[Insight-developers] Re: [Insight-users] bad free() reading Signa images

kent williams norman-k-williams at uiowa.edu
Thu May 10 09:55:02 EDT 2007 

This behavior is my fault, since GE4ImageIO was a conversion of code from
the BRAINS2 application, which worked this way. I can understand how that
would interact poorly with the ImageSeriesReader.

It brings up an issue that has bothered me about how Image IO works in ITK:
While it promises and delivers a fair amount of image format independence,
it exposes the difference between monolithic image file formats and image
formats that are built from a series of 2D image files.

What I've started for brains2 is a separate layer of abstraction that takes
a path (a filename, a directory name), deduces how that can be translated
into an image volume, and then loads it.

In most ITK-based research software, the developer knows the incoming image
format, and can choose whether to use an image series reader or an image
reader.  That's all well and good, but in the case of more general image
processing applications, you want to be able to 'take all comers.'

And really, there's no good reason for using the same 10 or 15 lines of
boilerplate code every time you want to open a file.  Much better to write
it once and re-use it everywhere.  For novices, knowing which 10 or 15 lines
of boilerplate code to drop in is part of the learning curve to using ITK --
one that isn't strictly necessary.


On 5/9/07 5:05 PM, "Neil Weisenfeld" <neil at bwh.harvard.edu> wrote:

> Hi all,
> 
> (devs maybe read the last paragraph)
> 
> Here's why using GE4ImageIO with itkImageSeriesReader causes core-dumps:
> 
> itkImageSeriesReader reader attempts to read in an image slice by
> slice, however GE4ImageIO, when given the filename of a single slice,
> ***actually reads in the whole volume***
> 
> So while itkImageSeriesReader correctly allocates an output volume
> for the slice dimensions x #slices, it then reads the entire volume
> #slices many times obviously (well, not obviously, but see below)
> trampling over unallocated memory.
> 
> One solution is to use GE4ImageIO with itkImageReader, rather than
> itkImageSeriesReader, supply only the *first* filename in the series,
> and then let the internal magic pick up all of the slices.
> 
> Maybe stylistically, it would be good if the loop in
> itkImageSeriesReader that actually copies pixels checked both the
> input AND output arrays for IsAtEnd().  That would, at least, avoid
> the core dump.
> 
> 
> Regards,
> Neil
> 
> 
> 
> 
> 
> 
> 
> On May 8, 2007, at 2:29 PM, Neil Weisenfeld wrote:
> 
>> 
>> Hi all,
>> 
>> I'm trying to read some Signa 4.x images using the GE4ImageIO and
>> this seems to result in:
>> 
>> *** glibc detected *** free(): invalid pointer: 0x0000002a98ab1010 ***
>> 
>> which is coming from the destruction of itkImportImageContainer.
>> 
>> 
>> An e-mail that someone sent to this list in January about reading
>> Signa 5.x images yielded the response that the Insight Applications
>> program ConvertBetweenFileFormats does this, however it seems to me
>> that ConvertBetweenFileFormats uses GDCM to read DICOM series
>> images, but not Signa(?)
>> 
>> Furthermore, the tests included for GE4ImageIO only test reading of
>> a single slice, not a series, so I'm wondering if there is actually
>> a bug.
>> 
>> Here's the code excerpt.  ***NOTE: I've tried this both via factory
>> traversal and by specifying a GE4ImageIO "manually" and both
>> produce the same results.
>> 
>> I'd appreciate any suggestions and apologise if I did something
>> incredibly stupid.
>> 
>> 
>> 
>> Regards,
>> Neil
>> 
>> 
>> 
>> 
>>   itk::ObjectFactoryBase::RegisterFactory
>> ( itk::GEAdwImageIOFactory::New() );
>>   itk::ObjectFactoryBase::RegisterFactory
>> ( itk::GE4ImageIOFactory::New() );
>>   itk::ObjectFactoryBase::RegisterFactory
>> ( itk::GE5ImageIOFactory::New() );
>> 
>>   typedef unsigned short PixelType;
>>   typedef itk::Image< PixelType, 3> ImageType;
>>   typedef itk::NumericSeriesFileNames GeneratorType;
>>   typedef itk::ImageSeriesReader< ImageType > SeriesReaderType;
>> 
>>   GeneratorType::Pointer names = GeneratorType::New();
>>   names->SetStartIndex( start );
>>   names->SetIncrementIndex( inc ) ;
>>   names->SetEndIndex( end );
>>   names->SetSeriesFormat( inpPat.c_str() );
>> 
>>   SeriesReaderType::Pointer reader = SeriesReaderType::New();
>>   reader->SetFileNames( names->GetFileNames() );
>> 
>>   try
>>     {
>>     reader->SetDebug(true);
>>     reader->Update();
>> 
>> 
>> code trace:
>> 
>> #0  0x000000329cf2e21d in raise () from /lib64/tls/libc.so.6
>> (gdb) bt
>> #0  0x000000329cf2e21d in raise () from /lib64/tls/libc.so.6
>> #1  0x000000329cf2fa1e in abort () from /lib64/tls/libc.so.6
>> #2  0x000000329cf63291 in __libc_message () from /lib64/tls/libc.so.6
>> #3  0x000000329cf68eae in _int_free () from /lib64/tls/libc.so.6
>> #4  0x000000329cf691f6 in free () from /lib64/tls/libc.so.6
>> #5  0x000000329fcae20e in operator delete () from /usr/lib64/libstdc
>> ++.so.6
>> #6  0x0000000000441e15 in ~ImportImageContainer (this=0x659f70)
>>     at /opt/x86_64/pkgs/itk/3.2.0/gcc-release/include/
>> InsightToolkit/Common/itkImportImageContainer.txx:45
>> #7  0x0000002a967ecf29 in itk::Object::UnRegister ()
>>    from /opt/x86_64/pkgs/itk/3.2.0/gcc-release/lib/InsightToolkit/
>> libITKCommon.so.3.2
>> #8  0x00000000004324e4 in
>> itk::SmartPointer<itk::ImportImageContainer<unsigned long, unsigned
>> short>
>>> ::UnRegister (this=0x65bc20)
>>     at /opt/x86_64/pkgs/itk/3.2.0/gcc-release/include/
>> InsightToolkit/Common/itkSmartPointer.h:148
>> #9  0x0000000000432109 in ~SmartPointer (this=0x65bc20)
>>     at /opt/x86_64/pkgs/itk/3.2.0/gcc-release/include/
>> InsightToolkit/Common/itkSmartPointer.h:65
>> #10 0x0000000000440373 in ~Image (this=0x65ba60)
>>     at /opt/x86_64/pkgs/itk/3.2.0/gcc-release/include/
>> InsightToolkit/Common/itkImage.h:365
>> #11 0x0000002a967ecf29 in itk::Object::UnRegister ()
>>    from /opt/x86_64/pkgs/itk/3.2.0/gcc-release/lib/InsightToolkit/
>> libITKCommon.so.3.2
>> #12 0x0000002a967f8143 in itk::ProcessObject::~ProcessObject$base ()
>>    from /opt/x86_64/pkgs/itk/3.2.0/gcc-release/lib/InsightToolkit/
>> libITKCommon.so.3.2
>> #13 0x0000000000430351 in ~ImageSource (this=0x659db0)
>>     at /opt/x86_64/pkgs/itk/3.2.0/gcc-release/include/
>> InsightToolkit/Common/itkImageSource.h:178
>> #14 0x0000000000448e74 in ~ImageFileReader (this=0x659db0)
>>     at /opt/x86_64/pkgs/itk/3.2.0/gcc-release/include/
>> InsightToolkit/IO/itkImageFileReader.txx:46
>> #15 0x0000002a967ecf29 in itk::Object::UnRegister ()
>>    from /opt/x86_64/pkgs/itk/3.2.0/gcc-release/lib/InsightToolkit/
>> libITKCommon.so.3.2
>> #16 0x00000000004462b0 in
>> itk::SmartPointer<itk::ImageFileReader<itk::Image<unsigned short,
>> 3u>, itk::DefaultConvertPixelTraits<unsigned short> > >::UnRegister
>> (this=0x7fbfffecc0)
>>     at /opt/x86_64/pkgs/itk/3.2.0/gcc-release/include/
>> InsightToolkit/Common/itkSmartPointer.h:148
>> #17 0x0000000000444069 in ~SmartPointer (this=0x7fbfffecc0)
>>     at /opt/x86_64/pkgs/itk/3.2.0/gcc-release/include/
>> InsightToolkit/Common/itkSmartPointer.h:65
>> #18 0x000000000043e928 in
>> itk::ImageSeriesReader<itk::Image<unsigned short, 3u>
>>> ::GenerateData (this=0x6428a0)
>>     at /opt/x86_64/pkgs/itk/3.2.0/gcc-release/include/
>> InsightToolkit/IO/itkImageSeriesReader.txx:290
>> #19 0x0000002a967f9516 in itk::ProcessObject::UpdateOutputData ()
>>    from /opt/x86_64/pkgs/itk/3.2.0/gcc-release/lib/InsightToolkit/
>> libITKCommon.so.3.2
>> #20 0x0000000000428421 in main (argc=6, argv=0x7fbffff518)
>>     at /home/ch116471/projects/svnwrk/crkit.org/trunk/tools/code/
>> crlSlicesToVolume.cxx:80
>> 
>> _______________________________________________
>> Insight-users mailing list
>> Insight-users at itk.org
>> http://www.itk.org/mailman/listinfo/insight-users
> 
> _______________________________________________
> Insight-developers mailing list
> Insight-developers at itk.org
> http://www.itk.org/mailman/listinfo/insight-developers

More information about the Insight-developers mailing list