Please pull SieXA10_Phi5dMRI into master #145
Conversation
(esp. for enhanced and unenhanced DICOM from Philips 2-5.)
…ieXA10_Phi5dMRI Conflicts: README.md
|
Rob- |
|
@yarikoptic / @chrisfilo / @jonclayden - could you test this version on some of your data. Rob's fantastic changes should allow it to support Siemens XA10 (at the moment only on the Vida) and new Philips 5.x features. However, lets make sure there are no unintended consequences. It handles my QA fine, and compiles seamlessly with other projects. By the way, for any sites that have a Siemens Vida: at the moment it is critical that you do not save fMRI and DTI data as "mosaics" on this platform. The mosaics created by XA10 are not valid DICOMs (they are missing required type 1 values like patient position). This has been reported to Siemens. For the moment, stick with the non-mosaic output. As a to-do: there are no CSA headers, so at some point we will need to decipher slice timing, effective readout time, etc. @mharms - any chance you would want to look into this? Since this is such a big patch, I have generated a new release v1.0.20171204 that consolidates all previous minor tweaks. |
Hello,
This branch was triggered by Siemens changing their DICOM format in the switch from VE11 to XA10. XA10 is currently only found on their Vida scanners, but when I asked Siemens about it they confirmed that XA10 is the next software version after VE11, and they are switching from mosaic + CSA style DICOM to enhanced DICOM. The new enhanced DICOM is a move in the direction of using more public tags for diffusion info, like Philips, but unlike Philips 4D series are still receiving 1 file per volume. Those files are enhanced, however, not mosaics. The Vida also has a "compatibility" mode, but we have not been able to extract diffusion b values or directions from it, and this branch only supports the enhanced DICOM mode.
It also includes some updates to improve dcm2niix's handling of diffusion MRI in either enhanced or unenhanced DICOM from Philips scanners, especially ones running the current software version (5). Notes have been added to clarify that the "ADC" volume commonly found in Philips diffusion MRI DICOM is likely really a trace volume, i.e. the average of the diffusion weighted volumes. (Trace volumes do not appear to be made by Philips if there are multiple b values.)
This branch passes the dcm_qa test script if compiled with OpenJPEG, and passes my own test suite that includes enhanced DICOM from Siemens XA10 and Philips 5, and not-enhanced DICOM from various software versions of GE, Philips, and Siemens MR scanners, for various sequences such as MPRAGE, FLAIR, diffusion, and functional MRI.