Image Processing of BRATS Data Using Pylearn2

128 views
Skip to first unread message

VIVEK RUHELA

unread,
Feb 13, 2016, 6:58:07 AM2/13/16
to pylearn-users
Hello Gyus,


          Currently I'm working on Brain Tumor Segmentation using Deep Learning using BRATS DATA with the help of Pylearn2. In BRATS DATA all the brain MRI are .mha image files. I have finished all the tutorials of Pylearn2 and I understood that how does it works but still I am facing lots of problem such as:


(A) I don't have the the BRATS DATA in binary form or I don't have the .pkl file of dataset. So How can I build those either binary files or .pkl files for BRATS DATA as per the procedure mentioned in Pylearn2 for Cifar10 and MNIST datasets. This is very important for starting work in Pylearn2.


(B) Same problem for .yaml files for BRATS DATA.


So guys please suggest me how to work with .mha files  or having customized dataset in Pylearn2.


Thanks

poorna....@gmail.com

unread,
Mar 7, 2016, 9:53:52 AM3/7/16
to pylearn-users
Hi Guys,
Even I am facing the above problem...

Frédéric Bastien

unread,
Mar 7, 2016, 12:28:28 PM3/7/16
to pylear...@googlegroups.com
Hi,

Just to tell you that there isn't support or development of Pylearn2 anymore. Check the warning of this page for possible replacements:

https://github.com/lisa-lab/pylearn2/

Fred

On Mon, Mar 7, 2016 at 9:53 AM, <poorna....@gmail.com> wrote:
Hi Guys,
                       Even I am facing the above problem...

--
You received this message because you are subscribed to the Google Groups "pylearn-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pylearn-user...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Message has been deleted

VIVEK RUHELA

unread,
Mar 7, 2016, 10:56:37 PM3/7/16
to pylearn-users
Hello,
  
       Thanks for reply sir. But I have started my work on pylearn2. As you said there is no support for pylearn2 so what are other alternatives that is useful for the same purpose.

Thanks.

Frédéric Bastien

unread,
Mar 8, 2016, 10:22:38 AM3/8/16
to pylear...@googlegroups.com
See the warning on the URL I gave, it give some alternative.

Fred

Thanks.

vivek ruhela

unread,
Mar 8, 2016, 1:35:53 PM3/8/16
to pylear...@googlegroups.com
What about Theano. Theano also deals with deep learning and block, keras are also theano based. Should I go with Theano

From: Frédéric Bastien
Sent: ‎08-‎03-‎2016 08:52 PM
To: pylear...@googlegroups.com
Subject: Re: [pylearn-users] Re: Image Processing of BRATS Data Using Pylearn2

You received this message because you are subscribed to a topic in the Google Groups "pylearn-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/pylearn-users/S5k4JAk7_NY/unsubscribe.
To unsubscribe from this group and all its topics, send an email to pylearn-user...@googlegroups.com.

Frédéric Bastien

unread,
Mar 8, 2016, 4:45:57 PM3/8/16
to pylear...@googlegroups.com
If you should go with Theano or not depend on many things. Blocs and keras are build on top of Theano to give an higher level of abstraction. Theano is more low level then those 2. Keras and Blocs are more like the pylearn2 level. Pylearn2 was also on top of Theano.

If you want to replace pylearn2 and keep the same high level abstraction, go with keras/blocs or lasagne. If you want more control and lower level abstraction, then use Theano directly.

Fred
Message has been deleted

VIVEK RUHELA

unread,
Mar 8, 2016, 10:31:03 PM3/8/16
to pylearn-users
Thanks for reply sir. Actually I am working in pylearn from last two months and now I have not much time left to switch. So I have time and hardware constraints. I have two months to finish my work and I have i7-5th generation , 64 core processor without gpus as a high end computer in my college. So is it possible to switch to block or keras with this time and hardware constraints. I was using pylearn because of its simplicity. I expect the same for blocks and keras have the same way of application for medical image processing. Thanks sir.
Message has been deleted
Message has been deleted
Message has been deleted

Frédéric Bastien

unread,
Mar 10, 2016, 1:17:59 PM3/10/16
to pylear...@googlegroups.com
I can't tell you for sure. It depend for many thinks. I have the impression keras is faster to learn, but this isn't sure. I don't know if any of them have support for medical stuff, but both of them have support for images.

Fred

On Tue, Mar 8, 2016 at 10:31 PM, VIVEK RUHELA <vivek...@gmail.com> wrote:
Thanks for reply sir. Actually I am working in pylearn from last two months and now I have not much time left to switch. So I have time and hardware constraints. I have two months to finish my work and I have i7-5th generation , 64 core processor without gpus as a high end computer in my college. So is it possible to switch to block or keras with this time and hardware constraints. I was using pylearn because of its simplicity. I expect the same for blocks and keras have the same way of application for medical image processing. Thanks sir.

keilal...@gmail.com

unread,
Nov 9, 2016, 4:10:44 AM11/9/16
to pylearn-users

Hi vivek.
I would like to know how you resolved the problem of using the images in .mha format. I really need to know if I can use .mha format on pylearn2 or Theano

Thanks

Frédéric Bastien

unread,
Nov 9, 2016, 8:35:22 AM11/9/16
to pylear...@googlegroups.com

That is the wrong question I think. You should ask if you can load in Python that file format. Then make sure to have your minibatch of examples in numpy.ndarray format and it will work.


--
You received this message because you are subscribed to the Google Groups "pylearn-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pylearn-users+unsubscribe@googlegroups.com.

Poornachandra Sandur

unread,
Nov 9, 2016, 12:12:21 PM11/9/16
to pylear...@googlegroups.com

Hi Frederick,
                       You quoted a pointer that " minibatch of numpy.ndarray " .I want to know how to create the minibatches for the .mha files of the BRATS DATASET using Python. Please give me some pointers on this ...

Frédéric Bastien

unread,
Nov 9, 2016, 1:14:16 PM11/9/16
to pylear...@googlegroups.com
I never used that file format, so I can't give more detail then this high level guidance.

Search for python lib that read that format. Then select a small part of the example as a minibatch.

Fred

Poornachandra Sandur

unread,
Nov 9, 2016, 8:52:40 PM11/9/16
to pylear...@googlegroups.com

Hi Frederick ,
                         Thanks for the reply ... I even want to know it for the natural images .. can I please  get some pointers on that ...

Reply all
Reply to author
Forward
0 new messages