import torch
import torch.nn as nn
from torch.utils.data import DataLoader
from torchvision.datasets import MNIST
from torchvision import transforms
from torchvision.utils import save_image
# 定义模型
class AutoEncoder(nn.Module):
def __init__(self):
super(AutoEncoder, self).__init__()
self.encoder = nn.Sequential(
nn.Linear(28*28, 128),
nn.ReLU(),
nn.Linear(128, 64),
nn.ReLU(),
nn.Linear(64, 12),
nn.ReLU(),
nn.Linear(12, 3), # 假设我们将图像编码为3维向量
)
self.decoder = nn.Sequential(
nn.Linear(3, 12),
nn.ReLU(),
nn.Linear(12, 64),
nn.ReLU(),
nn.Linear(64, 128),
nn.ReLU(),
nn.Linear(128, 28*28),
nn.Sigmoid(), # 使用Sigmoid函数将输出映射到[0, 1]区间
)
def forward(self, x):
encoded = self.encoder(x)
decoded = self.decoder(encoded)
return encoded, decoded
# 准备数据
transform = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307,), (0.3081,)),
])
train_set = MNIST('data', train=True, download=True, transform=transform)
train_loader = DataLoader(train_set, shuffle=True, batch_size=128)
# 实例化模型、损失函数和优化器
model = AutoEncoder().to(device)
optimizer = torch.optim.Adam(model.parameters(), lr=0.001)
loss_func = nn.MSELoss()
# 训练模型
for epoch in range(10):
for batch_idx, (data, _) in enumerate(train_loader):
data = data.view(-1, 28*28).to(device)
optimizer.zero_grad()
encoded, decoded = model(data)
loss = loss_func(decoded, data)
loss.backward()
optimizer.step()
if batch_idx % 10 == 0:
print('Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format(
epoch, batch_idx * len(data), len(train_loader.dataset),
0. * batch_idx / len(train_loader), loss.item()))
# 保存编码后的特征
with torch.no_grad():
encoded_images = model.encoder(train_set[:3].view(-1, 28*28).to(device))
save_image(encoded_images.view(-1, 28, 28), 'encoded_images.png')
# 注意：这里的代码仅作为示例，实际应用中需要根据医学影像数据的特点进行调整这段代码展示了如何使用PyTorch来训练一个自编码器模型，它将医学影像数据（这里以MNIST手写数据作为例子）转换为一个低维空间中的向量表示，并尝试从这个向量重建原始图像。在实际应用中，影像数据需要进行预处理，包括图像尺寸调整、归一化等步骤，并且损失函数、优化器和模型架构都需要根据医学影像数