Correlation Coefficient on Stacks?

[Jacob Keller]

Dear List,

is there a way to calculate correlation coefficients between all images in
a stack, in my case to evaluate quality of registration? I am open to other
scoring methods for registration as well if this is misguided or there is a
better way to do it...

Jacob

--
*******************************************
Jacob Pearson Keller, PhD
Postdoctoral Associate
HHMI Janelia Farms Research Campus
email: [hidden email]
*******************************************

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[GCH-2]

Try the Image CorrelationJ plugin.

http://www.gcsca.net/IJ/ImageCorrelationJ.html

Gary.

Sent from my iPhone

On 25. jan. 2013, at 18:52, Jacob Keller <[hidden email]> wrote:

> Dear List,
>
> is there a way to calculate correlation coefficients between all images in
> a stack, in my case to evaluate quality of registration? I am open to other
> scoring methods for registration as well if this is misguided or there is a
> better way to do it...
>
> Jacob
>
> --
> *******************************************
> Jacob Pearson Keller, PhD
> Postdoctoral Associate
> HHMI Janelia Farms Research Campus
> email: [hidden email]
> *******************************************
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Jacob Keller]

Let me clarify my question: is there a capacity in imagej simply to
calculate the correlation coefficient between two images, i.e., between all
pixels x in image1 with pixels y in image2? Seems really simple, but I just
can't find where it is. Seems like it should be part of the image
calculator function, where one can perform simple operations between two
images, but the list there is limited. Couldn't the list be expanded simply
enough to include CC and some other operations, as a basis for a lot of
possible macros? Where does one find the code for the image calculator
itself--maybe I can try to hack it?

Jacob

On Sat, Jan 26, 2013 at 11:29 AM, GCH <[hidden email]> wrote:

> Try the Image CorrelationJ plugin.
>
> http://www.gcsca.net/IJ/ImageCorrelationJ.html
>
> Gary.
>
> Sent from my iPhone
>
> On 25. jan. 2013, at 18:52, Jacob Keller <[hidden email]>
> wrote:
>
> > Dear List,
> >
> > is there a way to calculate correlation coefficients between all images
> in
> > a stack, in my case to evaluate quality of registration? I am open to
> other
> > scoring methods for registration as well if this is misguided or there
> is a
> > better way to do it...
> >
> > Jacob
> >
> > --
> > *******************************************
> > Jacob Pearson Keller, PhD
> > Postdoctoral Associate
> > HHMI Janelia Farms Research Campus
> > email: [hidden email]
> > *******************************************
> >
> > --
> > ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
*******************************************
Jacob Pearson Keller, PhD
Postdoctoral Associate
HHMI Janelia Farms Research Campus
email: [hidden email]
*******************************************

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Herbie-2]

Jacob,

the (non-normalized) cross-correlation coefficient of two signals is
defined as the integral of their product. That said, you need to
multiply the two images and then take the sum of the values of all
pixels of the product image. Of course this is best done with 32 bit images.

0. step:
Go to "Set Measurements" and check "Integrated Density"

1. step:
Use image calculator to multiply the images.

2. step:
Use "Measure" and the cross-correlation coefficient shows up in the
results table.

HTH

Herbie

_______________________________
On 28.01.13 04:37, Jacob Keller wrote:

> Let me clarify my question: is there a capacity in imagej simply to
> calculate the correlation coefficient between two images, i.e., between all
> pixels x in image1 with pixels y in image2? Seems really simple, but I just
> can't find where it is. Seems like it should be part of the image
> calculator function, where one can perform simple operations between two
> images, but the list there is limited. Couldn't the list be expanded simply
> enough to include CC and some other operations, as a basis for a lot of
> possible macros? Where does one find the code for the image calculator
> itself--maybe I can try to hack it?
>
> Jacob
>
> On Sat, Jan 26, 2013 at 11:29 AM, GCH<[hidden email]>  wrote:
>
>> Try the Image CorrelationJ plugin.
>>
>> http://www.gcsca.net/IJ/ImageCorrelationJ.html
>>
>> Gary.
>>
>> Sent from my iPhone
>>
>> On 25. jan. 2013, at 18:52, Jacob Keller<[hidden email]>
>> wrote:
>>
>>> Dear List,
>>>
>>> is there a way to calculate correlation coefficients between all images
>> in
>>> a stack, in my case to evaluate quality of registration? I am open to
>> other
>>> scoring methods for registration as well if this is misguided or there
>> is a
>>> better way to do it...
>>>
>>> Jacob
>>>
>>> --
>>> *******************************************
>>> Jacob Pearson Keller, PhD
>>> Postdoctoral Associate
>>> HHMI Janelia Farms Research Campus
>>> email: [hidden email]
>>> *******************************************
>>>
>>> --
>>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>>
>> --
>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>>
>
>
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[ctrueden]

Hi Jacob,

> Seems like it should be part of the image calculator function

Good idea. We have added a new type of plugin (CalculatorOp) to ImageJ2
which can be used to extend the operations available to the Image
Calculator.

Interested developers can find the code at: http://git.io/etSltA

Unfortunately, I did not have time to implement correlation coefficient as
an operation, but the point is that now it should be easier for anyone else
to do so, should they feel so inclined, without modifying the core source
code.

Regards,
Curtis


On Sun, Jan 27, 2013 at 9:37 PM, Jacob Keller <
[hidden email]> wrote:

> Let me clarify my question: is there a capacity in imagej simply to
> calculate the correlation coefficient between two images, i.e., between all
> pixels x in image1 with pixels y in image2? Seems really simple, but I just
> can't find where it is. Seems like it should be part of the image
> calculator function, where one can perform simple operations between two
> images, but the list there is limited. Couldn't the list be expanded simply
> enough to include CC and some other operations, as a basis for a lot of
> possible macros? Where does one find the code for the image calculator
> itself--maybe I can try to hack it?
>
> Jacob
>
> On Sat, Jan 26, 2013 at 11:29 AM, GCH <[hidden email]> wrote:
>
> > Try the Image CorrelationJ plugin.
> >
> > http://www.gcsca.net/IJ/ImageCorrelationJ.html
> >
> > Gary.
> >
> > Sent from my iPhone
> >
> > On 25. jan. 2013, at 18:52, Jacob Keller <[hidden email]
> >
> > wrote:
> >
> > > Dear List,
> > >
> > > is there a way to calculate correlation coefficients between all images
> > in
> > > a stack, in my case to evaluate quality of registration? I am open to
> > other
> > > scoring methods for registration as well if this is misguided or there
> > is a
> > > better way to do it...
> > >
> > > Jacob
> > >
> > > --
> > > *******************************************
> > > Jacob Pearson Keller, PhD
> > > Postdoctoral Associate
> > > HHMI Janelia Farms Research Campus
> > > email: [hidden email]
> > > *******************************************
> > >
> > > --
> > > ImageJ mailing list: http://imagej.nih.gov/ij/list.html
> >
> > --
> > ImageJ mailing list: http://imagej.nih.gov/ij/list.html
> >
>
>
>
> --
> *******************************************
> Jacob Pearson Keller, PhD
> Postdoctoral Associate
> HHMI Janelia Farms Research Campus
> email: [hidden email]
> *******************************************
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html