search

faustomilletari / VNet

GNU General Public License v3.0
284 stars 123 forks source link

How to convert MRI images into lmdb #12

Closed sagarhukkire closed 7 years ago

sagarhukkire commented 7 years ago

Hi

I am trying to use Vnet for image segmentation MRI images. Do you have any script to convert MRI images into lmdb format for caffe? It will be great help

faustomilletari commented 7 years ago

I would use h5. In that case it's easy(er)

Fausto Milletarì Sent from my iPhone

On 25 Jan 2017, at 08:00, sagarax009 notifications@github.com wrote:

Hi

I am trying to use Vnet for image segmentation MRI images. Do you have any script to convert MRI images into lmdb format for caffe? It will be great help

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

sagarhukkire commented 7 years ago

Hi Fausto

Sorry to ask you again. As I am new to this and I found tough to get blogs which are referring DIGITS and all. to use h5 ? can you please explain little bit

faustomilletari commented 7 years ago

I guess that you should load the mri images with simpleITK and then dump them into a lmdb or h5 file by importing and using the necessary library in your python script. How to use these libraries can be found online. With h5 I remember it was quite easy because you would dump a dictionary of volumes and a dictionary of label as far as I remember into a h5 file: A 5 lines long script really. With lmdb I think I recall it was more difficult or there were problems..

Why not use tensorflow?

Fausto Milletarì Sent from my iPhone

On 25 Jan 2017, at 08:08, sagarax009 notifications@github.com wrote:

Hi Fausto

Sorry to ask you again. As I am new to this and I found tough to get blogs which are referring DIGITS and all. to use h5 ? can you please explain little bit

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

sagarhukkire commented 7 years ago

Hi

I am able to get volumes of all MRI images ..as you mentioned volumetric in vnet approach. I am not getting how I should pass this volumes as input to caffe.

Please guide and thanks in advance

Sagar Hukkire

On Wed, Jan 25, 2017 at 2:03 PM, Fausto Milletari notifications@github.com wrote:

I would use h5. In that case it's easy(er)

Fausto Milletarì Sent from my iPhone

On 25 Jan 2017, at 08:00, sagarax009 notifications@github.com wrote:

Hi

I am trying to use Vnet for image segmentation MRI images. Do you have any script to convert MRI images into lmdb format for caffe? It will be great help

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/faustomilletari/VNet/issues/12#issuecomment-275102668, or mute the thread https://github.com/notifications/unsubscribe-auth/ATFxy_feb34P2KXA9IbGSdFIomB7757iks5rV0gmgaJpZM4LtdRx .

faustomilletari commented 7 years ago

Vnet code loads them from a directory.have a look at data manager class

Fausto Milletarì Sent from my iPhone

On 29 Jan 2017, at 16:14, sagarax009 notifications@github.com wrote:

Hi

I am able to get volumes of all MRI images ..as you mentioned volumetric in vnet approach. I am not getting how I should pass this volumes as input to caffe.

Please guide and thanks in advance

Sagar Hukkire

On Wed, Jan 25, 2017 at 2:03 PM, Fausto Milletari notifications@github.com wrote:

I would use h5. In that case it's easy(er)

Fausto Milletarì Sent from my iPhone

On 25 Jan 2017, at 08:00, sagarax009 notifications@github.com wrote:

Hi

I am trying to use Vnet for image segmentation MRI images. Do you have any script to convert MRI images into lmdb format for caffe? It will be great help

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/faustomilletari/VNet/issues/12#issuecomment-275102668, or mute the thread https://github.com/notifications/unsubscribe-auth/ATFxy_feb34P2KXA9IbGSdFIomB7757iks5rV0gmgaJpZM4LtdRx .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

sagarhukkire commented 7 years ago

Great man you did great. The code is self explanation. 1)Currently i have fetched volumes of Psoyas major MRI images into .stl format so I need to convert into sitK format? 2) what about parameters, what is right choice ? many thanks

Sagar Hukkire

On Sun, Jan 29, 2017 at 10:16 PM, Fausto Milletari <notifications@github.com

wrote:

Vnet code loads them from a directory.have a look at data manager class

Fausto Milletarì Sent from my iPhone

On 29 Jan 2017, at 16:14, sagarax009 notifications@github.com wrote:

Hi

I am able to get volumes of all MRI images ..as you mentioned volumetric in vnet approach. I am not getting how I should pass this volumes as input to caffe.

Please guide and thanks in advance

Sagar Hukkire

On Wed, Jan 25, 2017 at 2:03 PM, Fausto Milletari < notifications@github.com> wrote:

I would use h5. In that case it's easy(er)

Fausto Milletarì Sent from my iPhone

On 25 Jan 2017, at 08:00, sagarax009 notifications@github.com wrote:

Hi

I am trying to use Vnet for image segmentation MRI images. Do you have any script to convert MRI images into lmdb format for caffe? It will be great help

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/faustomilletari/VNet/issues/ 12#issuecomment-275102668, or mute the thread https://github.com/notifications/unsubscribe-auth/ATFxy_ feb34P2KXA9IbGSdFIomB7757iks5rV0gmgaJpZM4LtdRx .

— You are receiving this because you commented.

Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/faustomilletari/VNet/issues/12#issuecomment-275946717, or mute the thread https://github.com/notifications/unsubscribe-auth/ATFxy4mfSrE9OZpjpLk1gQQnnTnsjAvqks5rXQHKgaJpZM4LtdRx .

faustomilletari commented 7 years ago

Hello,

I hope it's fine. So simple itk reads all the formats that itk reads. Mha, mhd and so on. In other words all the formats that 3d slicer can read.

Regards,

Fausto Milletarì Sent from my iPhone

On 29 Jan 2017, at 16:32, sagarax009 notifications@github.com wrote:

Great man you did great. The code is self explanation. 1)Currently i have fetched volumes of Psoyas major MRI images into .stl format so I need to convert into sitK format? 2) what about parameters, what is right choice ? many thanks

Sagar Hukkire

On Sun, Jan 29, 2017 at 10:16 PM, Fausto Milletari <notifications@github.com

wrote:

Vnet code loads them from a directory.have a look at data manager class

Fausto Milletarì Sent from my iPhone

On 29 Jan 2017, at 16:14, sagarax009 notifications@github.com wrote:

Hi

I am able to get volumes of all MRI images ..as you mentioned volumetric in vnet approach. I am not getting how I should pass this volumes as input to caffe.

Please guide and thanks in advance

Sagar Hukkire

On Wed, Jan 25, 2017 at 2:03 PM, Fausto Milletari < notifications@github.com> wrote:

I would use h5. In that case it's easy(er)

Fausto Milletarì Sent from my iPhone

On 25 Jan 2017, at 08:00, sagarax009 notifications@github.com wrote:

Hi

I am trying to use Vnet for image segmentation MRI images. Do you have any script to convert MRI images into lmdb format for caffe? It will be great help

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/faustomilletari/VNet/issues/ 12#issuecomment-275102668, or mute the thread https://github.com/notifications/unsubscribe-auth/ATFxy_ feb34P2KXA9IbGSdFIomB7757iks5rV0gmgaJpZM4LtdRx .

— You are receiving this because you commented.

Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/faustomilletari/VNet/issues/12#issuecomment-275946717, or mute the thread https://github.com/notifications/unsubscribe-auth/ATFxy4mfSrE9OZpjpLk1gQQnnTnsjAvqks5rXQHKgaJpZM4LtdRx .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

sagarhukkire commented 7 years ago

Hi Fausto

thanks man now i am bit relieved , what about parameters ,what will be right choice?

is any standards for MRI images

Sagar Hukkire

On Sun, Jan 29, 2017 at 11:15 PM, Fausto Milletari <notifications@github.com

wrote:

Hello,

I hope it's fine. So simple itk reads all the formats that itk reads. Mha, mhd and so on. In other words all the formats that 3d slicer can read.

Regards,

Fausto Milletarì Sent from my iPhone

On 29 Jan 2017, at 16:32, sagarax009 notifications@github.com wrote:

Great man you did great. The code is self explanation. 1)Currently i have fetched volumes of Psoyas major MRI images into .stl format so I need to convert into sitK format? 2) what about parameters, what is right choice ? many thanks

Sagar Hukkire

On Sun, Jan 29, 2017 at 10:16 PM, Fausto Milletari < notifications@github.com

wrote:

Vnet code loads them from a directory.have a look at data manager class

Fausto Milletarì Sent from my iPhone

On 29 Jan 2017, at 16:14, sagarax009 notifications@github.com wrote:

Hi

I am able to get volumes of all MRI images ..as you mentioned volumetric in vnet approach. I am not getting how I should pass this volumes as input to caffe.

Please guide and thanks in advance

Sagar Hukkire

On Wed, Jan 25, 2017 at 2:03 PM, Fausto Milletari < notifications@github.com> wrote:

I would use h5. In that case it's easy(er)

Fausto Milletarì Sent from my iPhone

On 25 Jan 2017, at 08:00, sagarax009 notifications@github.com wrote:

Hi

I am trying to use Vnet for image segmentation MRI images. Do you have any script to convert MRI images into lmdb format for caffe? It will be great help

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/faustomilletari/VNet/issues/ 12#issuecomment-275102668, or mute the thread https://github.com/notifications/unsubscribe-auth/ATFxy_ feb34P2KXA9IbGSdFIomB7757iks5rV0gmgaJpZM4LtdRx .

— You are receiving this because you commented.

Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/faustomilletari/VNet/issues/ 12#issuecomment-275946717, or mute the thread https://github.com/notifications/unsubscribe-auth/ ATFxy4mfSrE9OZpjpLk1gQQnnTnsjAvqks5rXQHKgaJpZM4LtdRx

.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/faustomilletari/VNet/issues/12#issuecomment-275950702, or mute the thread https://github.com/notifications/unsubscribe-auth/ATFxy4WZ6v0_tKCKzmaA7O87hP_KiVfkks5rXQ9ogaJpZM4LtdRx .

faustomilletari commented 7 years ago

I think you can first try with the parameters I have used.

I wish I could be more useful but I don't currently work with this anymore. I should code up everything in tensor flow to make everything easier for you guys, but I don't have time!

Fausto Milletarì Sent from my iPhone

On 29 Jan 2017, at 17:25, sagarax009 notifications@github.com wrote:

Hi Fausto

thanks man now i am bit relieved , what about parameters ,what will be right choice?

is any standards for MRI images

Sagar Hukkire

On Sun, Jan 29, 2017 at 11:15 PM, Fausto Milletari <notifications@github.com

wrote:

Hello,

I hope it's fine. So simple itk reads all the formats that itk reads. Mha, mhd and so on. In other words all the formats that 3d slicer can read.

Regards,

Fausto Milletarì Sent from my iPhone

On 29 Jan 2017, at 16:32, sagarax009 notifications@github.com wrote:

Great man you did great. The code is self explanation. 1)Currently i have fetched volumes of Psoyas major MRI images into .stl format so I need to convert into sitK format? 2) what about parameters, what is right choice ? many thanks

Sagar Hukkire

On Sun, Jan 29, 2017 at 10:16 PM, Fausto Milletari < notifications@github.com

wrote:

Vnet code loads them from a directory.have a look at data manager class

Fausto Milletarì Sent from my iPhone

On 29 Jan 2017, at 16:14, sagarax009 notifications@github.com wrote:

Hi

I am able to get volumes of all MRI images ..as you mentioned volumetric in vnet approach. I am not getting how I should pass this volumes as input to caffe.

Please guide and thanks in advance

Sagar Hukkire

On Wed, Jan 25, 2017 at 2:03 PM, Fausto Milletari < notifications@github.com> wrote:

I would use h5. In that case it's easy(er)

Fausto Milletarì Sent from my iPhone

On 25 Jan 2017, at 08:00, sagarax009 notifications@github.com wrote:

Hi

I am trying to use Vnet for image segmentation MRI images. Do you have any script to convert MRI images into lmdb format for caffe? It will be great help

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/faustomilletari/VNet/issues/ 12#issuecomment-275102668, or mute the thread https://github.com/notifications/unsubscribe-auth/ATFxy_ feb34P2KXA9IbGSdFIomB7757iks5rV0gmgaJpZM4LtdRx .

— You are receiving this because you commented.

Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/faustomilletari/VNet/issues/ 12#issuecomment-275946717, or mute the thread https://github.com/notifications/unsubscribe-auth/ ATFxy4mfSrE9OZpjpLk1gQQnnTnsjAvqks5rXQHKgaJpZM4LtdRx

.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/faustomilletari/VNet/issues/12#issuecomment-275950702, or mute the thread https://github.com/notifications/unsubscribe-auth/ATFxy4WZ6v0_tKCKzmaA7O87hP_KiVfkks5rXQ9ogaJpZM4LtdRx .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.