本章以BraTs数据集为例子,详细地讲解对于2D网络,医学三维数据且多模态多标签该如何预处理,并用代码实现
预处理方法各种各样,而我的步骤主要以下几步:
1、对各个模态进行标准化
2、对各模态及其GT数据进行裁剪
3、对各模态及其GT数据进行切片,并抛无病灶切片,最后合并各模态的切片,然后保存为Numpy
一、标准化多模态
brats中的四个序列是不同模态的图像,因此图像对比度也不一样,所以采用z-score方式来对每个模态图像进行标准化,即将每个模态的数据标准化为零均值和单位标准差 ,但是GT文件是不需要进行标准化的.函数实现代码如下
def normalize(slice, bottom=99, down=1):
"""
normalize image with mean and std for regionnonzero,and clip the value into range
:param slice:
:param bottom:
:param down:
:return:
"""
#有点像“去掉最低分去掉最高分”的意思,使得数据集更加“公平”
b = np.percentile(slice, bottom)
t = np.percentile(slice, down)
slice = np.clip(slice, t, b)#限定范围numpy.clip(a, a_min, a_max, out=None)
#除了黑色背景外的区域要进行标准化
image_nonzero = slice[np.nonzero(slice)]
if np.std(slice) == 0 or np.std(image_nonzero) == 0:
return slice
else:
tmp = (slice - np.mean(image_nonzero)) / np.std(image_nonzero)
# since the range of intensities is between 0 and 5000 ,
# the min in the normalized slice corresponds to 0 intensity in unnormalized slice
# the min is replaced with -9 just to keep track of 0 intensities
# so that we can discard those intensities afterwards when sampling random patches
tmp[tmp == tmp.min()] = -9 #黑色背景区域
return tmp
二、裁剪
裁剪这个步骤根据自己的数据集进行选择,对于BraTs数据集我觉得是有必要的,下图是一个序列MR图像,其中灰色部分为脑部区域，黑色则为背景，背景信息在整幅图像中的比例较大，而且背景对于分割没有任何帮助。但是要对每一个像素进行分类，图像中肿瘤区域占比非常小，因此会出现严重的数据不平衡。为了提升模型分割的性能，我觉得裁剪有必要。从医生角度来看这个ＭＲ图像，会自动过滤掉这个背景信息，把所有目光集中在脑部区域，因此去除脑部区域周围的背景信息是必要的
def crop_ceter(img,croph,cropw):
#for n_slice in range(img.shape[0]):
height,width = img[0].shape
starth = height//2-(croph//2)
startw = width//2-(cropw//2)
return img[:,starth:starth+croph,startw:startw+cropw]
三、切片、抛无病灶切片、合并各模态的切片、保存为Numpy
由于大多数医学图像都是三维数据,所以只有切成2D数据,才能适应2D网络,此外切片中不含有病灶部分的可以舍弃,同样也是为了缓解类别不均衡问题,又由于是多模态,因此要将各模态的切片组合成多通道,最后保存为npy, 而对于其相应的GT切片我是直接保存为npy
扫描二维码关注公众号，回复：
11227245 查看本文章
#切片处理,并去掉没有病灶的切片,合并多模态组合多通道
for n_slice in range(flair_crop.shape[0]):
if np.max(mask_crop[n_slice,:,:]) != 0:
maskImg = mask_crop[n_slice,:,:]
FourModelImageArray = np.zeros((flair_crop.shape[1],flair_crop.shape[2],4),np.float)
flairImg = flair_crop[n_slice,:,:]
flairImg = flairImg.astype(np.float)
FourModelImageArray[:,:,0] = flairImg
t1Img = t1_crop[n_slice,:,:]
t1Img = t1Img.astype(np.float)
FourModelImageArray[:,:,1] = t1Img
t1ceImg = t1ce_crop[n_slice,:,:]
t1ceImg = t1ceImg.astype(np.float)
FourModelImageArray[:,:,2] = t1ceImg
t2Img = t2_crop[n_slice,:,:]
t2Img = t2Img.astype(np.float)
FourModelImageArray[:,:,3] = t2Img
imagepath = outputImg_path + "\\" + str(pathlgg_list[subsetindex]) + "_" + str(n_slice) + ".npy"
maskpath = outputMask_path + "\\" + str(pathlgg_list[subsetindex]) + "_" + str(n_slice) + ".npy"
np.save(imagepath,FourModelImageArray)#(160,160,4) np.float dtype('float64')
np.save(maskpath,maskImg) # (160, 160) dtype('uint8') 值为0 1 2 4
以BraTs18数据集的预处理为例的完整代码
用 jupyter notebook 执行
https://download.csdn.net/download/weixin_40519315/12275459
Pytorch怎么读取上面处理好的Numpy文件
外层部分主要是这么干,train_test_split函数是将数据集分成训练集和验证集的
# Data loading code
img_paths = glob(r'D:\Project\CollegeDesign\dataset\Brats2018FoulModel2D\trainImage\*')
mask_paths = glob(r'D:\Project\CollegeDesign\dataset\Brats2018FoulModel2D\trainMask\*')
train_img_paths, val_img_paths, train_mask_paths, val_mask_paths = train_test_split(img_paths, mask_paths, test_size=0.2, random_state=41)
print("train_num:%s"%str(len(train_img_paths)))
print("val_num:%s"%str(len(val_img_paths)))
train_dataset = Dataset(args, train_img_paths, train_mask_paths, args.aug)
val_dataset = Dataset(args, val_img_paths, val_mask_paths)
train_loader = torch.utils.data.DataLoader(
train_dataset,
batch_size=args.batch_size,
shuffle=True,
pin_memory=True,
drop_last=True)
val_loader = torch.utils.data.DataLoader(
val_dataset,
batch_size=args.batch_size,
shuffle=False,
pin_memory=True,
drop_last=False)
...
for epoch in range(args.epochs):
# train for one epoch
train_log = train(args, train_loader, model, criterion, optimizer, epoch)
# evaluate on validation set
val_log = validate(args, val_loader, model, criterion)
...
由于用到了 Dataset ,需要人工去复现这个类中的 __getitem__函数,代码如下
def __getitem__(self, idx):
img_path = self.img_paths[idx]
mask_path = self.mask_paths[idx]
#读numpy数据(npy)的代码
npimage = np.load(img_path)
npmask = np.load(mask_path)
npimage = npimage.transpose((2, 0, 1))
WT_Label = npmask.copy()
WT_Label[npmask == 1] = 1.
WT_Label[npmask == 2] = 1.
WT_Label[npmask == 4] = 1.
TC_Label = npmask.copy()
TC_Label[npmask == 1] = 1.
TC_Label[npmask == 2] = 0.
TC_Label[npmask == 4] = 1.
ET_Label = npmask.copy()
ET_Label[npmask == 1] = 0.
ET_Label[npmask == 2] = 0.
ET_Label[npmask == 4] = 1.
nplabel = np.empty((160, 160, 3))
nplabel[:, :, 0] = WT_Label
nplabel[:, :, 1] = TC_Label
nplabel[:, :, 2] = ET_Label
nplabel = nplabel.transpose((2, 0, 1))
nplabel = nplabel.astype("float32")
npimage = npimage.astype("float32")
return npimage,nplabel
