Re: [Plastimatch] plastimatch crashed with mha files

[Paolo Zaffino]

Hi, The problem seems to be the img1.mha, not the img0.mha Baybe it should be coded into an unrecognized data type. Could you show the img1.mha header? Paolo Inviato da Yahoo! Mail su Android

---

From: Jim Chang <jim.c...@gmail.com>;
To: <plast...@googlegroups.com>;
Subject: [Plastimatch] plastimatch crashed with mha files
Sent: Sun, Feb 24, 2013 12:56:45 AM

I tried to run plastimatch to register two images but it immediately crashed (see the attached output). It appears that something wrong with the mha creation or my ITK (v2.30.1)   What I did is the following:   1. use matlab code to create a raw file. The code is: files=dir('*.dcm'); num_f=length(files); img_raw=zeros(512,512,length(files)); for i=1:length(files)   img=dicomread(files(i).name);   img=img.';   img_raw(:,:,i)=img; end fid=fopen('img0.raw','wb');  fwrite(fid,img_raw,'float'); fclose(fid); 2. use the raw_to_mha to convert the raw file to a mha file: raw_to_mha img0.raw img0.mha 3 "0.95 0.95 2.5" 3. Follow to run registration using the command plastimatch register command_file.txt with the example command_file.txt: # command_file.txt [GLOBAL] fixed=img1.mha moving=img0.mha img_out=warped_img0.mha xform_out=bspline_coefficients.txt [STAGE] xform=bspline impl=plastimatch threading=openmp max_its=30 regularization_lambda=0.005 grid_spac=100 100 100 res=4 4 2 Would greatly appreciate it if anybody can help. Jim     ===========================================   Loading fixed image: img1.mha ElementType required and not defined. MetaObject: Read: MET_Read Failed MetaImage: M_Read: Error parsing file MetaImage: Read: Cannot parse file ITK exception. itk::ExceptionObject (01C3F394) Location: "void __thiscall itk::MetaImageIO::ReadImageInformation(void)" File: ..\..\..\InsightToolkit-3.20.1\Code\IO\itkMetaImageIO.cxx Line: 97 Description: itk::ERROR: MetaImageIO(00540348): File cannot be read: img1.mha for reading. Reason: Invalid argument Error: unhandled file type for loading image (0) phase5.mha Press any key to continue . . . -- You received this message because you are subscribed to the Google Groups "Plastimatch" group. To unsubscribe from this group and stop receiving emails from it, send an email to plastimatch...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.

[Paolo Zaffino]

Sorry for the mistakes but the smartphone changes what I type. "Maybe it could" is what I wanted to type.

Inviato da Yahoo! Mail su Android

---

From: Paolo Zaffino <p.za...@yahoo.it>;
To: plast...@googlegroups.com <plast...@googlegroups.com>; jim.c...@gmail.com <jim.c...@gmail.com>;
Subject: Re: [Plastimatch] plastimatch crashed with mha files
Sent: Sun, Feb 24, 2013 9:07:22 AM

[Jim Chang]

The header of both mha files looks like:

 

ObjectType = Image

NDims = 3

BinaryData = True

BinaryDataByteOrderMSB = False

TransformMatrix = 1 0 0 0 1 0 0 0 1

Offset = 0 0 0

ElementSpacing = 0.97656 0.97656 2.5

DimSize = 3

AnatomicalOrientation = RPI

ElementType = MET_UCHAR

ElementDataFile = LOCAL

[Marta Peroni]

Hi Jim...
I think your DimSize is wrong...it should contain the number of pixels in each dimension, something like:
DimSize = 512 512 58

Hope this helps!
Marta

*******************************************************************
Marta Peroni, PhD

contacts:
mobile: +393488202136 (IT)

********************************************************************

[Jim Chang]

Hi Marta,

 

You are right! I re-generated mha files from the raw files using the command:

 

raw_to_mha img0.raw img0.mha "512 512 120" "0.95 0.95 2.5"

 

instead of

 

raw_to_mha img0.raw img0.mha 3 "0.95 0.95 2.5"

 

That does the trick! Thanks very much.

 

Jim

[Jim Chang]

Ooops. Apparently I replied too quickly. Although the dimensions are correct, the raw data type got changed from float to unsigned char. So the mha image is different from the raw image. The raw_to_mha command does not specify how to keep the format of the original raw file. Is there anything I missed?

 

Thanks,

Jim 

[Marta Peroni]

Hi Jim!
Another thing you can use to convert from dcm to mha, given you are already using matlab, is the function writemha.m (you can find it in plastimatch_directory/extra/matlab). You can directly pass the matrix of data, element spacing, data type, offset to the function and it will write the corresponding mha.
This is what I would do to get it working :)
Thanks
Marta

[Paolo Zaffino]

Hi all,

anyway you have another two possibilities:

1) use plastimatch convert to convert the dicom files into an mha file

2) use directly the dicom folder in the parameters file.

Paolo

---

Da: Marta Peroni <m.pe...@gmail.com>
A: Jim Chang <jim.c...@gmail.com>
Cc: plast...@googlegroups.com
Inviato: Domenica 24 Febbraio 2013 23:50
Oggetto: Re: [Plastimatch] plastimatch crashed with mha files

[Jim Chang]

Thanks Paolo.

 

I followed Marta's suggestion using matlab writemha and it worked! Thanks Marta.

 

Jim

[Paolo Zaffino]

Hi Jim,

I'm glad to hear this.

Paolo

---

Da: Jim Chang <jim.c...@gmail.com>
A: Paolo Zaffino <p.za...@yahoo.it>
Cc: "plast...@googlegroups.com" <plast...@googlegroups.com>
Inviato: Lunedì 25 Febbraio 2013 0:37

[Jim Chang]

Hi Paolo & Marta,

 

I have a slightly different question regarding creating an mha image from a series of dicom images. The mha image I have  looks wrong. Please refer to the attached figure on saggittal and coronal views. Obviously something goes wrong. This does not always happens, but does occur.

 

Anybody knows what can be done to have correct images? Many thanks!

 

Jim

[Paolo Zaffino]

Hi Jim,

did you generate this image using matlab?

If yes please try this plastimatch command:

plastimatch convert --input=dicom_folder --output-img=img.mha

In my opinion the problem is due to a "flip" of the 2D image.

If you generated it via matlab maybe the 2D matrix (the dicom file) has been rotated.

Hope this helps.

Paolo

---

Da: Jim Chang <jim.c...@gmail.com>

A: plast...@googlegroups.com
Cc: Paolo Zaffino <p.za...@yahoo.it>
Inviato: Lunedì 25 Febbraio 2013 22:21

Hi Jim!
Another thing you can use to convert from dcm to mha, given you are already using matlab, is the function writemha.m (you can find it in plastimatch_directory/extra/ matlab). You can directly pass the matrix of data, element spacing, data type, offset to the function and it will write the corresponding mha.

--
****************************** ****************************** *******

Marta Peroni, PhD

contacts:
mobile: +393488202136 (IT)

****************************** ****************************** ********

--
****************************** ****************************** *******
Marta Peroni, PhD

contacts:
mobile: +393488202136 (IT)

****************************** ****************************** ********

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

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

[Marta Peroni]

Hi!
it could be due to 2 things:
1) each 2D matrix image shall be transposed in matlab (as paolo suggested)
2) did you correctly ordered the CC axis according to Image Position Patient?

Marta

--
*******************************************************************

[Jim Chang]

Hi Paolo & Marta,

 

I used plastimatch command instead of using matlab code and got the right mha image.

You both are right that one/some of the dicom images are flipped. Is there an easy way

to detect it/them?

 

At any rate, I am good at this point, but may come back for more questions. :)

 

Thanks a lot, guys!

Jim

[Marta Peroni]

Well, the easiest solution is to use plastimatch convert and/or directly dicoms in the plastimatch parameter file.
If you do want to use matlab, than for the flip in CC direction, simply re-order (sort) the Image Position Patient and pile your images accordingly. For the 2D image rotation, it depends from the way matlab thinks (i.e. columwise) :)

Shoot a question whenever you fill like :)
marta

[Paolo Zaffino]

This is a good news!

If you didn't filp/transpose some 2D matrix during the "building" of the 3D matrix I think that you should look into the dicom header, in order to check if the image orentation  is always the same (the positive direction along which the voxel are counted) and in case "correct" it.

Anyway if you are interested in mha manipulation there is a python module that execute the same thing of the matlab files (read/write mha files).

The file is extra/python/mha.py

Have a nice day.

Paolo

---

Da: Jim Chang <jim.c...@gmail.com>
A: plast...@googlegroups.com

Inviato: Lunedì 25 Febbraio 2013 23:44