[vtkusers] help for rendering a large amount of images

[David Gobbi]

Hi Ang,

If I was going to use memory mapping to read a DICOM file, then I wouldn't
use vtkDICOMReader.  The whole point of memory mapping is to _avoid_ using
a traditional file reader.

Instead, I would use vtkDICOMParser to find out where in the file the
PixelData starts, and then I would directly memory-map the PixelData from
disk.  This requires the use of uncompressed DICOM images, of course.

Or, if you need to map multiple files as one contiguous block of memory
(e.g. for vtkImageReslice), then you can do what I suggested earlier in
this thread: combine the PixelData from all those files into one large flat
file, and then memory-map the new file.

 - David


On Tue, Apr 11, 2017 at 9:05 PM, Ang <ysa0829 at gmail.com> wrote:

> Hi David,
>
> Could you give me advanced tips for memory mapping file ?
>
> in my understanding, the process is
>
> vtkDICOMReader(retrieve pixel data)  =>  CreateFileMapping => MapViewOfFile
> => put the pixel data into  File => OpenFileMapping => MapViewOfFile =>
> vtkImageImport => vtkImageReslice
>
> Is it correct?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/vtkusers/attachments/20170411/cbbf9ff2/attachment.html>