# from concurrent.futures import ThreadPoolExecutor # import sys # import os # import numpy as np # import torch # from torch.utils.data.dataset import Dataset # import glob # import cv2 # import torchio as tio # import matplotlib.pyplot as plt # from torch.utils.data import DataLoader # from sdf import compute_sdf # PREPROCESSING_TRANSORMS_CT = tio.Compose([ # tio.Clamp(out_min=-250, out_max=800), # tio.RescaleIntensity(in_min_max=(-250, 800), # out_min_max=(-1.0, 1.0)), # tio.CropOrPad(target_shape=(64, 64, 64)) # ]) # PREPROCESSING_TRANSORMS_MRI = tio.Compose([ # tio.Clamp(out_min=0, out_max=1000), # tio.RescaleIntensity(in_min_max=(0, 1000), # out_min_max=(-1.0, 1.0)), # tio.CropOrPad(target_shape=(64, 64, 64)) # ]) # PREPROCESSING_MASK_TRANSORMS = tio.Compose([ # tio.CropOrPad(target_shape=(64, 64, 64)) # ]) # class MMWHS_Dataset(Dataset): # def __init__(self, root_dir='', data_type='', mode = ''): # self.root_dir = root_dir # self.data_type = data_type # self.mode = mode # self.file_names = self.get_file_names() # self.preprocessing_img_ct = PREPROCESSING_TRANSORMS_CT # self.preprocessing_img_mri = PREPROCESSING_TRANSORMS_MRI # self.preprocessing_mask = PREPROCESSING_MASK_TRANSORMS # @staticmethod # def create_mask(shape): # return torch.zeros(shape, dtype=torch.uint8) # @staticmethod # def project_to_2d(mask): # projection = torch.max(mask, dim=0)[0] # return projection.numpy() # @staticmethod # def min_enclosing_circle(projection): # points = np.column_stack(np.where(projection > 0)) # points = points.astype(np.float32) # print(points.shape) # (x, y), radius = cv2.minEnclosingCircle(points.astype(np.float32)) # center = (int(x), int(y)) # radius = int(radius) # return center, radius # @staticmethod # def create_circle_mask_2d(shape, center, radius): # mask = np.zeros(shape, dtype=np.uint8) # cv2.circle(mask, center, radius, 1, thickness=-1) # return mask # @staticmethod # def apply_circle_mask_to_3d(mask, circle_mask_2d): # for i in range(mask.shape[0]): # mask[i] = torch.from_numpy(circle_mask_2d) # return mask # def train_transform(self, image, label, p): # TRAIN_TRANSFORMS = tio.Compose([ # tio.RandomFlip(axes=(1), flip_probability=p), # ]) # image = TRAIN_TRANSFORMS(image) # label = TRAIN_TRANSFORMS(label) # return image, label # def get_file_names(self): # all_img_names = glob.glob(os.path.join(self.root_dir, './*image.nii.gz'), recursive=True) # return all_img_names # def __len__(self): # return len(self.file_names) # def __getitem__(self, index): # img_path = self.file_names[index] # mask_path = img_path.replace("image.nii.gz", "label.nii.gz") # sdf_path = img_path.split() # img = tio.ScalarImage(img_path) # mask = tio.LabelMap(mask_path) # name = img_path.split('/')[-1] # name = name.split('.nii')[0] # if self.data_type.upper() == 'CT': # img = self.preprocessing_img_ct(img) # elif self.data_type.upper() == 'MRI': # img = self.preprocessing_img_mri(img) # else: # raise ValueError("Wrong Data Type!") # mask = self.preprocessing_mask(mask) # # p = np.random.choice([0, 1]) # # if self.mode == 'train': # # img, mask = self.train_transform(img, mask, p) # affine = img.affine # mask = mask.data # img = img.data # label_1 = (mask == 1).float() # label_1_sdf = compute_sdf(label_1) # label_2 = (mask == 2).float() # label_2_sdf = compute_sdf(label_2) # label_3 = (mask == 3).float() # label_3_sdf = compute_sdf(label_3) # label_4 = (mask == 4).float() # label_4_sdf = compute_sdf(label_4) # label_5 = (mask == 5).float() # label_5_sdf = compute_sdf(label_5) # label = torch.cat((label_1, label_2, label_3, label_4, label_5), dim=0) # label_sdf = torch.cat((torch.tensor(label_1_sdf), torch.tensor(label_2_sdf), torch.tensor(label_3_sdf), torch.tensor(label_4_sdf), torch.tensor(label_5_sdf)), dim=0) # sdf_to_save = tio.ScalarImage(tensor=label_sdf, channels_last=False, affine=affine) # os.makedirs(self.root_dir, exist_ok=True) # sdf_to_save.save(os.path.join(self.root_dir, f"{name.replace('image', 'sdf')}.nii.gz")) # return { # 'name': name, # 'img': img, # 'mask_sdf': label_sdf, # 'mask': label, # 'affine': affine # } # def get_MMWHS_dataloader(root_dir, data_type, mode, batch_size=1, drop_last=False): # dataset = MMWHS_Dataset(root_dir=root_dir, data_type=data_type, mode=mode) # if mode == 'train': # shuffle = True # return dataset # elif mode == 'test': # shuffle = False # else: # raise ValueError('No Such Mode') # loader = DataLoader( # dataset, batch_size=batch_size, shuffle=shuffle, num_workers=20, pin_memory=True, drop_last=drop_last # ) # return loader # def test_dataset(): # dl = get_MMWHS_dataloader(root_dir='/mnt/data2/data/store/MSN/data_v2/MMWHS-CT/MMWHS-CT-resize-try-tiny-train', data_type='CT', mode='test') # for batch in iter(dl): # print(batch['name']) # print(len(dl)) # if __name__ == '__main__': # test_dataset() from concurrent.futures import ThreadPoolExecutor import sys import os import numpy as np import torch from torch.utils.data.dataset import Dataset import glob import cv2 import torchio as tio import matplotlib.pyplot as plt from torch.utils.data import DataLoader from sdf import compute_sdf PREPROCESSING_TRANSORMS_CT = tio.Compose([ tio.Clamp(out_min=-250, out_max=800), tio.RescaleIntensity(in_min_max=(-250, 800), out_min_max=(-1.0, 1.0)), tio.CropOrPad(target_shape=(64, 64, 64)) ]) PREPROCESSING_TRANSORMS_MRI = tio.Compose([ tio.Clamp(out_min=0, out_max=1800), tio.RescaleIntensity(in_min_max=(0, 1800), out_min_max=(-1.0, 1.0)), tio.CropOrPad(target_shape=(64, 64, 64)) ]) PREPROCESSING_MASK_TRANSORMS = tio.Compose([ tio.CropOrPad(target_shape=(64, 64, 64)) ]) class MMWHS_Dataset(Dataset): def __init__(self, root_dir='', data_type='', mode = ''): self.root_dir = root_dir self.data_type = data_type self.mode = mode self.file_names = self.get_file_names() self.preprocessing_img_ct = PREPROCESSING_TRANSORMS_CT self.preprocessing_img_mri = PREPROCESSING_TRANSORMS_MRI self.preprocessing_mask = PREPROCESSING_MASK_TRANSORMS @staticmethod def create_mask(shape): return torch.zeros(shape, dtype=torch.uint8) @staticmethod def project_to_2d(mask): projection = torch.max(mask, dim=0)[0] return projection.numpy() @staticmethod def min_enclosing_circle(projection): points = np.column_stack(np.where(projection > 0)) points = points.astype(np.float32) print(points.shape) (x, y), radius = cv2.minEnclosingCircle(points.astype(np.float32)) center = (int(x), int(y)) radius = int(radius) return center, radius @staticmethod def create_circle_mask_2d(shape, center, radius): mask = np.zeros(shape, dtype=np.uint8) cv2.circle(mask, center, radius, 1, thickness=-1) return mask @staticmethod def apply_circle_mask_to_3d(mask, circle_mask_2d): for i in range(mask.shape[0]): mask[i] = torch.from_numpy(circle_mask_2d) return mask def train_transform(self, image, label, sdf, p): TRAIN_TRANSFORMS = tio.Compose([ tio.RandomFlip(axes=(1), flip_probability=p), ]) image = TRAIN_TRANSFORMS(image) label = TRAIN_TRANSFORMS(label) sdf = TRAIN_TRANSFORMS(sdf) return image, label, sdf def get_file_names(self): all_img_names = glob.glob(os.path.join(self.root_dir, './*image.nii.gz'), recursive=True) return all_img_names def __len__(self): return len(self.file_names) def __getitem__(self, index): img_path = self.file_names[index] mask_path = img_path.replace("image.nii.gz", "label.nii.gz") sdf_path = img_path.replace("image.nii.gz", "sdf.nii.gz") img = tio.ScalarImage(img_path) mask = tio.LabelMap(mask_path) sdf = tio.ScalarImage(sdf_path) name = img_path.split('/')[-1] name = name.split('.nii')[0] if self.data_type.upper() == 'CT': img = self.preprocessing_img_ct(img) elif self.data_type.upper() == 'MRI': img = self.preprocessing_img_mri(img) else: raise ValueError("Wrong Data Type!") mask = self.preprocessing_mask(mask) p = np.random.choice([0, 1]) if self.mode == 'train': img, mask, sdf = self.train_transform(img, mask, sdf, p) affine = img.affine mask = mask.data img = img.data sdf = sdf.data label_1 = (mask == 1).float() # label_1_sdf = compute_sdf(label_1) label_2 = (mask == 2).float() # label_2_sdf = compute_sdf(label_2) label_3 = (mask == 3).float() # label_3_sdf = compute_sdf(label_3) label_4 = (mask == 4).float() # label_4_sdf = compute_sdf(label_4) label_5 = (mask == 5).float() # label_5_sdf = compute_sdf(label_5) label = torch.cat((label_1, label_2, label_3, label_4, label_5), dim=0) # label_sdf = torch.cat((torch.tensor(label_1_sdf), torch.tensor(label_2_sdf), torch.tensor(label_3_sdf), torch.tensor(label_4_sdf), torch.tensor(label_5_sdf)), dim=0) label_sdf = sdf # sdf_to_save = tio.LabelMap(tensor=label_sdf, channels_last=False, affine=affine) # os.makedirs(self.root_dir, exist_ok=True) # sdf_to_save.save(os.path.join(self.root_dir, f"{name.replace('image', 'sdf')}.nii.gz")) return { 'name': name, 'img': img, 'mask_sdf': label_sdf, 'mask': label, 'affine': affine } def get_MMWHS_dataloader(root_dir, data_type, mode, batch_size=1, drop_last=False): dataset = MMWHS_Dataset(root_dir=root_dir, data_type=data_type, mode=mode) if mode == 'train': shuffle = True return dataset elif mode == 'test': shuffle = False else: raise ValueError('No Such Mode') loader = DataLoader( dataset, batch_size=batch_size, shuffle=shuffle, num_workers=20, pin_memory=True, drop_last=drop_last ) return loader # def test_dataset(): # dl = get_MMWHS_dataloader(root_dir='/mnt/data2/data/store/MSN/data/MMWHS-MRI/mr_train_clean_resize_to_64_train_10percent', data_type='MRI', mode='test') # for batch in iter(dl): # print(batch['name']) # print(len(dl)) # if __name__ == '__main__': # test_dataset()