JIRA [IMG-1520] Use curvelets to measure add'l features

[thouis]

Original reporter: mbray Jira link: https://jira.broadinstitute.org/browse/IMG-1520 Fix version: future Assignee: leek

Comment from mbray on Thu, 8 Sep 2011 09:43:18 -0400: On Thu, Sep 8, 2011 at 8:18 AM, Mark Bray<mbray@broadinstitute.org> wrote:
> Re: curvelets - When I was at LOCI, Kevin introduced me to a student who
> worked on using curvelets for extracting features from collagen fibers
> (http://www.loci.wisc.edu/software/curvelet-based-alignment-analysis). I
> was impressed by the results, however, her program is MATLAB-based.
> After talking to Vebjorn briefly about it, I found a licensed Python
> implementation:https://slimweb.eos.ubc.ca/~hegilles/software.html,but
> never followed up on it.
> -Mark
>
> On 9/7/2011 5:41 PM, Anne Carpenter wrote:
>> Zhang, B.& Pham, T. Phenotype Recognition with Combined Features and
>> Random Subspace Classifier Ensemble. BMC Bioinformatics 12, 128
>> (2011).http://www.ncbi.nlm.nih.gov/pubmed/21529372
>>
>> ----> Uses Ilya Goldberg lab's WND CHARM benchmark sets to test a
>> method that similarly avoids segmentation and instead uses curvelets
>> in combination with Haralick texture features for phenotype
>> classification, which is done using the random subspace (RS) ensemble
>> method. Seems to work better than original features used by WND-CHARM.
>> Might be interesting for those interested in curvelets.

[AnneCarpenter]

See also #103

[dlogan]

Closing now, for future consideration.