Area 51 » Science

125

followers

Follow It!

Announcements

If you're interested in this proposal going forward, please vote on the questions below and/or add your own. – jonsca Jun 16 '12 at 13:22

35 Example Questions

active oldest votes

up vote 22 down vote

added by Vladimir Feb 1 '12 at 16:51

I used four manually labeled, human brain MRI datasets in an evaluation study and have links to the them here: mindboggle.info/papers/… – Arno Klein Feb 2 '12 at 23:57

And there are MANY unlabeled brain MRI volumes available!: nitrc.org/projects/fcon_1000, oasis-brains.org, nitrc.org/projects/multimodal – Arno Klein Feb 3 '12 at 0:05

Data from 101 manually labeled brain image volumes and surfaces will be released soon as part of the Mindboggle project: mindboggle.info – Arno Klein Feb 3 '12 at 0:09

Both NIF (www.neuinfo.org) broadly for all of neuroscience and NITRC (www.nitrc.org) offer listings of available data. Searching NITRC for 'data resources' (nitrc.org/search/…) generates 45 data sources, including the ones nicely identified by Arno Klein in a prior comment. – dnkennedy Feb 3 '12 at 20:39

Check out brain-development.org for a variety of resources. – soundray Sep 17 '12 at 21:10

show 1 more comment

up vote 17 down vote

added by Vladimir Feb 1 '12 at 20:14

There is a recent article from the LONI group (UCLA) that includes a review of common formats: frontiersin.org/Neuroinformatics/10.3389/… (Frontiers in Neuroinformatics, March 2012) – Raphael Sep 27 '12 at 13:44

up vote 14 down vote

added by Isaiah Feb 1 '12 at 17:39

I think the first question to ask is what kind of large imaging study are you trying to support? I'm guessing that this is for DICOM data, no? XNAT is the system I am most familiar with and most likely to recommend for a neuroimaging study, but NBIA might be better if you are working with an oncology study and the sites are already familiar with it. – Nolan Nichols Feb 3 '12 at 0:12

up vote 10 down vote

added by satra Jan 31 '12 at 3:11

	e.g. a single subject fMRI dataset - 100MB - 2GB (uncompressed) what about mouse, rat, monkey, eeg, etc.,.? – satra Feb 3 '12 at 19:53
	Single subject MEG recording ( ~ 10-25 minutes) on a CTF 275 channel systems are typically 200-500 Mb – Ian Feb 16 '12 at 9:36
	anybody with estimates on microscopy datasets? – satra Feb 26 '12 at 15:54
	1 cm^2 scanned with a zeiss mirax at 40x and stored as a pyramidal virtual slide with 7 layers is about 130-150MB. – Isaiah Apr 30 '12 at 16:09

up vote 10 down vote

added by Vladimir Feb 1 '12 at 16:54

One can always try checking NITRC (www.nitrc.com); and get a listing of all resources that claim to do statistical analysis (nitrc.org/search/…) – dnkennedy Feb 3 '12 at 20:30

For a complete preprocessing stream that handles resting-state fMRI data you can check out the scripts that are part of the 1000 Functional Connectomes Project: fcon_1000.projects.nitrc.org – Maarten Feb 8 '12 at 9:59

There are a growing number of neuroimaging tools for fMRI analysis in R. R is an open source project so no expensive license to purchase such as for matlab. Still not mature enough for a comprehensive neuroimaging platform. – Ian Feb 16 '12 at 9:33

up vote 10 down vote

added by Isaiah Feb 1 '12 at 17:43

this looks like an awesome tools questions. These sort of questions are on-topic on cogsci.SE as are all other neuroscience and related tools questions. We are trying to test how receptive the community is to neuroinformatics questions. Do you have the time to write up a full question on the topic there? – Artem Kaznatcheev May 14 '12 at 21:09

up vote 10 down vote

added by Isaiah, edited by WikiSpeedia hang-around Sep 16 '12 at 18:41

	You should check out ANTS and Gang Song one of the developers has definitely done slice co-registration to build 3d mouse models. picsl.upenn.edu/ANTS – satra Feb 1 '12 at 20:55
	TrakEM2 is your friend: ini.uzh.ch/~acardona/trakem2_manual.html – unidesigner Feb 5 '12 at 13:57

up vote 10 down vote

added by Ailton Andrade de Oliveira Feb 1 '12 at 19:33

Well, any registration algorithm can be used to transform anatomical labels to an unlabeled target brain image. I performed several evaluation studies evaluating registration-based neuroanatomical labeling. Please see mindboggle.info/papers/… and mindboggle.info/papers/… – Arno Klein Feb 3 '12 at 0:58

In a few months we'll be releasing the new Mindboggle software to automate cortical labeling: mindboggle.info – Arno Klein Feb 3 '12 at 1:01

Sorry to beat the NITRC drum, but, hey, that's what I do... Searching NITRC for resources that claim to do 'segmentation' generates 48 answers (including mindboggle!) (nitrc.org/search/…) It would take a little more manual human intervention to resolve which of these offer an automated approach, and which support 'whole brain' analysis, but it seems like a good start... – dnkennedy Feb 3 '12 at 20:33

up vote 10 down vote

added by memyself Feb 2 '12 at 10:51

up vote 10 down vote

added by unidesigner Feb 2 '12 at 15:18

If you need speed, have a look at Nest (nest-initiative.org). The other option would be brian simulator (briansimulator.org/) – deniz Mar 5 at 12:44

up vote 10 down vote

added by dnkennedy Feb 2 '12 at 15:58

I think it can on a global level. If you have for example total brain volume (structural MRI) and mean Franctional Anisotropy (Diffusion Tensor Imaging), you could maybe create a compound score. I don't know what that would gain you though. Another thing that comes to mind which is not exactly pooling is the use of combined EEG and fMRI to have both high spatial as well as temporal resolution. – Vincent Apr 6 at 19:14

up vote 10 down vote

added by unidesigner Feb 4 '12 at 8:59

This is also a good question for the proposed community "Digital Preservation". – Jennifer Davis Apr 25 '12 at 4:33

This would be a good question to ask on the existing Academia.SE since it is not neuroinformatics specific. – Artem Kaznatcheev Jan 21 at 0:40

up vote 10 down vote

added by memyself, edited by WikiSpeedia hang-around Sep 16 '12 at 18:40

up vote 10 down vote

added by satra, edited by WikiSpeedia hang-around Sep 16 '12 at 18:40

The human brain consumes between 20-25 W (hypertextbook.com/facts/2001/JacquelineLing.shtml). A Pentium 4 2.8 GHz consumes up to 85 W, but some low power CPUs such as the IXC1100 can consume as little as 2.4 W – Yann Apr 2 '12 at 16:58

up vote 10 down vote

added by tdc, edited by WikiSpeedia hang-around Sep 16 '12 at 18:40

see nipy.org. within that umbrella we have been developing nipype which is a python framework for integrating analyses across existing tools. also see pymvpa.org for machine learning tools applied to neuroimaging data – satra Feb 26 '12 at 15:58

@satra thanks, will take a look – tdc Feb 27 '12 at 8:31

up vote 10 down vote

added by Nate Apr 27 '12 at 15:28

up vote 10 down vote

added by Nate Apr 27 '12 at 15:35

up vote 10 down vote

added by Nolan Nichols May 2 '12 at 20:49

cocomac.g-node.org – satra Sep 13 '12 at 9:59

up vote 9 down vote

added by Nolan Nichols Feb 9 '12 at 4:38

up vote 7 down vote

added by jonsca Jun 18 '12 at 9:32

up vote 4 down vote

added by btel Oct 25 '12 at 13:07

up vote 4 down vote

added by btel Oct 25 '12 at 13:20

take a look at ipython.org. they recently received funding from the sloan foundation to extend the notebook in many different directions. you can also see how such a notebook can be useful nbviewer.ipython.org. – satra Mar 16 at 12:42

up vote 3 down vote

added by jonsca Jun 18 '12 at 9:30

up vote 3 down vote

added by jonsca Jun 18 '12 at 9:32

up vote 2 down vote

added by unidesigner Feb 4 '12 at 9:16

up vote 2 down vote

added by satra, edited by WikiSpeedia hang-around Sep 16 '12 at 18:39

	e.g., iRODS (+iDrop) - see: irods.org – satra Feb 26 '12 at 16:01
	amazon S3 - aws.amazon.com/s3 – Nolan Nichols May 2 '12 at 22:04

up vote 2 down vote

added by jonsca Jun 18 '12 at 9:36

up vote 2 down vote

added by btel Oct 25 '12 at 13:19

up vote 1 down vote

added by unidesigner Feb 2 '12 at 15:03

up vote 1 down vote

added by unidesigner Feb 4 '12 at 9:06

Is this a neuroscience question or a neuroinformatics question? – Matt Feb 6 '12 at 22:19

True, it is a neuroscience question. – unidesigner Feb 11 '12 at 17:26

but if there was a database describing those tradoffs, i'd argue that would be a neuroinformatics tool – Nolan Nichols May 2 '12 at 20:51

This question would be on-topic at cogsci.SE. That site is in public beta, so feel free to post it there as a full question. – Artem Kaznatcheev May 14 '12 at 21:13

up vote 1 down vote

added by Chris Filo Gorgolewski Mar 19 '12 at 10:27

up vote 0 down vote

added by Chris Filo Gorgolewski May 2 '12 at 13:44

up vote 0 down vote

added by mih Sep 16 '12 at 7:26

from my experience, there has been two main blockers: – Pierre Bellec Sep 21 '12 at 20:31

(1) legacy. Analyze was sort of a de facto standard at the end of the 90s, and anything different had a hard time emerging. (2) although MINC is supposed to be just a specific incarnation of HDF5, the MINC developer community has consistently pushed its own C-based API rather than using the HDF5 libraries native in python/matlab/R/etc. The consequence is that MINC interfaces with various languages are long to develop, difficult to install, and not always actively maintained. – Pierre Bellec Sep 21 '12 at 20:43