mohamed-ezz / saratan Goto Github PK

Hi,
I searched through the repo and couldnt find any link to a pretrained model that I can run and reproduce results. Pardon me if I oversaw and missed it

Where can I download these two files, which are used in "notebooks/realtime_train.ipynb"?
Thanks.

I'm using below protoxt, what's wrong? I couldn't able to predict lesion:

name: "phseg_v5"
force_backward: true

layer {
name: "data"
type: "Python"
top: "data"
top: "label"
python_param {
module: "numpy_data_layer"
layer: "NumpyTrainDataLayer"
}
include: { phase: TRAIN }
}

layer {
name: "data"
type: "Python"
top: "data"
top: "label"
python_param {
module: "numpy_data_layer"
layer: "NumpyTestDataLayer"
}
include: { phase: TEST }
}

layer {
name: "conv_d0a-b"
type: "Convolution"
bottom: "data"
top: "d0b"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 64
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}

layer {
name: "relu_d0b"
type: "ReLU"
bottom: "d0b"
top: "d0b"
}
layer {
name: "conv_d0b-c"
type: "Convolution"
bottom: "d0b"
top: "d0c"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 64
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}

layer {
name: "relu_d0c"
type: "ReLU"
bottom: "d0c"
top: "d0c"
}
layer {
name: "pool_d0c-1a"
type: "Pooling"
bottom: "d0c"
top: "d1a"
pooling_param {
pool: MAX
kernel_size: 2
stride: 2
}
}
layer {
name: "conv_d1a-b"
type: "Convolution"
bottom: "d1a"
top: "d1b"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 128
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}

layer {
name: "relu_d1b"
type: "ReLU"
bottom: "d1b"
top: "d1b"
}
layer {
name: "conv_d1b-c"
type: "Convolution"
bottom: "d1b"
top: "d1c"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 128
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}

layer {
name: "relu_d1c"
type: "ReLU"
bottom: "d1c"
top: "d1c"
}
layer {
name: "pool_d1c-2a"
type: "Pooling"
bottom: "d1c"
top: "d2a"
pooling_param {
pool: MAX
kernel_size: 2
stride: 2
}
}
layer {
name: "conv_d2a-b"
type: "Convolution"
bottom: "d2a"
top: "d2b"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 256
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}

layer {
name: "relu_d2b"
type: "ReLU"
bottom: "d2b"
top: "d2b"
}
layer {
name: "conv_d2b-c"
type: "Convolution"
bottom: "d2b"
top: "d2c"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 256
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}

layer {
name: "relu_d2c"
type: "ReLU"
bottom: "d2c"
top: "d2c"
}
layer {
name: "pool_d2c-3a"
type: "Pooling"
bottom: "d2c"
top: "d3a"
pooling_param {
pool: MAX
kernel_size: 2
stride: 2
}
}
layer {
name: "conv_d3a-b"
type: "Convolution"
bottom: "d3a"
top: "d3b"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 512
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}

layer {
name: "relu_d3b"
type: "ReLU"
bottom: "d3b"
top: "d3b"
}
layer {
name: "conv_d3b-c"
type: "Convolution"
bottom: "d3b"
top: "d3c"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 512
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}

layer {
name: "relu_d3c"
type: "ReLU"
bottom: "d3c"
top: "d3c"
}

layer {
name: "pool_d3c-4a"
type: "Pooling"
bottom: "d3c"
top: "d4a"
pooling_param {
pool: MAX
kernel_size: 2
stride: 2
}
}
layer {
name: "conv_d4a-b"
type: "Convolution"
bottom: "d4a"
top: "d4b"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 1024
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}

layer {
name: "relu_d4b"
type: "ReLU"
bottom: "d4b"
top: "d4b"
}
layer {
name: "conv_d4b-c"
type: "Convolution"
bottom: "d4b"
top: "d4c"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 1024
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}

layer {
name: "relu_d4c"
type: "ReLU"
bottom: "d4c"
top: "d4c"
}

layer {
name: "upconv_d4c_u3a"
type: "Deconvolution"
bottom: "d4c"
top: "u3a"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 512
pad: 0
kernel_size: 2
stride: 2
weight_filler {
type: "xavier"
}
}
}

layer {
name: "relu_u3a"
type: "ReLU"
bottom: "u3a"
top: "u3a"
}
layer {
name: "crop_d3c-d3cc"
type: "Crop"
bottom: "d3c"
bottom: "u3a"
top: "d3cc"

}
layer {
name: "concat_d3cc_u3a-b"
type: "Concat"
bottom: "u3a"
bottom: "d3cc"
top: "u3b"
}
layer {
name: "conv_u3b-c"
type: "Convolution"
bottom: "u3b"
top: "u3c"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 512
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}
layer {
name: "relu_u3c"
type: "ReLU"
bottom: "u3c"
top: "u3c"
}
layer {
name: "conv_u3c-d"
type: "Convolution"
bottom: "u3c"
top: "u3d"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 512
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}
layer {
name: "relu_u3d"
type: "ReLU"
bottom: "u3d"
top: "u3d"
}
layer {
name: "upconv_u3d_u2a"
type: "Deconvolution"
bottom: "u3d"
top: "u2a"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 256
pad: 0
kernel_size: 2
stride: 2
weight_filler {
type: "xavier"
}
}
}
layer {
name: "relu_u2a"
type: "ReLU"
bottom: "u2a"
top: "u2a"
}
layer {
name: "crop_d2c-d2cc"
type: "Crop"
bottom: "d2c"
bottom: "u2a"
top: "d2cc"

}
layer {
name: "concat_d2cc_u2a-b"
type: "Concat"
bottom: "u2a"
bottom: "d2cc"
top: "u2b"
}
layer {
name: "conv_u2b-c"
type: "Convolution"
bottom: "u2b"
top: "u2c"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 256
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}
layer {
name: "relu_u2c"
type: "ReLU"
bottom: "u2c"
top: "u2c"
}
layer {
name: "conv_u2c-d"
type: "Convolution"
bottom: "u2c"
top: "u2d"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 256
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}
layer {
name: "relu_u2d"
type: "ReLU"
bottom: "u2d"
top: "u2d"
}
layer {
name: "upconv_u2d_u1a"
type: "Deconvolution"
bottom: "u2d"
top: "u1a"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 128
pad: 0
kernel_size: 2
stride: 2
weight_filler {
type: "xavier"
}
}
}
layer {
name: "relu_u1a"
type: "ReLU"
bottom: "u1a"
top: "u1a"
}
layer {
name: "crop_d1c-d1cc"
type: "Crop"
bottom: "d1c"
bottom: "u1a"
top: "d1cc"

}
layer {
name: "concat_d1cc_u1a-b"
type: "Concat"
bottom: "u1a"
bottom: "d1cc"
top: "u1b"
}
layer {
name: "conv_u1b-c"
type: "Convolution"
bottom: "u1b"
top: "u1c"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 128
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}
layer {
name: "relu_u1c"
type: "ReLU"
bottom: "u1c"
top: "u1c"
}
layer {
name: "conv_u1c-d"
type: "Convolution"
bottom: "u1c"
top: "u1d"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 128
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}
layer {
name: "relu_u1d"
type: "ReLU"
bottom: "u1d"
top: "u1d"
}
layer {
name: "upconv_u1d_u0a_NEW"
type: "Deconvolution"
bottom: "u1d"
top: "u0a"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 64
pad: 0
kernel_size: 2
stride: 2
weight_filler {
type: "xavier"
}
}
}
layer {
name: "relu_u0a"
type: "ReLU"
bottom: "u0a"
top: "u0a"
}
layer {
name: "crop_d0c-d0cc"
type: "Crop"
bottom: "d0c"
bottom: "u0a"
top: "d0cc"

}
layer {
name: "concat_d0cc_u0a-b"
type: "Concat"
bottom: "u0a"
bottom: "d0cc"
top: "u0b"
}
layer {
name: "conv_u0b-c_New"
type: "Convolution"
bottom: "u0b"
top: "u0c"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 64
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}
layer {
name: "relu_u0c"
type: "ReLU"
bottom: "u0c"
top: "u0c"
}
layer {
name: "conv_u0c-d_New"
type: "Convolution"
bottom: "u0c"
top: "u0d"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 64
pad: 0
kernel_size: 3
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}
layer {
name: "relu_u0d"
type: "ReLU"
bottom: "u0d"
top: "u0d"
}
layer {
name: "conv_u0d-score_New"
type: "Convolution"
bottom: "u0d"
top: "score"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 2
pad: 0
kernel_size: 1
weight_filler {
type: "xavier"
}
engine: CAFFE
}
}

layer {
name: "loss"
type: "SoftmaxWithLoss"
bottom: "score"
bottom: "label"
top: "loss"
include {
phase: TRAIN
}
softmax_param {engine: CAFFE}
loss_param {
weight_by_label_freqs: true
ignore_label: 2
class_weighting: 0.25
class_weighting: 9.664
}

}
layer {
name: "prob"
type: "Softmax"
bottom: "score"
top: "prob"
include {
phase: TEST
}
}

layer {
name: "accuracy"
type: "Accuracy"
bottom: "score"
bottom: "label"
top: "accuracy"
accuracy_param {
ignore_label: 2
}
include {
phase: TEST
}
}

i have two questions 👍

1. where to download your train-data(like 3Dircab1)
   2.Is CRF necessary or not(if necessary,how to run it)? and should i train with your Recompiled-caffenet or offical version？does this matter？
   Thank you

Hi, Mohamed,
I noticed that you use NLopt to tuning CRF's parameters. Once I use this library to tune the parameters and I got bad result.
I use your CRF parameter after liver inference and got good result. But when I tuned the parameter the dice was going to approach the dice which got without CRF. So does the lesion inference.
I wonder know why. Thank you!@mohamed-ezz

Hi,
I'm learning your codes in these days and I added CRF after net1 and got a good result.
But When I add CRF after net2 I got {'voe': 1.0, 'dice': 0.0, 'rvd': -1.0, 'assd': 0, 'jaccard': 0.0, 'msd': 0}.
Could you please tell why or would you mind giving me your CRF processing code after net2 ?

Thanks

Hello,
I read your codes and I noticed that you said in the function step1_preprocess_img_slice has this comment below:
1- Rotate the input volume so the the liver is on the left, spine is at the bottom of the image
I wonder know how can you make sure the liver is always on the left. I'll appricate if you could help me. Thanks.

Hello. I want to use your net to train my own dataset from scratch. BuT no code for training is available. Please tell me how to configure this net for my own training.

Thanks

Hi, I see your source and your paper, both of them were very useful.
I saw you used thresholding in the last part of network training, but I couldn't find anything in the paper, about it.
Could you please explain the matter of using thresholding and the type of thresholding you used?

On giving the CRF an image of W,H,D and its probability mask of W,H,D,L where L = 1 I am getting a matrix of 0's, I tried running the CRF optimizer but I got a matrix of 0's for all iterations
Can you help me out with the same also can this CRF be used for datasets other than the liver as a general segmentation improvement postprocessing step?

I only have 3D-IRCADb dataset.
(1) I should how to process my data so that i can train?
(2) So i should how to read '/data' and '/data/layers' folders?
I only know the function of numpy_data_layer.py.
Because i dont have .nii format data in 3D-IRCADb dataset, I annotated the code
# image_filename = os.path.join(volume_fulldirname,"image"+volume_id+".nii")
# flip_volume(image_filename, os.path.join(OUTPUT_PATH, "image%.2d"%int(volume_id)+".nii")) in irca_to_nii.py.

(3) What is the link to fire3 dataset?

I fellow this script data/irca_to_nii.py.
I downloaded the 3D-IRCADb-01 dataset ran this scipt.
But there are some errrors.

➜  3Dircadb1 python ../CFCN/saratan/data/irca_to_nii.py
Volume 3Dircadb1.1
	 Flipping left-right ./3Dircadb1.1/image1.nii to ./niftis_segmented_all/image01.nii
Traceback (most recent call last):
  File "../CFCN/saratan/data/irca_to_nii.py", line 86, in <module>
    flip_volume(image_filename, os.path.join(OUTPUT_PATH, "image%.2d"%int(volume_id)+".nii"))
  File "../CFCN/saratan/data/irca_to_nii.py", line 27, in flip_volume
    volume = nibabel.load(input_filename).get_data()
  File "/usr/local/lib/python2.7/dist-packages/nibabel/loadsave.py", line 40, in load
    raise FileNotFoundError("No such file: '%s'" % filename)
nibabel.py3k.FileNotFoundError: No such file: './3Dircadb1.1/image1.nii'

so, where is the 3Dircadb1.1/image1.nii file?

How many dataset did you use for training the model, test the model?