好的，根据您的要求，以下是一个处理PTB心电数据集的Python代码，使用CNN和RNN进行特征提取，并使用PyTorch实现：
```python
import numpy as np
import matplotlib.pyplot as plt
import wfdb
import pywt
import torch
import torch.nn as nn
import torch.optim as optim
from torch.utils.data import DataLoader, Dataset
# 数据预处理
def preprocess(data):
# 滤波器
b, a = signal.butter(4, [0.5, 40], 'bandpass')
# 去噪
data = signal.filtfilt(b, a, data)
# 小波变换
coeffs = pywt.wavedec(data, 'db4', level=5)
data = pywt.waverec(coeffs[:-1] + [None]*3, 'db4')
return data
# 加载PTB心电数据集
class PTBDataset(Dataset):
def __init__(self, path):
self.records = wfdb.get_record_list('ptbdb/' + path)
def __len__(self):
return len(self.records)
def __getitem__(self, idx):
record = self.records[idx]
data, _ = wfdb.rdsamp('ptbdb/' + record)
data = preprocess(data[:, 0])
data = torch.from_numpy(data).float()
return data.unsqueeze(0)
# CNN+RNN模型
class CNNRNN(nn.Module):
def __init__(self):
super(CNNRNN, self).__init__()
self.cnn = nn.Sequential(
nn.Conv1d(1, 64, kernel_size=5, stride=2, padding=2),
nn.BatchNorm1d(64),
nn.ReLU(),
nn.Conv1d(64, 128, kernel_size=5, stride=2, padding=2),
nn.BatchNorm1d(128),
nn.ReLU(),
nn.Conv1d(128, 256, kernel_size=5, stride=2, padding=2),
nn.BatchNorm1d(256),
nn.ReLU(),
)
self.rnn = nn.GRU(256, 128, bidirectional=True, batch_first=True)
self.fc = nn.Linear(256, 1)
self.sigmoid = nn.Sigmoid()
def forward(self, x):
x = self.cnn(x)
x = x.permute(0, 2, 1)
x, _ = self.rnn(x)
x = self.fc(x)
x = self.sigmoid(x)
return x
# 训练模型
def train_model(model, train_loader, optimizer, criterion, device):
model.train()
train_loss = 0
for data in train_loader:
data = data.to(device)
optimizer.zero_grad()
output = model(data)
loss = criterion(output.squeeze(), torch.ones_like(output))
loss.backward()
optimizer.step()
train_loss += loss.item() * data.size(0)
train_loss /= len(train_loader.dataset)
return train_loss
# 测试模型
def test_model(model, test_loader, criterion, device):
model.eval()
test_loss = 0
with torch.no_grad():
for data in test_loader:
data = data.to(device)
output = model(data)
loss = criterion(output.squeeze(), torch.ones_like(output))
test_loss += loss.item() * data.size(0)
test_loss /= len(test_loader.dataset)
return test_loss
# 主函数
if __name__ == '__main__':
# 设置超参数
batch_size = 64
learning_rate = 0.001
num_epochs = 10
# 加载数据集
train_dataset = PTBDataset('ptbdb_train')
test_dataset = PTBDataset('ptbdb_test')
train_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)
test_loader = DataLoader(test_dataset, batch_size=batch_size, shuffle=False)
# 初始化模型、优化器和损失函数
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
model = CNNRNN().to(device)
optimizer = optim.Adam(model.parameters(), lr=learning_rate)
criterion = nn.BCELoss()
# 训练模型
for epoch in range(num_epochs):
train_loss = train_model(model, train_loader, optimizer, criterion, device)
test_loss = test_model(model, test_loader, criterion, device)
print(f'Epoch {epoch+1}/{num_epochs}, Train Loss: {train_loss:.4f}, Test Loss: {test_loss:.4f}')
```
这段代码包含了以下步骤：
1. 数据预处理：使用带通滤波器、小波变换和去噪等方法对数据进行预处理。
2. 加载PTB心电数据集：使用wfdb库加载数据集。
3. CNN+RNN模型：使用CNN和RNN进行特征提取。
4. 训练模型：使用Adam优化器和BCELoss损失函数训练模型。
注意：这段代码仅供参考，实际应用中可能需要对数据预处理和模型结构进行调整。
最低0.47元/天 解锁文章
5855
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
