如何使用python脚本批量对png格式图片进行肺部CT图像的肺实质分割？

    pip install numpy opencv-python scikit-image matplotlib tensorflow
    
    
      
    

    import os
    import cv2
    import numpy as np
    import matplotlib.pyplot as plt
    from skimage import measure, filters
    from tensorflow.keras.models import load_model
    from skimage.morphology import closing, square
    
    # 定义一个函数，用于加载模型（如果使用深度学习模型进行分割）
    def load_segmentation_model(model_path):
    return load_model(model_path)
    
    # 定义一个函数进行肺部CT图像分割
    def lung_segmentation(image, model=None, save_intermediate_steps=False):
    # 1. 转为灰度图
    gray_image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    
    # 2. 预处理：去噪 (中值滤波)
    preprocessed_image = cv2.medianBlur(gray_image, 5)
    
    # 3. 阈值化 (Otsu's方法)
    _, thresholded = cv2.threshold(preprocessed_image, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)
    
    # 4. 开闭运算：去除小噪声
    closed_image = closing(thresholded, square(3))
    
    if save_intermediate_steps:
        # 保存预处理图像
        cv2.imwrite("step_preprocessed.png", preprocessed_image)
        cv2.imwrite("step_thresholded.png", thresholded)
        cv2.imwrite("step_closed.png", closed_image)
    
    # 5. 连通域分析：标记肺区域
    labeled_image = measure.label(closed_image)
    
    # 6. 根据区域大小过滤噪声（可选）
    regions = measure.regionprops(labeled_image)
    for region in regions:
        if region.area < 500:  # 小于500像素的区域认为是噪声
            for coordinates in region.coords:
                labeled_image[coordinates[0], coordinates[1]] = 0
    
    # 7. 显示分割结果
    if save_intermediate_steps:
        plt.imshow(labeled_image, cmap='nipy_spectral')
        plt.title('Lung Segmentation Result')
        plt.axis('off')
        plt.savefig('step_final_segmentation.png')
    
    return labeled_image
    
    # 批量处理文件夹中的 PNG 图像
    def batch_process_images(input_folder, output_folder, model=None, save_steps=False):
    if not os.path.exists(output_folder):
        os.makedirs(output_folder)
    
    # 遍历文件夹中的所有 PNG 图像
    for filename in os.listdir(input_folder):
        if filename.endswith(".png"):
            # 读取图像
            img_path = os.path.join(input_folder, filename)
            image = cv2.imread(img_path)
    
            # 分割肺部CT图像
            segmented_image = lung_segmentation(image, model=model, save_intermediate_steps=save_steps)
    
            # 保存分割结果
            output_path = os.path.join(output_folder, f"segmented_{filename}")
            cv2.imwrite(output_path, segmented_image)
    
            print(f"Processed {filename} and saved to {output_path}")
    
    # 使用深度学习模型进行肺实质分割（如有需要，使用模型）
    def deep_learning_segmentation(image, model):
    # 如果你有训练好的深度学习模型，可以在这里调用它进行分割
    # image = np.expand_dims(image, axis=0)  # 根据需要调整图像形状
    # prediction = model.predict(image)
    # return prediction
    pass
    
    if __name__ == "__main__":
    # 输入文件夹（包含 PNG 图像）
    input_folder = 'path_to_input_images'
    # 输出文件夹（保存分割结果）
    output_folder = 'path_to_output_images'
    
    # 如果你有训练好的模型，加载它
    # model = load_segmentation_model('path_to_trained_model.h5')
    
    # 批量处理图像
    batch_process_images(input_folder, output_folder, model=None, save_steps=True)