info on directionality plugin

classic Classic list List threaded Threaded
7 messages Options
Reply | Threaded
Open this post in threaded view
|

info on directionality plugin

Durgam, Sushmitha S.

Hello all,

I would like to use the directionality plug in to compare orientation of elastin (red stain) in image A and image B


Would appreciate any input on how best to use the plug in to obtain quantitative information on the direction of elastin in image A and B

Many thanks,
Sushmitha

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

Image A.tif (1M) Download Attachment
Image B.tif (1M) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: info on directionality plugin

Krs5
Dear Sushmitha,

Did you see https://imagej.net/Directionality?

Best wishes

Kees

Dr Ir K.R. Straatman
Senior Experimental Officer
Advanced Imaging Facility
Centre for Core Biotechnology Services
University of Leicester
www.le.ac.uk/advanced-imaging-facility

ImageJ workshops 15 and 16 July, Leicester, UK. For registration visit https://www2.le.ac.uk/colleges/medbiopsych/facilities-and-services/cbs/AIF/workshops/imagej-workshops-July-2019 


-----Original Message-----
From: Durgam, Sushmitha S. <[hidden email]>
Sent: 09 April 2019 06:50
To: [hidden email]
Subject: info on directionality plugin


Hello all,

I would like to use the directionality plug in to compare orientation of elastin (red stain) in image A and image B


Would appreciate any input on how best to use the plug in to obtain quantitative information on the direction of elastin in image A and B

Many thanks,
Sushmitha

--
ImageJ mailing list: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&amp;data=02%7C01%7Ckrs5%40leicester.ac.uk%7C3c96cef01e1c4dca124708d6bcaf651c%7Caebecd6a31d44b0195ce8274afe853d9%7C0%7C1%7C636903858878541586&amp;sdata=9Twfljq5mZfozTK0f4g6pTg7dJDniQH%2FWuyEhq4w%2BBA%3D&amp;reserved=0

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: info on directionality plugin

Herbie
In reply to this post by Durgam, Sushmitha S.
Good day Sushmitha,

attached please find results (Orientation Salience Functions) of the
provided sample images. The applied approaches are explained in a report
available from here:
<www.gluender.de/Writings/WritingsTexts/HardText.html#Gl-2013-1>

Please note that the angle starts from the horizontal (0deg) and
increases counter-clockwise!

Your images show severe clipping (especially image B) at gray-value 85.
Hence they aren't well-suited for scientific analyses. You should
improve the image acquisition process.

I've used two highpass settings for every sample image with equivalent
cut-off frequencies of 2% and 10% of the Nyquist frequency.
Highpass-filtering removes low-frequency fluctuations in images that
otherwise are analyzed as well. You need to decide what setting is
appropriate. My guess is that no or only slight highpass-filtering (e.g.
2%) will do in your cases.

Please note that the Directionality-plugin performs sub-optimum
orientation analyses when compared to the approaches suggested here.
Besides other deficiencies it doesn't provide an option for defined
highpass filtering.

In any case you should be aware of the fact that Orientation Salience is
a measure that not only depends on the geometric orientation of image
structures but on their contrast as well. (Actually the
Directionality-plugin provides Orientation Salience Functions as well,
i.e. not orientation histograms in the pure geometric sense, although it
isn't mentioned in the docs.)

The computation of proper Orientation Salience Functions requires a
certain mathematical competence and coding skills. If you have only a
moderate number of images to analyze I offer to do the job for you.

Regards

Herbie

::::::::::::::::::::::::::::::::::::::::::::::::::
Am 09.04.19 um 07:50 schrieb Durgam, Sushmitha S.:

>
> Hello all,
>
> I would like to use the directionality plug in to compare orientation of elastin (red stain) in image A and image B
>
>
> Would appreciate any input on how best to use the plug in to obtain quantitative information on the direction of elastin in image A and B
>
> Many thanks,
> Sushmitha
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>


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

Orientation Salience of "ImageA_norm_HP02".png (35K) Download Attachment
Orientation Salience of "ImageA_norm_HP10".png (37K) Download Attachment
Orientation Salience of "ImageB_32_norm_HP02".png (42K) Download Attachment
Orientation Salience of "ImageB_32_norm_HP10".png (44K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: info on directionality plugin

Curtis Rueden
In reply to this post by Durgam, Sushmitha S.
Hi Herbie,

> The computation of proper Orientation Salience Functions requires a
> certain mathematical competence and coding skills. If you have only a
> moderate number of images to analyze I offer to do the job for you.

Please post the source code of your analysis, to facilitate reproduction
and validation of your approach.

-Curtis

--
Curtis Rueden
LOCI software architect - https://loci.wisc.edu/software
ImageJ2 lead, Fiji maintainer - https://imagej.net/User:Rueden
Have you tried the Image.sc Forum? https://forum.image.sc/



On Tue, Apr 9, 2019 at 3:59 AM Herbie <[hidden email]> wrote:

> Good day Sushmitha,
>
> attached please find results (Orientation Salience Functions) of the
> provided sample images. The applied approaches are explained in a report
> available from here:
> <www.gluender.de/Writings/WritingsTexts/HardText.html#Gl-2013-1>
>
> Please note that the angle starts from the horizontal (0deg) and
> increases counter-clockwise!
>
> Your images show severe clipping (especially image B) at gray-value 85.
> Hence they aren't well-suited for scientific analyses. You should
> improve the image acquisition process.
>
> I've used two highpass settings for every sample image with equivalent
> cut-off frequencies of 2% and 10% of the Nyquist frequency.
> Highpass-filtering removes low-frequency fluctuations in images that
> otherwise are analyzed as well. You need to decide what setting is
> appropriate. My guess is that no or only slight highpass-filtering (e.g.
> 2%) will do in your cases.
>
> Please note that the Directionality-plugin performs sub-optimum
> orientation analyses when compared to the approaches suggested here.
> Besides other deficiencies it doesn't provide an option for defined
> highpass filtering.
>
> In any case you should be aware of the fact that Orientation Salience is
> a measure that not only depends on the geometric orientation of image
> structures but on their contrast as well. (Actually the
> Directionality-plugin provides Orientation Salience Functions as well,
> i.e. not orientation histograms in the pure geometric sense, although it
> isn't mentioned in the docs.)
>
> The computation of proper Orientation Salience Functions requires a
> certain mathematical competence and coding skills. If you have only a
> moderate number of images to analyze I offer to do the job for you.
>
> Regards
>
> Herbie
>
> ::::::::::::::::::::::::::::::::::::::::::::::::::
> Am 09.04.19 um 07:50 schrieb Durgam, Sushmitha S.:
> >
> > Hello all,
> >
> > I would like to use the directionality plug in to compare orientation of
> elastin (red stain) in image A and image B
> >
> >
> > Would appreciate any input on how best to use the plug in to obtain
> quantitative information on the direction of elastin in image A and B
> >
> > Many thanks,
> > Sushmitha
> >
> > --
> > 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
Reply | Threaded
Open this post in threaded view
|

Re: info on directionality plugin

Durgam, Sushmitha S.
In reply to this post by Herbie

Hello,

Many thanks for these plots.

I am hoping to do some directionality assessments on these 7 images.

Would like to compare how and if directionality changes in 1, 2, 3 and 4, 5, 6 ?
If we use these assessments how can I do that? and what quantitative index can be used to express that

I very much appreciate your offer and will take you up on it. I wanted to do Oval Profile Plot and directionality assessment on these images.

Thanks,
Sushmitha
________________________________
From: ImageJ Interest Group <[hidden email]> on behalf of Herbie <[hidden email]>
Sent: Tuesday, April 9, 2019 4:58 AM
To: [hidden email]
Subject: Re: info on directionality plugin

Good day Sushmitha,

attached please find results (Orientation Salience Functions) of the
provided sample images. The applied approaches are explained in a report
available from here:
<www.gluender.de/Writings/WritingsTexts/HardText.html#Gl-2013-1<http://www.gluender.de/Writings/WritingsTexts/HardText.html#Gl-2013-1>>

Please note that the angle starts from the horizontal (0deg) and
increases counter-clockwise!

Your images show severe clipping (especially image B) at gray-value 85.
Hence they aren't well-suited for scientific analyses. You should
improve the image acquisition process.

I've used two highpass settings for every sample image with equivalent
cut-off frequencies of 2% and 10% of the Nyquist frequency.
Highpass-filtering removes low-frequency fluctuations in images that
otherwise are analyzed as well. You need to decide what setting is
appropriate. My guess is that no or only slight highpass-filtering (e.g.
2%) will do in your cases.

Please note that the Directionality-plugin performs sub-optimum
orientation analyses when compared to the approaches suggested here.
Besides other deficiencies it doesn't provide an option for defined
highpass filtering.

In any case you should be aware of the fact that Orientation Salience is
a measure that not only depends on the geometric orientation of image
structures but on their contrast as well. (Actually the
Directionality-plugin provides Orientation Salience Functions as well,
i.e. not orientation histograms in the pure geometric sense, although it
isn't mentioned in the docs.)

The computation of proper Orientation Salience Functions requires a
certain mathematical competence and coding skills. If you have only a
moderate number of images to analyze I offer to do the job for you.

Regards

Herbie

::::::::::::::::::::::::::::::::::::::::::::::::::
Am 09.04.19 um 07:50 schrieb Durgam, Sushmitha S.:

>
> Hello all,
>
> I would like to use the directionality plug in to compare orientation of elastin (red stain) in image A and image B
>
>
> Would appreciate any input on how best to use the plug in to obtain quantitative information on the direction of elastin in image A and B
>
> Many thanks,
> Sushmitha
>
> --
> 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
Reply | Threaded
Open this post in threaded view
|

Re: info on directionality plugin

Herbie
In reply to this post by Curtis Rueden
Good day Curtis!

The essential ImageJ-plugins are available and the processing steps are
described in great detail in my report.

As mentioned earlier, the source code is not available for commercial
reasons.

I have no idea how a processing with many case-dependent parameters
could ever be validated. As mentioned, the topic is none-trivial and
requires quite some knowledge of mathematical signal theory for the
proper *application* of the approach. The latter problem can't be
remedied by providing source code: The problem is data dependent. Most
of them are treated in the very report and the highpass question has
been discussed in contributions to this list and to the Forum.

Providing my expertise and a detailed report as well as my offer to
process images for free is more than posters can usually expect from a
list-member -- no?

Whether proposed results suit the needs of posters, is up to them
because I can only suggest parameter settings, e.g. the highpass cut-off
that depends on what exact image structures posters are interested in.
Funny enough, it turns out that quite often the latter question poses a
considerable problem for posters because its solution requires a
thorough understanding of their data.

In my opinion image-/signal-processing or -analysis in general is more a
question of mathematical signal and systems theory than of coding. Of
course, any eventual coding is to be done wisely, an aspect that is
treated in my report as well. (In fact it was the very reason I've
written it.)

Regards

Herbie

:::::::::::::::::::::::::::::::::::::::::::
Am 11.04.19 um 22:29 schrieb Curtis Rueden:

> Hi Herbie,
>
>> The computation of proper Orientation Salience Functions requires a
>> certain mathematical competence and coding skills. If you have only a
>> moderate number of images to analyze I offer to do the job for you.
>
> Please post the source code of your analysis, to facilitate reproduction
> and validation of your approach.
>
> -Curtis
>
> --
> Curtis Rueden
> LOCI software architect - https://loci.wisc.edu/software
> ImageJ2 lead, Fiji maintainer - https://imagej.net/User:Rueden
> Have you tried the Image.sc Forum? https://forum.image.sc/
>
>
>
> On Tue, Apr 9, 2019 at 3:59 AM Herbie <[hidden email]> wrote:
>
>> Good day Sushmitha,
>>
>> attached please find results (Orientation Salience Functions) of the
>> provided sample images. The applied approaches are explained in a report
>> available from here:
>> <www.gluender.de/Writings/WritingsTexts/HardText.html#Gl-2013-1>
>>
>> Please note that the angle starts from the horizontal (0deg) and
>> increases counter-clockwise!
>>
>> Your images show severe clipping (especially image B) at gray-value 85.
>> Hence they aren't well-suited for scientific analyses. You should
>> improve the image acquisition process.
>>
>> I've used two highpass settings for every sample image with equivalent
>> cut-off frequencies of 2% and 10% of the Nyquist frequency.
>> Highpass-filtering removes low-frequency fluctuations in images that
>> otherwise are analyzed as well. You need to decide what setting is
>> appropriate. My guess is that no or only slight highpass-filtering (e.g.
>> 2%) will do in your cases.
>>
>> Please note that the Directionality-plugin performs sub-optimum
>> orientation analyses when compared to the approaches suggested here.
>> Besides other deficiencies it doesn't provide an option for defined
>> highpass filtering.
>>
>> In any case you should be aware of the fact that Orientation Salience is
>> a measure that not only depends on the geometric orientation of image
>> structures but on their contrast as well. (Actually the
>> Directionality-plugin provides Orientation Salience Functions as well,
>> i.e. not orientation histograms in the pure geometric sense, although it
>> isn't mentioned in the docs.)
>>
>> The computation of proper Orientation Salience Functions requires a
>> certain mathematical competence and coding skills. If you have only a
>> moderate number of images to analyze I offer to do the job for you.
>>
>> Regards
>>
>> Herbie
>>
>> ::::::::::::::::::::::::::::::::::::::::::::::::::
>> Am 09.04.19 um 07:50 schrieb Durgam, Sushmitha S.:
>>>
>>> Hello all,
>>>
>>> I would like to use the directionality plug in to compare orientation of
>> elastin (red stain) in image A and image B
>>>
>>>
>>> Would appreciate any input on how best to use the plug in to obtain
>> quantitative information on the direction of elastin in image A and B
>>>
>>> Many thanks,
>>> Sushmitha
>>>
>>> --
>>> 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
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Software verification and validation

Herbie
In reply to this post by Curtis Rueden
For lister interested in this topic.

Curtis Rueden brought the question of "Software validation" to the
attention of this list.
(The topic has appeared several times already...)

In view of my global approaches to orientation analysis he wrote:
"Please post the source code of your analysis, to facilitate
reproduction and validation of your approach."(CR)

Here we get some insights to the meaning of the terms "software
verification and validation" (V&V):
<https://en.wikipedia.org/wiki/Software_verification_and_validation>

 From this wiki:

*Software verification is:*
"Assuming we should build X, does our software achieve its goals without
any bugs or gaps?"
"*Verification:* Are we building the product right?"

*Software validation is:*
"Was X what we should have built? Does X meet the high level requirements?"
"*Validation*: Are we building the right product?"


According to this definition, *software validation* is highly
application- or task-specific. For rather universal software tools, such
as ImageJ/Fiji, the task isn't well-defined. This also holds for certain
ImageJ-plugins and, as it turned out in the past, even for software with
a rather restricted scope such as my approaches to orientation analysis.
(I've tried to explain some reasons in the past.)

The above definition of *software verification* essentially deals with
the question whether a piece of software is correctly coded (actually it
reaches beyond this question). For this purpose it is necessary to
extensively test the software and perhaps to even check the source code.


In his plea (CR above) Curtis asked me for "validation" and for
"reproduction". The latter appears rather opaque to me in the context of
orientation analyses and, according to the above definition, source code
isn't required for software validation.

Although not explicitly asked for by Curtis: *Software verification* of
my approaches of orientation analyses is nicely solved in a unique
fashion. Unique, because it isn't generally possible but happened to
work for the task in question. In short, there are at least two
fundamentally different approaches to the proposed orientation analysis
that theoretically lead to identical results. (I've published the
corresponding proof in 1986.) Because both approaches have nothing in
common, neither mathematically nor computationally, it is
straightforward to check one by the other. Besides many other tests,
this is what I've done and explained in detail in my report (2013) that
meanwhile was downloaded by many posters of this list and of the forum.
<www.gluender.de/Writings/WritingsTexts/HardText.html#Gl-2013-1>
Consequently, those who have really studied this report should know how
to test their corresponding implementations.
Of course the implementations I use have been tested accordingly,
otherwise I couldn't have included the extensive test results in the report.

Hopefully this rather incomplete and somehow specific excursion into the
quality control of software was stimulating.

Best

Herbie

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