NIfTI: — Neuroimaging Informatics Technology Initiative

phorum - NIfTI Message Board - Re: Size limitations in NIfTI

NIfTI Message Board

New Topic | Go to Top | Go to Topic | Search

Re: Size limitations in NIfTI

Author: Cinly OOI (---.medschl.cam.ac.uk)
Date: 03-19-09 15:16

After reading the whole email, I do not think you have any way around this problem. What you are describing is a sparse matrix problem and people pay good money to be able to store them properly. Not to mention storing sparse matrix efficiently (memory and disk) and quick access to the elements are subject of active research in computer sciences.

Certainly give GifTI a look. But I don't think it will help you. Zipping data up only take care of disk efficiency problem, at the expense of longer read time since unzipping is necessary and so far I cannot find any algorithms that can jump to specific section and unzip that section only. Most has to start from the beginning.

If you really want to use NifTI, then I suggest this:
Choose a sparse matrix implementation for your programming language of choice. Find one that can actually read/write itself to file. The purpose here is to eliminate the need to code the mask for the sparse matrix yourselves and use a specialist implementation that is more efficient (use less disk space, squeeze more data into memory, faster element access)

Then derive a new "intent_code". The nim->data will read/write whatever is your sparse matrix implementation read/write to the file.

I acknowledge that most sparse matrix implementation is for 2D and we need 3D or 4D data. The way around it is to write an intermediate layer that translate your 3D/4D data into a 2D storage array automatically.

If this is the case, use nim->dim[] to store the dimension information, but use intent_p[123] to store the dimension of your 2D sparse matrix.

HTH
Cinly

Reply To This Message | Flat View

Newer Topic | Older Topic

Topics