计算PET脑nii图像分割白质和灰质的Dice系数代码

以下是一个Python代码示例，用于计算PET脑nii图像分割白质和灰质的Dice系数。

import numpy as np
import nibabel as nib

def dice_coef(y_true, y_pred, smooth=1):
    intersection = np.sum(np.abs(y_true * y_pred))
    return (2. * intersection + smooth) / (np.sum(np.square(y_true)) + np.sum(np.square(y_pred)) + smooth)

# Load the PET brain nii image and the ground truth segmentation mask
pet_nii = nib.load('pet_brain.nii')
seg_nii = nib.load('seg_mask.nii')

# Get the image data and the segmentation mask data as numpy arrays
pet_data = pet_nii.get_fdata()
seg_data = seg_nii.get_fdata()

# Normalize the PET data to have values between 0 and 1
pet_data = (pet_data - np.min(pet_data)) / (np.max(pet_data) - np.min(pet_data))

# Threshold the segmentation mask data to get binary masks for white matter and gray matter
white_matter_seg = np.where(seg_data == 1, 1, 0)
gray_matter_seg = np.where(seg_data == 2, 1, 0)

# Threshold the PET data to get binary masks for white matter and gray matter
white_matter_pet = np.where(pet_data >= 0.5, 1, 0)
gray_matter_pet = np.where(pet_data < 0.5, 1, 0)

# Calculate the Dice coefficients for white matter and gray matter segmentation
white_matter_dice = dice_coef(white_matter_seg, white_matter_pet)
gray_matter_dice = dice_coef(gray_matter_seg, gray_matter_pet)

# Print the results
print('White Matter Dice Coefficient:', white_matter_dice)
print('Gray Matter Dice Coefficient:', gray_matter_dice)

该代码首先加载PET脑nii图像和地面真实分割掩码，然后将它们转换为numpy数组。然后，PET数据被归一化到0到1之间的值，并使用阈值将分割掩码数据转换为二进制掩码，以获取白质和灰质掩码。接下来，使用定义的Dice系数函数计算白质和灰质的分割Dice系数。最后，代码将打印结果。