dicom转png代码

import os
from glob import glob

import cv2
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import pydicom
from pydicom.pixel_data_handlers.util import apply_voi_lut
from tqdm.notebook import tqdm
import imageio
from PIL import Image

import warnings
warnings.filterwarnings("ignore")

def dicom2array(path, voi_lut=True, fix_monochrome=True):
    # Use the pydicom library to read the DICOM file
    dicom = pydicom.read_file(path)

    # VOI LUT (if available by DICOM device) is used to transform raw DICOM data to "human-friendly" view
    if voi_lut:
        data = apply_voi_lut(dicom.pixel_array, dicom)
    else:
        data = dicom.pixel_array

    # Depending on this value, X-ray may look inverted - fix that
    if fix_monochrome and dicom.PhotometricInterpretation == "MONOCHROME1":
        data = np.amax(data) - data

    # Normalize the image array
    data = data - np.min(data)
    data = data / np.max(data)
    data = (data * 255).astype(np.uint8)

    return data


def resize(array, size, keep_ratio=False, resample=Image.LANCZOS):
    image = Image.fromarray(array)

    if keep_ratio:
        image.thumbnail((size, size), resample)
    else:
        image = image.resize((size, size), resample)

    return image

image_ids = []
height = []
width = []

load_dir = '/home/lincong/桌面/yxq/kaggle_new/bad_img'
save_dir = '/home/lincong/桌面/yxq/kaggle_new/xml_img'

os.makedirs(save_dir, exist_ok=True)

image_paths = glob(os.path.join(load_dir, '*.dicom'))

for image_path in image_paths:
    basename = os.path.basename(image_path).split('.')[0]
    image = dicom2array(image_path)
    #image_r = resize(image, size=SIZE, keep_ratio=True)
    imageio.imwrite(os.path.join(save_dir, basename + ".png"), image)