多张png合成为nii.gz

最新推荐文章于 2024-04-19 23:53:45 发布

litchi&&mango

最新推荐文章于 2024-04-19 23:53:45 发布

阅读量1.9k

点赞数 2

CC 4.0 BY-SA版权

分类专栏： pytorch 文章标签： pytorch 深度学习 python

本文链接：https://blog.youkuaiyun.com/qq_45862064/article/details/119935476

pytorch 专栏收录该内容

6 篇文章

订阅专栏

这段代码展示了如何使用Python将多张PNG图像按(h,w,c)维度组合成NII.gz文件。首先，定义了一个DataSet类，继承自torch.utils.data.Dataset，用于加载和转换图像。接着，遍历所有PNG图像，将其转换为Tensor并堆叠成一个3维数组。最后，使用nibabel库保存为Nifti1Image格式的nii.gz文件。该过程适用于灰度图像，对于RGB图像需要先将其转换为单通道。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

将多张png图片按照(h, w, c)组装成nii.gz

import numpy as np
import nibabel as nib
from glob import glob
from torch.utils.data import Dataset
from torchvision import transforms


class DataSet(Dataset):  # 方便读数据
    def __init__(self, label_path):
        self.label_path = label_path
        self.transform = transforms.Compose([transforms.ToTensor()],)

    def __len__(self):
        return len(self.label_path)

    def __getitem__(self, idx):
        label = self.label_path[idx]
        label = Image.open(label)
        label = self.transform(label)
        return label


label_path = './label_test/1/*'
label_path = glob(label_path)
dataset = DataSet(label_path)
label = dataset[0]
z, x, y = label.shape
z = len(dataset)  # 通道数
allImage = np.zeros([x, y, z], dtype='uint8')
# print(x, y, z)
for i in range(z):
    label = dataset[i]
    singleImage = label.numpy()
    allImage[:, :, i] = singleImage
# print(allImage.shape)
new_image = nib.Nifti1Image(allImage, np.eye(4))
nib.save(new_image, 'data2.nii.gz')

（适用于灰度图像，RGB图像需转为单通道）

image = Image.open(singleImage).convert('L')

参考：
1.https://blog.youkuaiyun.com/weixin_40519315/article/details/106022220?spm=1001.2014.3001.5501

2.https://blog.youkuaiyun.com/u010882121/article/details/94336919?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522162996141016780262532497%2522%252C%2522scm%2522%253A%252220140713.130102334.pc%255Fall.%2522%257D&request_id=162996141016780262532497&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2_allfirst_rank_ecpm_v1~rank_v29_ecpm-1-94336919.first_rank_v2_pc_rank_v29&utm_term=png%E5%90%88%E6%88%90%E4%B8%BAnii&spm=1018.2226.3001.4187