影像组学图像预处理

针对一个大文件夹以及子文件内所有的dicom格式图像

1.重采样，目的使空间分辨率一致

2.归一化（最大-最小值缩放）：(x-min(x)) / (max(x)-min(x))

目的消除图像亮度差异，使像素值缩放到[0, 1]之间

3.Z-score标准化：(x-x(averge))/标准差

目的使图像具有统一的特征

import os
import SimpleITK as sitk
import numpy as np

 # 进行归一化和Z-score标准化处理
def normalize_and_standardize(image_array):
    # 归一化
    min_val = np.min(image_array)
    max_val = np.max(image_array)
    if min_val == max_val:# 以防出现归一化时分母为0
        normalized_array = np.zeros_like(image_array)# 如果相等，则将所有值设为0
    else:
        normalized_array = (image_array - min_val) / (max_val - min_val)
    # Z-score
    mean = np.mean(normalized_array)
    std = np.std(normalized_array)
    if std == 0:
        standardized_array = np.zeros_like(normalized_array)# 避免标准差为0而报错
    else:
        standardized_array = (normalized_array - mean) / std
    return standardized_array

CT_path = r'C:\Users\lyp\Desktop\CT03802522'
file_list = []
for root, dirs, files in os.walk(CT_path):
    for file in files:
        if file.endswith('dcm'):
            file_list.append(os.path.join(root, file))

for image_path in file_list:
    image = sitk.ReadImage(image_path)
    
    new_spacing = [1, 1, 1]# 定义新的空间分辨率
    original_size = image.GetSize()#获取原始图像像素尺寸和空间分辨率，均为一个三元素元组（）
    original_spacing = image.GetSpacing()
    # 计算新的像素尺寸
    new_size = [int(np.round(original_size[0] * original_spacing[0] / new_spacing[0])),
                int(np.round(original_size[1] * original_spacing[1] / new_spacing[1])),
                int(np.round(original_size[2] * original_spacing[2] / new_spacing[2]))
               ]# 分别计算X，Y，Z轴大小
    
    resample = sitk.ResampleImageFilter()
    resample.SetOutputSpacing(new_spacing)
    resample.SetSize(new_size)
    resample.SetOutputDirection(image.GetDirection())
    resample.SetOutputOrigin(image.GetOrigin())
    resample.SetTransform(sitk.Transform())
    resample.SetDefaultPixelValue(image.GetPixelIDValue())# 以上四个操作均采用保留原始图像属性
    resample.SetInterpolator(sitk.sitkBSpline)# 插值方法，这种为双线性插值，大多数情况下使用，或者线性插值（sitk.sitkLinear）
    
    resample_image = resample.Execute(image)
    
    resample_array = sitk.GetArrayFromImage(resample_image)
    standardized_array = normalize_and_standardize(resample_array)
    
    processed_image = sitk.GetImageFromArray(standardized_array)# 倒回来，从array得到图像
    processed_image.SetSpacing(new_spacing)
    processed_image.SetOrigin(resample_image.GetOrigin())
    processed_image.SetDirection(resample_image.GetDirection())# 确保处理后的图像是正确的
    
    # 将图像强制转换为指定的像素类型，防止报错
    processed_image = sitk.Cast(processed_image, sitk.sitkInt16)
    save_file = image_path.replace('.dcm', '_processed.dcm')
    
    sitk.WriteImage(processed_image, save_file)