Acquisition numbering in ISMRMRD

[kor]

I have a few doubts about numbering of acquisition in ISMRMRD dataset.
Is the acquisition number same as first phase encoding step number?

How to calculate the acquisition number in below example scenarios,
1)If I have a k-space data of five consecutive slices of size 512 x  256, created with 2d pulse sequence
 Kspace2d(512, 256, 5), then how many total acquisitions will be there. (is it 256 x 5 )

2) if have a k-space data of five slices of size 512 x  256, created with 3d pulse sequence
 Kspace3d(512, 256, 5), then how many total acquisitions will be there. (is it 256 )

3)if have a k-space data of five slices of size 512 x  256 read with 10 coils, created with 3d pulse sequence
 Kspace3d(512, 256, 5, 10), then how many total acquisitions will be there. (is it 256 )

4) 3)if have a k-space data of five slices of size 512 x  256 read with 10 coils and 3 echos, created with 3d pulse sequence

How to calculate the acquisition number in case of a k-space dataset with combinations of multiple repetition, multi channel, multiple echo, multiple slices (3D) case scenarios?


Also,  what is the difference between the fields in encoding counter
kspace_escode_step_2 and slices
average and repetitions

Is kspace_escode_step_1 and number of first phase encoding steps same?

[Michael Hansen]

Please read the paper describing the data format, which will give you most of your answers:

http://onlinelibrary.wiley.com/doi/10.1002/mrm.26089/abstract

Here are some answers:

"Is the acquisition number same as first phase encoding step number?"

In general no. The acquisition number is just a sequential counter of the acquisitions. If it is a 2D acquisition with a single frame, they could be the same, but it is accidental. 

"How to calculate the acquisition number in below example scenarios,"

In general this statement hides some potential misunderstandings. The format does not guarantee that you can calculate the number of acquisitions. In some case you can get some idea of how many to expect, but the acquisition could be undersampled in some way and it may not be possible to calculate. It could also be a free running sequence (golden angle or something) where such calculations don't make sense. 

"1)If I have a k-space data of five consecutive slices of size 512 x  256, created with 2d pulse sequence 
 Kspace2d(512, 256, 5), then how many total acquisitions will be there. (is it 256 x 5 )"

Again, it depends on a number of things: is it fully sampled or undersampled? Are you using partial fourier acquisition? Do you have reduced resolution in the phase encoding direction?, etc., etc. Do get a clearer idea you have to inspect the encoded matrix size of the header and the encoding limits. Please refer to the paper for a description of these or look at the examples in the code. However, if it is fully sampled, then yes, it would be 256x5.

"2) if have a k-space data of five slices of size 512 x  256, created with 3d pulse sequence 
 Kspace3d(512, 256, 5), then how many total acquisitions will be there. (is it 256 )"

Again, it depends on a number of things. Do you have oversampling in the slice direction (as is commonly the case in 3D), and all the concerns above. You need to consult the encoded matrix size and encoding limits of the data.  However, assuming no oversampling in the slice direction and fully sampled in-plane, then there would be 256x5 profiles. 

"3)if have a k-space data of five slices of size 512 x  256 read with 10 coils, created with 3d pulse sequence 
 Kspace3d(512, 256, 5, 10), then how many total acquisitions will be there. (is it 256 )"

Number of coils does not increase the number of acquisitions. An acquisition object is a multi-coil object. So it would be 256x5 again. (taken all the points above into consideration).

"4) 3)if have a k-space data of five slices of size 512 x  256 read with 10 coils and 3 echos, created with 3d pulse sequence"

There would probably be 256x5x3 acquisitions. But again, there could be a lot of other considerations at work here and more details would be needed. 

"How to calculate the acquisition number in case of a k-space dataset with combinations of multiple repetition, multi channel, multiple echo, multiple slices (3D) case scenarios?"

Again, it depends. All these questions are variations of the same question to which there may not be an answer. But you need to look at the encoding limits of those dimensions and take undersampling into considerations. But assuming full sampling, you would have

number of encoding step 1 lines X number of encoding step 2 lines x number of echos (contrasts) x number of repetitions

Or something like that. 

"Also,  what is the difference between the fields in encoding counter
kspace_escode_step_2 and slices
average and repetitions

Is kspace_escode_step_1 and number of first phase encoding steps same? "

Please read the paper. But:

kspace_encode_step_1 = ky

kspace_encode_step_2 = kz

slice = slice/slab(3D)

repetition = repetion of experiment (typically time point)

average = average (usually line averaging)

Again, please, consult the paper.

Hope this helps,

Michael