Storing of coil sensitivity data within the ISMRMRD raw data h5 file

[kor]

How can store coil sensitivity matrix data and its associated details(like matrix size) within the same ISMRMRD dataset (H5 file) where k-space data is stored?
Have any inbuilt mechanism in ISMRMRD for it?
Also, can we include the image data within the same dataset (H5 file) where k-space data is stored?
I'm working in Matlab environment, I couldn't find  any interface for image data operations in Dataset class definition in ISMRMRD Matlab library, but in online API reference document, details of image operation functions like "ismrmrd_append_image" are mentioned in Dataset class.
I expect,  they are included in the ISMRMRD C/C++  library implementation only.

[Michael Hansen]

Couple of comments:

1. If you are storing coil sensitivity maps as images in the files (which is possible, see below), you are probably doing it wrong. I know this is how the Philips users like to think about things, but the coil sensitivity maps came from some raw data at some point, were reconstructed and then used in the reconstruction. If you follow the ISMRMRD philosophy, you should store the raw data for the sensitivity maps in a separate ISMRMRD file and used the dependency sections of the XML header to indicate in subsequent files that it needs sensitivity information from a different file. The point here is that sensitivity maps don't fall from the sky, they are not "available" on the scanner through any other means than reconstruction of acquisition data. 

2. You can store images. There are two approaches. You can just store some arrays of data using the generic HDF5 arrays using the standard HDF5 API. You can put this data anywhere in the file. Alternatively you can use the ISMRMRD images. You are right, the Matlab API does not at present support this image format at the moment, but the C/C++-api does. It should be pretty trivial to either add the image functions to the Matlab API, we will add it to the list to do that at some point. 

Anyways, my point is that adding sensitivity maps directly to the ISMRMRD file is a bit of a hack. The right thing to do is to add the raw data for the sensitivity maps and reconstruct them and then use the dependency fields. 

Hope this helps. 

Michael