nnUNet复现遇到Error: Spacing mismatch between segmentation and corresponding images.（LiTS数据集）

数据集：公开数据集LiTS

背景：已经完成了nnUNet的数据转换的部分（即Task040_Liver已经转换成Dataset040_Liver）

下一步执行数据预处理操作的时候，出现了错误Error: Spacing mismatch between segmentation and corresponding images.

数据预处理：nnUNetv2_plan_and_preprocess -d 040 --verify_dataset_integrity

原因：分割标签和原始图像的空间分辨率（体素间距）不一致
解决：因此我们要将label和原始图像的空间分辨率（体素间距）统一

from nibabel.processing import resample_from_to
import nibabel as nib
import os

# imgs_path是Dataset040_Liver的imagesTr
# lab_path是Dataset040_Liver的labelsTr
# save_path是自己设置的。统一体素间距后label保存的路径，可以自定义，后面取代Dataset040_Liver的labelsTr

imgs_path = r'F:\nnUNet-master\nnUNet-master\nnUNetFrame\nnUNet_raw\Dataset040_Liver\imagesTr'
lab_path = r'F:\nnUNet-master\nnUNet-master\nnUNetFrame\nnUNet_raw\Dataset040_Liver\labelsTr'
save_path = r'F:\nnUNet-master\nnUNet-master\nnUNetFrame\nnUNet_raw\Dataset040_Liver\labelsTr\label'

for image in os.listdir(imgs_path):
    img_path = imgs_path + '/' + image   # image:Liver_000_0000.nii.gz
    original_img = nib.load(img_path)    # 原始图像的路径
    base_name = image.split("_")[:-1]
    label_name = "_".join(base_name) + ".nii.gz"   # Liver_000.nii.gz
    label_path = lab_path + '/' + label_name  # 原始label的路径
    segmentation_img = nib.load(label_path)
    target_affine = original_img.affine
    target_shape = original_img.shape
    resampled_seg = resample_from_to(segmentation_img, (target_shape, target_affine), order=0)
    save_path_ = save_path + '/' + label_name
    nib.save(resampled_seg, save_path_)
    print("Finish:", save_path_)

PS: 看注释