请将LiTS2017数据集的图像和标签处理成2D的png格式，并且标签需要拆分成灰度图，背景灰度值0，肝脏灰度值1，肝脏肿瘤灰度值2，给出完整代码，我的LiTS2017数据集路径为C:\Users\Administrator\Desktop\LiTS2017，里面包含Training Batch 1，里面有图像，图像拓展名.nii。和Training Batch 2，里面有标签,标签拓展名.nii，保存路径为C:\Users\Administrator\Desktop\2D-LiTS2017，里面有image和mask两个文件夹，请将转换好的保存到对应的文件夹

下面是将LiTS2017数据集的图像和标签处理成2D的png格式，并且标签拆分成灰度图的完整代码：

import os
import numpy as np
import nibabel as nib
from PIL import Image

data_path = 'C:/Users/Administrator/Desktop/LiTS2017'
save_path = 'C:/Users/Administrator/Desktop/2D-LiTS2017'

if not os.path.exists(save_path):
    os.makedirs(os.path.join(save_path, 'image'))
    os.makedirs(os.path.join(save_path, 'mask'))

def normalize(volume):
    """Normalize the volume"""
    min_val = volume.min()
    max_val = volume.max()
    volume = (volume - min_val) / (max_val - min_val)
    return volume

def save_image(image, save_path, name):
    """Save a single image"""
    image = (image * 255).astype(np.uint8)
    Image.fromarray(image).save(os.path.join(save_path, name))

def process_data(data_path, save_path):
    """Process the LiTS2017 dataset"""
    image_path = os.path.join(data_path, 'Training Batch 1')
    mask_path = os.path.join(data_path, 'Training Batch 2')

    for i in range(131):
        # Load image
        image_file = os.path.join(image_path, f'volume-{i:03d}.nii')
        image = nib.load(image_file).get_fdata()
        image = normalize(image)

        # Save image slices
        for j in range(image.shape[2]):
            image_slice = image[:, :, j]
            save_image(image_slice, os.path.join(save_path, 'image'), f'image-{i:03d}-{j:03d}.png')

        # Load mask
        mask_file = os.path.join(mask_path, f'segment_liver_mask-{i:03d}.nii')
        mask = nib.load(mask_file).get_fdata()

        # Split mask into liver and tumor regions
        liver_mask = (mask == 1).astype(np.uint8)
        tumor_mask = (mask == 2).astype(np.uint8)

        # Save mask slices
        for j in range(mask.shape[2]):
            liver_slice = liver_mask[:, :, j]
            tumor_slice = tumor_mask[:, :, j]
            mask_slice = np.zeros((liver_slice.shape[0], liver_slice.shape[1], 3), dtype=np.uint8)
            mask_slice[:, :, 0] = 255 - liver_slice * 255
            mask_slice[:, :, 1] = 255 - tumor_slice * 255
            save_image(mask_slice, os.path.join(save_path, 'mask'), f'mask-{i:03d}-{j:03d}.png')

process_data(data_path, save_path)

这段代码中，我们首先创建了保存处理后数据的文件夹，然后定义了一个函数`normalize`，用于将体积数据归一化到0-1之间；另外定义了一个函数`save_image`，用于将单张图像保存为png格式。接下来，我们遍历每个体积数据，加载图像和标签，并将它们分别保存为png格式的2D图像。在保存标签时，我们将其拆分成灰度图，背景灰度值为0，肝脏灰度值为1，肝脏肿瘤灰度值为2，并将它们保存为RGB格式的图像，以便于后续可视化。