This part of your call:
, 'readdata', false);
is telling ft_read_cifti to not load the data matrix, that is why it isn't
there.
If you are just as comfortable in C++, then yes, CiftiLib is likely the
best solution, it is directly based on the code we use in connectome
workbench for cifti files. If what you need to do with the dconn organizes
well with the on-disk structure of the matrix (i.e., not needing more than
a few rows in memory at once, never needing to read columns), then it will
work efficiently with almost no memory footprint (as long as you don't tell
it to read the entire file into memory).
I don't think we distribute parcellated versions of the connectome yet, as
there hasn't exactly been a consensus on a parcellation for the entire
cortex. You can use -cifti-parcellate to apply a parcellation to the
dconn, if you have one that you favor.
Tim
On Tue, Nov 24, 2015 at 5:45 AM, David Dalmazzo <davm...@gmail.com> wrote:
> Hello,
> With fieldTrip reading using this line
> mycifti = ft_read_cifti(
> '/Users/David/Desktop/HumanConnectomeProject/HCP_S500_R468_rfMRI_MIGP_d4500ROW_zcorr/HCP_S500_R468_MIGPd4500ROW.dconn.nii',
> 'readdata', false);
> I get: dimond (pos_pos), hdr (1x1 struct), unit ('mm'), brain structure,
> brainstructurelabel, dim, pos, transform. But the connectivity matrix is
> not there.
>
> I'm going to check the C++ library. It seems to be the solution.
> Seems to be very difficult to work with this 33GB file. Is there in HCP
> datasets any other file with connectome information, that have a
> connectivity matrix or an array of region's connections? The Hagmann
> version seems to be very old for the models we need to implement.
>
> Thanks for the help
> David
>
>
>
> On Mon, Nov 23, 2015 at 9:14 PM, Rose Tharail John <
> mailboxofr...@gmail.com> wrote:
>
>> Thank you!
>>
>> On Mon, Nov 23, 2015 at 3:03 PM, Timothy Coalson <tsc...@mst.edu> wrote:
>>
>>> To get a dconn loaded in matlab, you of course need to have a large
>>> amount of available memory. Freezing is what I would expect if you don't
>>> have enough memory, and start using a lot of swap space (some ways of
>>> loading might initially load it as double precision, and need twice the
>>> memory).
>>>
>>> As you mention specifically c++ as your end goal, have a look at
>>> CiftiLib, as it is a c++ library for reading and writing cifti files, and
>>> has better support and features than currently available matlab cifti
>>> reading/writing (for instance, on disk reading/writing (not reading the
>>> entire file into memory first), easy construction and full support of
>>> dimension mappings of all supported types):
>>>
>>> https://github.com/Washington-University/CiftiLib
>>>
>>> Tim
>>>
>>>
>>> On Mon, Nov 23, 2015 at 11:18 AM, David Dalmazzo <davm...@gmail.com>
>>> wrote:
>>>
>>>> Hello,
>>>> I work in Specs Lab in Pompeu Fabra University as a Phd student.
>>>> I'm building an app for connectome visualisation and brain activity
>>>> simulation called BrainX3. The first version use Hagmann dataset based on
>>>> 998 nodes and ~14.000 bidirectional connections.
>>>>
>>>> For the new version I would like to use HCP S500 dataset. My main
>>>> problem is that following this two options about How to get CIFTI files
>>>> into MATLAB, I don't find the solution. The first way using fieldTrip gives
>>>> me some errors, but I'm already in contact with Robert Oostenveld who is
>>>> helping me. And the second option using Gifti+wb_command, just crash/freeze
>>>> my computer.
>>>>
>>>>
>>>> https://wiki.humanconnectome.org/display/PublicData/HCP+Users+FAQ#HCPUsersFAQ-2.HowdoyougetCIFTIfilesintoMATLAB
>>>> ?
>>>>
>>>> My question is how is the best way to extract the data, maybe I'm just
>>>> missing a tutorial where it's well explained?
>>>> My main purpose is to build an app in C++ of connectome visualization
>>>> with much more resolution than Hagmann's.
>>>>
>>>> Right now I'm in OS X 10.11.1 and Matlab R2015b (8.6.0.267246) 64-bit.
>>>>
>>>> Thanks for the support,
>>>>
>>>> David
>>>>
>>>> _______________________________________________
>>>> HCP-Users mailing list
>>>> HCP-Users@humanconnectome.org
>>>> http://lists.humanconnectome.org/mailman/listinfo/hcp-users
>>>>
>>>
>>> _______________________________________________
>>> HCP-Users mailing list
>>> HCP-Users@humanconnectome.org
>>> http://lists.humanconnectome.org/mailman/listinfo/hcp-users
>>>
>>
>>
>
_______________________________________________
HCP-Users mailing list
HCP-Users@humanconnectome.org
http://lists.humanconnectome.org/mailman/listinfo/hcp-users
