If a 3D+t image is imported from Chili into the MITK Plug-In the image and all its time steps where shown correctly in MITK. But if you save it as .pic file the saved file as well as the current data in the data tree get some black timesteps. That means some time steps only contain 0 pixel values instead of the image. The destroyed timesteps are not always the same, sometimes are more timesteps affected, sometimes less.
Test data e.g. Bernhauer, Tamara in pacs.
Cannot reproduce this bug. I tried "Seitel, Alexander", image no. 13. This is an image containing 400 time steps.
Something is wrong here with the image geometry, because "reinit" does not work -- "global reinit" DOES work. However, saving a PIC file and loading it does NOT produce any "black" time steps.
Can you find a patient/study/series (all information) and can you describe how you saved and loaded the image?
BTW this is probably not a SDK problem but something with saving images?!
Test patient is Keil,Oona in Chili.
Import from Chili and Save in Datamanager. Also the data in the datatree gets corrupted time steps. This can be seen by sliding through the timesteps with the time step slider.
Discussion with Ivo resulted in:
-> Possibly an update problem; maybe the Chili import reader is still connected to the image when writing? Or can block or reuse memory areas?
More details:
Severe bug in mitkImage. When loading data slice by slice, it is composed to greater units in mitkImage when necessary. When using GetChannelData(), after the creation of the first volume, ALL references to the imported slices are set to zero. ->All time steps above one that have not previously created are cleared.
Committed to 0.12.2 repository in Rev. 18329