[Insight-users] Efficiently comparing segmentation results

[Neal R. Harvey]

Nick pointed me to that yesterday. I took a look, but it still doesn't 
provide the
efficiency that I was hoping for. I am hoping that the LabelMap stuff 
will be able
to help, and I am revisiting that, otherwise I guess I will have to 
write my own
function to do it.

Andriy Fedorov wrote:
> Neal, you can also take a look at this publication in Insight Journal:
>
> "Introducing Dice, Jaccard, and Other Label Overlap Measures To ITK",
> Tustison N., Gee J.
> http://www.insight-journal.org/browse/publication/707
>
>   
>> Message: 6
>> Date: Thu, 9 Dec 2010 11:28:29 -0500
>> From: Sajendra <sajendra at gmail.com>
>> Subject: Re: [Insight-users] Efficiently comparing segmentation
>>        results
>> To: "Neal R. Harvey" <harve at lanl.gov>
>> Cc: insight-users at itk.org
>> Message-ID:
>>        <AANLkTikXrc2uYOR1Q+V09UX=DShrp+4zZg==hwwQoDjS at mail.gmail.com>
>> Content-Type: text/plain; charset=ISO-8859-1
>>
>> You could perhaps use Kappa Statistic Metric:
>> http://www.itk.org/Doxygen/html/classitk_1_1KappaStatisticImageToImageMetric.html#_details
>>
>> It allows you to set the foreground value for the label you are
>> interested in so you could either run it multiple times, or probably
>> modify the filter to do exactly what it is you needed.
>>
>> On Wed, Dec 8, 2010 at 2:22 PM, Neal R. Harvey <harve at lanl.gov> wrote:
>>     
>>> I am trying to write some code with which I can efficiently compare
>>> segmentation results with
>>> a known ground truth. There are several papers that describe interesting
>>> metrics for doing this.
>>> A common theme is finding intersection between segments.
>>> There seems to be most of the functionality within ITK for doing this, but I
>>> haven't been able to
>>> figure out how to do it effectively/efficiently.
>>> I want to be able to go through each region in a segmented image (i.e. the
>>> pixels that have the same
>>> value "i" in that image) and calculate the number of pixels within that
>>> region that intersect with
>>> each region in a ground truth image (i.e. the pixels that have the same
>>> value "j" in that image).
>>> I don't want to have to go through every pixel in the image and test to see
>>> if it is value "i"
>>> or "j", but want to have some way of limiting my processing to only those
>>> pixels that have the
>>> required values.
>>> I am sure that this should be possible within ITK, I just haven't been able
>>> to figure it out.
>>> If anyone has any suggestions, it would be very much appreciated.
>>>
>>> Cheers
>>>
>>> Harve
>>>
>>> --
>>> HARVE (Neal R. Harvey) ? ? ? ? ? Email: harve at lanl.gov
>>> ISR-2 ? ? ? ? ? ? ? ? ? ? ? ? ? ?Phone: 505 667 9077
>>> Los Alamos National Laboratory ? Fax: 505 664 0362
>>> Los Alamos, NM 87545, USA ? ? ? ?Mail Stop: D-436
>>> The generation of random numbers is too important to be left to chance.
>>>
>>> _____________________________________
>>> Powered by www.kitware.com
>>>
>>> Visit other Kitware open-source projects at
>>> http://www.kitware.com/opensource/opensource.html
>>>
>>> Kitware offers ITK Training Courses, for more information visit:
>>> http://www.kitware.com/products/protraining.html
>>>
>>> Please keep messages on-topic and check the ITK FAQ at:
>>> http://www.itk.org/Wiki/ITK_FAQ
>>>
>>> Follow this link to subscribe/unsubscribe:
>>> http://www.itk.org/mailman/listinfo/insight-users
>>>
>>>       
>> ------------------------------
>>
>> _______________________________________________
>> Insight-users mailing list
>> Insight-users at itk.org
>> http://www.itk.org/mailman/listinfo/insight-users
>>
>>
>> End of Insight-users Digest, Vol 80, Issue 15
>> *********************************************
>>
>>     


-- 
HARVE (Neal R. Harvey)           Email: harve at lanl.gov
ISR-2                            Phone: 505 667 9077
Los Alamos National Laboratory   Fax: 505 664 0362
Los Alamos, NM 87545, USA        Mail Stop: D-436
The generation of random numbers is too important to be left to chance.