File size: 5,231 Bytes
d60982d |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 |
import os
import torch
import torch.utils.data as data
import numpy as np
from PIL import Image, ImageFile
import random
from torchvision.transforms import ToTensor
from torchvision import transforms
import cv2
import pickle
import torch.nn.functional as F
ImageFile.LOAD_TRUNCATED_IMAGES = True
def collate_features(batch):
img = torch.cat([item[0] for item in batch], dim = 0)
coords = np.vstack([item[1] for item in batch])
return [img, coords]
def eval_transforms(pretrained=False):
if pretrained:
mean = (0.485, 0.456, 0.406)
std = (0.229, 0.224, 0.225)
else:
mean = (0.5,0.5,0.5)
std = (0.5,0.5,0.5)
trnsfrms_val = transforms.Compose(
[
transforms.Resize(224),
transforms.ToTensor(),
transforms.Normalize(mean = mean, std = std)
]
)
return trnsfrms_val
class GraphDataset(data.Dataset):
"""input and label image dataset"""
def __init__(self, root, ids, metadata_path, target_patch_size=-1):
super(GraphDataset, self).__init__()
"""
Args:
fileDir(string): directory with all the input images.
transform(callable, optional): Optional transform to be applied on a sample
"""
self.root = root
self.ids = ids
#self.target_patch_size = target_patch_size
self.classdict = pickle.load(open(os.path.join(metadata_path, 'label_map.pkl'), 'rb' )) # {'normal': 0, 'luad': 1, 'lscc': 2} #
#self.classdict = {'normal': 0, 'tumor': 1} #
#self.classdict = {'Normal': 0, 'TCGA-LUAD': 1, 'TCGA-LUSC': 2}
self._up_kwargs = {'mode': 'bilinear'}
def __getitem__(self, index):
sample = {}
info = self.ids[index].replace('\n', '')
#file_name, label = info.split('\t')[0].rsplit('.', 1)[0], info.split('\t')[1]
file_name, label = info.split('\t')[0], info.split('\t')[1]
sample['label'] = self.classdict[label]
sample['id'] = file_name
file_path = os.path.join(self.root, 'simclr_files')
#feature_path = os.path.join(self.root, file_name, 'features.pt')
feature_path = os.path.join(file_path, file_name, 'features.pt')
if os.path.exists(feature_path):
features = torch.load(feature_path, map_location=lambda storage, loc: storage)
else:
print(feature_path + ' not exists')
features = torch.zeros(1, 512)
#adj_s_path = os.path.join(self.root, file_name, 'adj_s.pt')
adj_s_path = os.path.join(file_path, file_name, 'adj_s.pt')
if os.path.exists(adj_s_path):
adj_s = torch.load(adj_s_path, map_location=lambda storage, loc: storage)
else:
print(adj_s_path + ' not exists')
adj_s = torch.ones(features.shape[0], features.shape[0])
#features = features.unsqueeze(0)
sample['image'] = features
sample['adj_s'] = adj_s #adj_s.to(torch.double)
# return {'image': image.astype(np.float32), 'label': label.astype(np.int64)}
return sample
def __len__(self):
return len(self.ids)
''' def __getitem__(self, index):
sample = {}
info = self.ids[index].replace('\n', '')
file_name, label = info.split('\t')[0].rsplit('.', 1)[0], info.split('\t')[1]
site, file_name = file_name.split('/')
# if site =='CCRCC':
# file_path = self.root + 'CPTAC_CCRCC_features/simclr_files'
if site =='LUAD' or site =='LSCC':
site = 'LUNG'
file_path = self.root + 'CPTAC_{}_features/simclr_files'.format(site) #_pre# with # rushin
# For NLST only
if site =='NLST':
file_path = self.root + 'NLST_Lung_features/simclr_files'
# For TCGA only
if site =='TCGA':
file_name = info.split('\t')[0]
_, file_name = file_name.split('/')
file_path = self.root + 'TCGA_LUNG_features/simclr_files' #_resnet_with
sample['label'] = self.classdict[label]
sample['id'] = file_name
#feature_path = os.path.join(self.root, file_name, 'features.pt')
feature_path = os.path.join(file_path, file_name, 'features.pt')
if os.path.exists(feature_path):
features = torch.load(feature_path, map_location=lambda storage, loc: storage)
else:
print(feature_path + ' not exists')
features = torch.zeros(1, 512)
#adj_s_path = os.path.join(self.root, file_name, 'adj_s.pt')
adj_s_path = os.path.join(file_path, file_name, 'adj_s.pt')
if os.path.exists(adj_s_path):
adj_s = torch.load(adj_s_path, map_location=lambda storage, loc: storage)
else:
print(adj_s_path + ' not exists')
adj_s = torch.ones(features.shape[0], features.shape[0])
#features = features.unsqueeze(0)
sample['image'] = features
sample['adj_s'] = adj_s #adj_s.to(torch.double)
# return {'image': image.astype(np.float32), 'label': label.astype(np.int64)}
return sample
''' |