【影像组学pyradiomics学习笔记】png图像提取组学特征

1、提取单张png图像组学特征示例：

import SimpleITK as sitk
import numpy as np
import matplotlib.pyplot as plt
from radiomics import featureextractor
import os
import cv2


def load_image(image_path):
    image = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)  # Load image in grayscale
    return sitk.GetImageFromArray(image)


if __name__ == '__main__':

    # 指定image和mask路径
    image_path = 'xxx_img.jpg'
    mask_path = 'xxx_mask.png'

    # 加载图像
    image = load_image(image_path)
    mask = load_image(mask_path)

    # 获取掩码
    mask = sitk.BinaryThreshold(mask, 1, 255, 1, 0)

    # 可视化
    plt.subplot(1, 2, 1)
    plt.imshow(sitk.GetArrayFromImage(image), cmap='gray')
    plt.title('Image')
    plt.subplot(1, 2, 2)
    plt.imshow(sitk.GetArrayFromImage(mask), cmap='gray')
    plt.title('Mask')
    plt.show()

    # 初始化特征提取器
    extractor = featureextractor.RadiomicsFeatureExtractor()

    # 提取特征
    features = extractor.execute(image, mask)

    # 打印特征
    for feature_name, feature_value in features.items():
        print(f'{feature_name}: {feature_value}')

2、遍历文件夹提取多张png图像组学特征示例（数据需要自行准备，一张原图对应一张掩膜图，原图命名格式为*_image.png，掩膜图命名格式为*_label.png，将图像数据放在一个文件夹内即可）：

import SimpleITK as sitk
import numpy as np
import matplotlib.pyplot as plt
from radiomics import featureextractor
import os
import cv2
import pandas as pd

def load_image(image_path):
    image = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)  # 加载灰度图像
    return sitk.GetImageFromArray(image)

# 遍历文件夹内的所有文件
#params = 'Params.yaml'
folder_path = 'data/all'
all_features = []
image_names = []

for root, dirs, files in os.walk(folder_path):
    for file in files:
        if file.endswith('_image.png'):
            image_path = os.path.join(root, file)
            mask_path = os.path.join(root, file.replace('_image.png', '_label.png'))

            # 加载图像和掩码
            image = load_image(image_path)
            mask = load_image(mask_path)

            # 获取掩码
            mask = sitk.BinaryThreshold(mask, 1, 255, 1, 0)

            # 初始化特征提取器
            extractor = featureextractor.RadiomicsFeatureExtractor()

            # 提取特征
            extractor.enableAllFeatures()
            extractor.enableAllImageTypes()
            extractor.enabledFeatures['glcm'] = ['Autocorrelation',
                                                 'JointAverage',
                                                 'ClusterProminence',
                                                 'ClusterShade',
                                                 'ClusterTendency',
                                                 'Contrast',
                                                 'Correlation',
                                                 'DifferenceAverage',
                                                 'DifferenceEntropy',
                                                 'DifferenceVariance',
                                                 'JointEnergy',
                                                 'JointEntropy',
                                                 'Imc1',
                                                 'Imc2',
                                                 'Idm',
                                                 'MCC',
                                                 'Idmn',
                                                 'Id',
                                                 'Idn',
                                                 'InverseVariance',
                                                 'MaximumProbability',
                                                 'SumEntropy',
                                                 'SumSquares']

            # 由于图像是2D，所以专门去除了GLCM中的3D特征

            features = extractor.execute(image, mask)
            features_dict = dict(features.items())  # 将特征转换为字典形式
            all_features.append(features_dict)  # 将字典添加到列表中
            image_names.append(file.replace('_image.png', ''))  # 记录图像名称

# 将所有特征合并到一个 DataFrame 中
df = pd.DataFrame.from_dict(all_features)  # 使用 from_dict 方法并指定 orient='index' 来实现转置

# 添加图像名称作为索引
df['ImageName'] = image_names
df.set_index('ImageName', inplace=True)


# 保存为 Excel 表格
df.to_excel('result.xlsx', index=False, header=True)  # 保留表头特征名称和索引

生成的特征数据为：