房颤心电数据库读取与转换

京元水煮肉

于 2025-01-08 20:44:57 发布

阅读量398

点赞数 6

文章标签：数据库 python 信号处理

本文链接：https://blog.csdn.net/mortinah/article/details/145017077

版权

设置数据库路径，这里主要使用MIT-BIH房颤数据库和正常窦性心律数据库。

mit_bih_af_path = 'F:/python/pycharm/AF_prediction/AF/data/MIT-BIHafdb/'
mit_bih_nsrdb_path = 'F:/python/pycharm/AF_prediction/AF/data/MIT-BIHnsrdb/'
path_save1='F:/python/pycharm/AF_prediction/AF/data/MIT-BIHafdb/csv/'
path_save2='F:/python/pycharm/AF_prediction/AF/data/MIT-BIHnsrdb/csv/'

DS1 = ['04015','04043','04048','04126','04746','04908','04936','05091','05121','05261','06426','06453','06995','07162','07859','07879','07910','08215','08219','08378','08405','08434','08455']
DS2 = ['16265','16272','16273','16420','16483','16539','16773','16786','16795','17052','17453','18177','18184','19088','19090','19093','19140','19830']

AFDB原始数据的读取：

def ReadTrainingSet1(DS):
    for ds in DS:
        dataSetLoc1 = mit_bih_af_path + ds
        record1 = wfdb.rdrecord(dataSetLoc1, physical=True)
        signal_annotation1 = wfdb.rdann(dataSetLoc1, 'atr')
        r_peak_annotation1 = wfdb.rdann(dataSetLoc1, 'qrs')
        # 获取关键信息
        af_ECG_rpeaks = r_peak_annotation1.sample
        af_ann_aux_note = signal_annotation1.aux_note
        af_ann_sample = signal_annotation1.sample
        ECG0 = record1.p_signal[:, 0]
        ECG1 = record1.p_signal[:, 1]
        trainingSet1[ds] = {'ECG0': ECG0, 'ECG1': ECG1, 'af_ann_aux_note': af_ann_aux_note, 'af_ann_sample': af_ann_sample, 'af_ECG_rpeaks': af_ECG_rpeaks}

trainingSet1 = {}
ReadTrainingSet1(DS1)

NSRDB原始数据的读取：

def ReadTrainingSet2(DS):
    for ds in DS:
        dataSetLoc2 = mit_bih_nsrdb_path + ds
        record2 = wfdb.rdrecord(dataSetLoc2, physical=True)
        signal_annotation2 = wfdb.rdann(dataSetLoc2, 'atr')
        # 获取关键信息
        nsr_ann_aux_note = signal_annotation2.aux_note
        nsr_ann_sample = signal_annotation2.sample
        nsr_ECG0 = record2.p_signal[:, 0]
        nsr_ECG1 = record2.p_signal[:, 1]
        trainingSet2[ds] = {'nsr_ECG0': nsr_ECG0, 'nsr_ECG1': nsr_ECG1, 'nsr_ann_aux_note': nsr_ann_aux_note, 'nsr_ann_sample': nsr_ann_sample}

trainingSet2 = {}
ReadTrainingSet2(DS2)

将数据保存为.csv格式，以AFDB为例：

headers = ['ECG0', 'ECG1', 'ann_aux_note', 'af_ann_sample', 'af_ECG_rpeaks']

for k, v in trainingSet1.items():
    df1 = pd.DataFrame({ 'af_ECG0': trainingSet1[k]['ECG0'], 'af_ECG1': trainingSet1[k]['ECG1']})
    df1.to_csv(path_save1 + k + '_afdb.csv', index=False)

    df2 = pd.DataFrame({'af_ann_aux_note': trainingSet1[k]['af_ann_aux_note'], 'af_ann_sample': trainingSet1[k]['af_ann_sample']})
    df2.to_csv(path_save1 + k + '_ann.csv', index=False)

    print(k,end = '')

检验数据的读取情况：

from matplotlib import pyplot as plt
# 检测数据的读取情况
plt.figure(figsize=(20, 6), dpi=80)
temp1 = trainingSet1['04015']['ECG0'][:5000]
plt.plot(temp1)
plt.show()

如图所示：