import os
import sys
from tqdm import tqdm
from tensorboardX import SummaryWriter
import shutil
import argparse
import logging
import time
import random
import numpy as np
import torch
import torch.optim as optim
from torchvision import transforms
import torch.nn.functional as F
import torch.backends.cudnn as cudnn
from torch.utils.data import DataLoader
from torchvision.utils import make_grid
from networks.vnet import VNet
from dataloaders.la_heart import LAHeart, RandomCrop, CenterCrop, RandomRotFlip, ToTensor, TwoStreamBatchSampler
from scipy.ndimage import distance_transform_edt as distance
from skimage import segmentation as skimage_seg
# Heart MR segmentation with boundary loss
parser = argparse.ArgumentParser()
parser.add_argument('--root_path', type=str, default='../data/2018LA_Seg_Training Set/', help='Name of Experiment')
parser.add_argument('--exp', type=str, default='vnet_dp_bd_sdf', help='model_name')
parser.add_argument('--max_iterations', type=int, default=20000, help='maximum epoch number to train')
parser.add_argument('--batch_size', type=int, default=4, help='batch_size per gpu')
parser.add_argument('--base_lr', type=float, default=0.001, help='maximum epoch number to train')
parser.add_argument('--deterministic', type=int, default=1, help='whether use deterministic training')
parser.add_argument('--seed', type=int, default=2019, help='random seed')
parser.add_argument('--gpu', type=str, default='0', help='GPU to use')
args = parser.parse_args()
train_data_path = args.root_path
snapshot_path = "../model_la/" + args.exp + "/"
os.environ['CUDA_VISIBLE_DEVICES'] = args.gpu
batch_size = args.batch_size * len(args.gpu.split(','))
max_iterations = args.max_iterations
base_lr = args.base_lr
if args.deterministic:
cudnn.benchmark = False
cudnn.deterministic = True
random.seed(args.seed)
np.random.seed(args.seed)
torch.manual_seed(args.seed)
torch.cuda.manual_seed(args.seed)
def dice_loss(score, target):
target = target.float()
smooth = 1e-5
intersect = torch.sum(score * target)
y_sum = torch.sum(target * target)
z_sum = torch.sum(score * score)
loss = (2 * intersect + smooth) / (z_sum + y_sum + smooth)
loss = 1 - loss
return loss
def compute_sdf1_1(img_gt, out_shape):
"""
compute the normalized signed distance map of binary mask
input: segmentation, shape = (batch_size, x, y, z)
output: the Signed Distance Map (SDM)
sdf(x) = 0; x in segmentation boundary
-inf|x-y|; x in segmentation
+inf|x-y|; x out of segmentation
normalize sdf to [-1, 1]
"""
img_gt = img_gt.astype(np.uint8)
normalized_sdf = np.zeros(out_shape)
for b in range(out_shape[0]): # batch size
# ignore background
for c in range(1, out_shape[1]):
posmask = img_gt[b]
negmask = 1-posmask
posdis = distance(posmask)
negdis = distance(negmask)
boundary = skimage_seg.find_boundaries(posmask, mode='inner').astype(np.uint8)
sdf = (negdis-np.min(negdis))/(np.max(negdis)-np.min(negdis)) - (posdis-np.min(posdis))/(np.max(posdis)-np.min(posdis))
sdf[boundary==1] = 0
normalized_sdf[b][c] = sdf
assert np.min(sdf) == -1.0, print(np.min(posdis), np.min(negdis), np.max(posdis), np.max(negdis))
assert np.max(sdf) == 1.0, print(np.min(posdis), np.min(negdis), np.max(posdis), np.max(negdis))
return normalized_sdf
def compute_sdf(img_gt, out_shape):
"""
compute the signed distance map of binary mask
input: segmentation, shape = (batch_size, x, y, z)
output: the Signed Distance Map (SDM)
sdf(x) = 0; x in segmentation boundary
-inf|x-y|; x in segmentation
+inf|x-y|; x out of segmentation
"""
img_gt = img_gt.astype(np.uint8)
gt_sdf = np.zeros(out_shape)
for b in range(out_shape[0]): # batch size
for c in range(1, out_shape[1]):
posmask = img_gt[b]
negmask = 1-posmask
posdis = distance(posmask)
negdis = distance(negmask)
boundary = skimage_seg.find_boundaries(posmask, mode='inner').astype(np.uint8)
sdf = negdis - posdis
sdf[boundary==1] = 0
gt_sdf[b][c] = sdf
return gt_sdf
def boundary_loss(outputs_soft, gt_sdf):
"""
compute boundary loss for binary segmentation
input: outputs_soft: softmax results, shape=(b,2,x,y,z)
gt_sdf: sdf of ground truth (can be original or normalized sdf); shape=(b,2,x,y,z)
output: boundary_loss; sclar
"""
pc = outputs_soft[:,1,...]
dc = gt_sdf[:,1,...]
multipled = torch.einsum('bxyz, bxyz->bxyz', pc, dc)
bd_loss = multipled.mean()
return bd_loss
patch_size = (112, 112, 80)
num_classes = 2
if __name__ == "__main__":
## make logger file
if not os.path.exists(snapshot_path):
os.makedirs(snapshot_path)
if os.path.exists(snapshot_path + '/code'):
shutil.rmtree(snapshot_path + '/code')
shutil.copytree('.', snapshot_path + '/code', shutil.ignore_patterns(['.git','__pycache__']))
logging.basicConfig(filename=snapshot_path+"/log.txt", level=logging.INFO,
format='[%(asctime)s.%(msecs)03d] %(message)s', datefmt='%H:%M:%S')
logging.getLogger().addHandler(logging.StreamHandler(sys.stdout))
logging.info(str(args))
net = VNet(n_channels=1, n_classes=num_classes, normalization='batchnorm', has_dropout=True)
net = net.cuda()
db_train = LAHeart(base_dir=train_data_path,
split='train',
num=16,
transform = transforms.Compose([
RandomRotFlip(),
RandomCrop(patch_size),
ToTensor(),
]))
def worker_init_fn(worker_id):
random.seed(args.seed+worker_id)
trainloader = DataLoader(db_train, batch_size=batch_size, shuffle=True, num_workers=4, pin_memory=True, worker_init_fn=worker_init_fn)
net.train()
optimizer = optim.SGD(net.parameters(), lr=base_lr, momentum=0.9, weight_decay=0.0001)
writer = SummaryWriter(snapshot_path+'/log', flush_secs=2)
logging.info("{} itertations per epoch".format(len(trainloader)))
iter_num = 0
alpha = 1.0
max_epoch = max_iterations//len(trainloader)+1
lr_ = base_lr
net.train()
for epoch_num in tqdm(range(max_epoch), ncols=70):
time1 = time.time()
for i_batch, sampled_batch in enumerate(trainloader):
time2 = time.time()
# print('fetch data cost {}'.format(time2-time1))
# volume_batch.shape=(b,1,x,y,z) label_patch.shape=(b,x,y,z)
volume_batch, label_batch = sampled_batch['image'], sampled_batch['label']
volume_batch, label_batch = volume_batch.cuda(), label_batch.cuda()
outputs = net(volume_batch)
loss_seg = F.cross_entropy(outputs, label_batch)
outputs_soft = F.softmax(outputs, dim=1)
loss_seg_dice = dice_loss(outputs_soft[:, 1, :, :, :], label_batch == 1)
# compute gt_signed distance function and boundary loss
with torch.no_grad():
# defalut using compute_sdf; however, compute_sdf1_1 is also worth to try;
gt_sdf_npy = compute_sdf(label_batch.cpu().numpy(), outputs_soft.shape)
gt_sdf = torch.from_numpy(gt_sdf_npy).float().cuda(outputs_soft.device.index)
# show signed distance map for debug
# import matplotlib.pyplot as plt
# plt.figure()
# plt.subplot(121), plt.imshow(gt_sdf_npy[0,1,:,:,40]), plt.colorbar()
# plt.subplot(122), plt.imshow(np.uint8(label_batch.cpu().numpy()[0,:,:,40]>0)), plt.colorbar()
# plt.show()
loss_boundary = boundary_loss(outputs_soft, gt_sdf)
loss = alpha*(loss_seg+loss_seg_dice) + (1 - alpha) * loss_boundary
optimizer.zero_grad()
loss.backward()
optimizer.step()
iter_num = iter_num + 1
writer.add_scalar('lr', lr_, iter_num)
writer.add_scalar('loss/loss_seg', loss_seg, iter_num)
writer.add_scalar('loss/loss_seg_dice', loss_seg_dice, iter_num)
writer.add_scalar('loss/loss_boundary', loss_boundary, iter_num)
writer.add_scalar('loss/loss', loss, iter_num)
writer.add_scalar('loss/alpha', alpha, iter_num)
logging.info('iteration %d : alpha : %f' % (iter_num, alpha))
logging.info('iteration %d : loss_seg_dice : %f' % (iter_num, loss_seg_dice.item()))
logging.info('iteration %d : loss_boundary : %f' % (iter_num, loss_boundary.item()))
logging.info('iteration %d : loss : %f' % (iter_num, loss.item()))
if iter_num % 2 == 0:
image = volume_batch[0, 0:1, :, :, 20:61:10].permute(3,0,1,2).repeat(1,3,1,1)
grid_image = make_grid(image, 5, normalize=True)
writer.add_image('train/Image', grid_image, iter_num)
image = outputs_soft[0, 1:2, :, :, 20:61:10].permute(3, 0, 1, 2).repeat(1, 3, 1, 1)
grid_image = make_grid(image, 5, normalize=False)
writer.add_image('train/Predicted_label', grid_image, iter_num)
image = label_batch[0, :, :, 20:61:10].unsqueeze(0).permute(3, 0, 1, 2).repeat(1, 3, 1, 1)
grid_image = make_grid(image, 5, normalize=False)
writer.add_image('train/Groundtruth_label', grid_image, iter_num)
image = gt_sdf[0, 1:2, :, :, 20:61:10].permute(3, 0, 1, 2).repeat(1, 3, 1, 1)
grid_image = make_grid(image, 5, normalize=True)
writer.add_image('train/gt_sdf', grid_image, iter_num)
## change lr
if iter_num % 2500 == 0:
lr_ = base_lr * 0.1 ** (iter_num // 2500)
for param_group in optimizer.param_groups:
param_group['lr'] = lr_
if iter_num % 1000 == 0:
save_mode_path = os.path.join(snapshot_path, 'iter_' + str(iter_num) + '.pth')
torch.save(net.state_dict(), save_mode_path)
logging.info("save model to {}".format(save_mode_path))
if iter_num > max_iterations:
break
time1 = time.time()
alpha -= 0.01
if alpha <= 0.01:
alpha = 0.01
if iter_num > max_iterations:
break
save_mode_path = os.path.join(snapshot_path, 'iter_'+str(max_iterations+1)+'.pth')
torch.save(net.state_dict(), save_mode_path)
logging.info("save model to {}".format(save_mode_path))
writer.close()
