DeepLesion / scripts /DL_save_nifti.py
farrell236's picture
log broken pngs
70dcf29
"""
A simple demo to load 2D 16-bit slices from DeepLesion and save to 3D nifti volumes.
The nifti volumes can be viewed in software such as 3D slicer and ITK-SNAP.
"""
import os
import cv2
import numpy as np
import pandas as pd
import SimpleITK as sitk
from tqdm import tqdm
dir_in = '../Images_png'
dir_out = '../Images_nifti'
info_fn = '../DL_info.csv'
def slices2nifti(ims, fn_out, spacing):
"""save 2D slices to 3D nifti file considering the spacing"""
image_itk = sitk.GetImageFromArray(np.stack(ims, axis=0))
image_itk.SetSpacing(spacing)
image_itk.SetDirection((1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, -1.0))
sitk.WriteImage(image_itk, os.path.join(dir_out, fn_out))
# print(f'{fn_out} saved')
def load_slices(dir, slice_idxs):
"""load slices from 16-bit png files"""
ims = []
for slice_idx in slice_idxs:
path = os.path.join(dir_in, dir, f'{slice_idx:03d}.png')
im = cv2.imread(path, cv2.IMREAD_UNCHANGED) # Read as 16-bit image
assert im is not None, f'Error reading: {path}'
# print(f'read {path}')
# the 16-bit png file has an intensity bias of 32768
ims.append((im.astype(np.int32) - 32768).astype(np.int16))
return ims
if __name__ == '__main__':
# Read spacings and image indices in DeepLesion
dl_info = pd.read_csv(info_fn)
spacings = dl_info['Spacing_mm_px_'].str.split(',', expand=True).astype(float).values
ranges = dl_info['Slice_range'].str.split(',', expand=True).astype(int).values
img_dirs = dl_info['File_name'].str.rsplit('_', n=1).str[0].values
if not os.path.exists(dir_out):
os.mkdir(dir_out)
for idx, (img_dir, range, spacing) in tqdm(enumerate(zip(img_dirs, ranges, spacings)), total=len(img_dirs)):
# Broken: ./Images_png/001821_07_01/372.png
# Broken: ./Images_png/002161_04_02/116.png
ims = load_slices(img_dir, np.arange(*range))
fn_out = f'{img_dir}_{range[0]:03d}-{range[-1]:03d}.nii.gz'
slices2nifti(ims, fn_out, spacing)