About dataset

[John1231983]

In your paper, you said

We evaluate the segmentation on the dataset of the MICCAI Multi-Atlas Labeling challenge1 (Landman and Warfield, 2012), which consists of T1-weighted MRI scans from 30 subjects of OASIS (Marcus et al., 2007). Manual segmentations were provided by Neuromorphometrics, Inc.2 under academic subscription.

I accessed the Neuromorphometrics website and it required to purchase about 1.500usd for academic subscription. It is very huge money for student. Do you really pay the money? Does it include raw image or just label image only? I also access the first link (http://masiweb.vuse.vanderbilt.edu/workshop2012/index.php?title=Main_Page&action=history) but it has no raw image

Thanks

[romainVala]

Hello,
did you manage to get the data ?
I would be interested too

Many thanks
Romain

[John1231983]

@romainVala : I still wait the response from author

[TJKlein]

You can get the data here .
Cheers,
Tassilo

[romainVala]

thanks
the form does not work any more, but I try a mail to bennett.landman I'll keep you informed if I succeed

[romainVala]

Google upgraded the forms. Here is a fixed link:

https://docs.google.com/forms/d/e/1FAIpQLSfwkdSt7hWo_tjHUDu2stDsxWTaWyLJIUiS_iapbtKaydEMIw/viewform?usp=sf_link

thanks to landman

[romainVala]

Hi now I have the data I realize you did not commit all your code
I just see the network definition but not the code for training testing the data.
Do I miss something?
Many thanks

[John1231983]

@romainVala: To run the code, you just download his caffe and compile it. After that, run training or just run develop (he already provide .caffemode)
@TJKlein: I want to ask you about the number of label. As the dataset, it contains 207 classes. But your code only consider 7 classes. Do we have any rule to map from 207 classes to 7 classes? Thanks

[YilinLiu97]

@John1231983 Have you successfully run the deployment code? I keep getting "Error parsing text-format caffe.NetParameter: 9:23: Message type "caffe.LayerParameter" has no field named "memory_datand_param" error.
Are there any pre-processing steps needed for running that code??

[John1231983]

@YilinLiu97 : Not yet. I cannot run it code. I am still confusing number of classes. Do you know how to convert 207 classes in the dataset to 7 classes in his code?

[YilinLiu97]

@John1231983 Are you attempting to re-train? otherwise, why do you need that step for just deploying the model??
btw which caffe interface are you using? Did you notice that memory data is used in this case and do you know how to deal with that in matcaffe/pycaffe? I got the error above even when I'm just trying to initialize the network. wired. I'm using matcaffe.

[YilinLiu97]

@John1231983 Could you please check out the second issue that I raised? I'm a newbie in caffe but It seems to me that there are some problems with the deployment prototxt, which is probably why it causes so many errors.

[John1231983]

He used his caffe code. Check it README

[YilinLiu97]

@John1231983 Thanks so much!