.join(path, 'A*.hea'),
no_ext=True,
sort=True)
else:
raise FileNotFoundError("Test files not found in 'tests/data/'!")
def teardown_module_load():
"""
Teardown module
"""
request.addfinalizer(teardown_module_load)
return ind, path
def test_load_wfdb_annotation(self, setup_module_load): #pylint: disable=redefined-outer-name
"""
Testing wfdb loader for annotation.
"""
# Arrange
path = setup_module_load[1]
ind = bf.FilesIndex(path=os.path.join(path, 'sel100.hea'),
no_ext=True,
sort=True)
batch = EcgBatch(ind)
# Act
batch = batch.load(fmt="wfdb",
components=["signal", "annotation", "meta"],
ann_ext="pu1")
# Assert
assert isinstance(batch.signal, np.ndarray)
assert isinstance(batch.meta, np.ndarray)
assert isinstance(batch.annotation, np.ndarray)
assert batch.signal.shape == (1, )
assert batch.annotation.shape == (1, )
assert batch.meta.shape == (1, )
assert isinstance(batch.signal[0], np.ndarray)
assert isinstance(batch.annotation[0], dict)
assert isinstance(batch.meta[0], dict)
assert 'annsamp' in batch.annotation[0]
assert 'anntype' in batch.annotation[0]
del batch
def setup_class_methods(request):
"""
Fixture to setup class to test EcgBatch methods separately.
"""
path = os.path.join(os.path.dirname(os.path.realpath(__file__)), 'data/')
ind = bf.FilesIndex(path=os.path.join(path, 'A*.hea'),
no_ext=True,
sort=True)
batch_loaded = (EcgBatch(ind, unique_labels=["A", "O", "N"]).load(
fmt="wfdb", components=["signal", "annotation", "meta"]))
def teardown_class_methods():
"""
Teardown class
"""
request.addfinalizer(teardown_class_methods)
return batch_loaded
def test_load_wav(self, setup_module_load): #pylint: disable=redefined-outer-name
"""
Testing EDF loader.
"""
# Arrange
path = setup_module_load[1]
ind = bf.FilesIndex(path=os.path.join(path, 'sample*.wav'),
no_ext=True,
sort=True)
batch = EcgBatch(ind)
# Act
batch = batch.load(fmt="wav",
components=["signal", "annotation", "meta"])
# Assert
assert isinstance(batch.signal, np.ndarray)
assert isinstance(batch.meta, np.ndarray)
assert isinstance(batch.annotation, np.ndarray)
assert batch.signal.shape == (1, )
assert batch.annotation.shape == (1, )
assert batch.meta.shape == (1, )
assert isinstance(batch.signal[0], np.ndarray)
assert isinstance(batch.annotation[0], dict)
assert isinstance(batch.meta[0], dict)
del batch
import os
import cardio.batchflow as bf
import warnings
import numpy as np
from cardio import EcgBatch
from my_tools import calculate_old_metrics
from PanTompkinsAlgorithm import testPipeline, PanTompkinsPipeline
warnings.filterwarnings('ignore')
SIGNALS_PATH = "data\\qt-database-1.0.0" # set path to QT database
SIGNALS_MASK = os.path.join(SIGNALS_PATH, "*.hea")
index = bf.FilesIndex(path=SIGNALS_MASK, no_ext=True, sort=True)
dtst = bf.Dataset(index, batch_class=EcgBatch)
dtst.split([0.1, 0.9])
pipeline = PanTompkinsPipeline(batch_size=len(dtst.train.indices),
annot="pan_tomp_annotation")
ppl_inits = (dtst.train >> pipeline).run()
batch: EcgBatch = ppl_inits.next_batch(len(dtst.train.indices))
parameters = calculate_old_metrics(batch, np.array(list([1]), np.int64), 0,
"pan_tomp_annotation")
print("Pan-Tompkins" + "\tsensitivity= " + str(parameters["sensitivity"]) +
"\tspecificity= " + str(parameters["specificity"]))
print("end")
from tensorflow.keras import optimizers
from sklearn.model_selection import train_test_split
#%%
# https://www.physionet.org/content/qtdb/1.0.0/
# where the data was saved
#SIGNALS_FOLDER = "\\qt-database-1.0.0" # the data was saved within the same folder
current_dir = os.getcwd()
SIGNALS_FOLDER = current_dir + "\\qt-database-1.0.0"
SIGNALS_MASK = os.path.join(SIGNALS_FOLDER, "*.hea") # .hea, header file, describing signal file contents and format
# when data comes from a file system, it might be convenient to use 'FileIndex',
# it is coming from batchflow module.
index = bf.FilesIndex(path=SIGNALS_MASK, # directory of the file location
no_ext=True, # when filenames contain extensions which are not part of the id, then they maybe stripped with this option
sort=True) # sort the order of your file index
dtst = bf.Dataset(index, batch_class=EcgBatch) # batch_class holds the data and contains processing functions. refer to documentation 'batch classes'
#%%
def expand_annotation(annsamp_0, anntype_0, length):
"""
Unravel annotation
anntype_0 and annsamp_0 are with the same size.