Quantcast

Otolith Analysis

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Otolith Analysis

Angela K
I am new to ImageJ and trying to figure out the best way to analyze fish otoliths (same idea as tree rings). I have downloaded the ObjectJ plugin, but can't get it to automatically mark rings, or record the lengths into columns. From the short tutorial video it seemed like these options would be ready to go when I downloaded the plugin, but it seems like I have to set everything up and I am struggling to do so.

Here's the general procedure I wish to accomplish using ImageJ - I have attached one of my images so you may have a better idea of what I'm working with: I will select a center point (if someone knows a good way to automatically select this, compared to just visually selecting the center, please share your thoughts), then mark points at each growth ring (along some sort of axis or straight line). I want to measure the increment lengths between each ring, as well as the total radius. I will be analyzing thousands of these otolith images so I would like to automate the process as much as possible to be consistent and save time.

It seems like the plot profile tool may be useful (that measures intensity along a line). If I could somehow select points along the intensity profile and have the points be applied to the image that would be GREAT. Another option could be the gel tool because the growth bands slightly resemble bands on a western blot.

Any suggestions will be greatly appreciated!
Thanks!
Angela

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

image010.jpg (3M) Download Attachment
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Otolith Analysis

Michael Schmid-3
Hi Angela,

from your image, as a non-expert I can't say for sure what one should
consider a valid growth ring and what is not a ring.  The rings look
quite different in different directions from the center.
Usually, if it is difficult for a human to decide, there is no way to do
it automatically.  After all, the image processing unit of mammals has
been developed in a massively parallel genetic algorithm over millions
of years.  This will beat even the best computer code...

I fear that the weakness of the features is also the reason why the
'Tree ring' functions of ObjectJ don't work well.  What you could try to
improve it:
- Convert to grayscale (optional, but speeds up the following)
- Select the background with the Versatile Wand Tool [1] and set it to
roughly the average color of the outer parts of the otolith (90-100 in
the example)
- Apply a high-pass filter [2] (large radius such as 50-100)
- Increase the contrast and apply
- Smooth by a Gaussian of suitable radius to make the substructure of
the rings disappear.
Instead of the highpass, you might also try smoothing first, then
'Subtract background' with 'sliding paraboloid' and a very large radius
(like 1000-100000).

It might also make sense to scale down the image to fewer pixels, which
would make processing faster and possibly detection easier for ObjectJ.
--

What I could also imagine, if this fails: you mark a series of points
(multipoint selection), with the first one being the center and the
others at the maxima of the growth rings, all in the same direction from
the center. Then you could have a macro that saves the coordinates,
calculates the distance or whatever.
--

Transferring points of a Profile Plot to the section:
That would be a great feature, I agree!
But also quite some work to implement it nicely! (Currently I am working
on the ImageJ Plot class, but I have absolutely no time to do add this
in the near future).

Best,

Michael


[1]
http://imagejdocu.tudor.lu/doku.php?id=plugin:segmentation:versatile_wand:start
[2]
http://imagejdocu.tudor.lu/doku.php?id=plugin:filter:highpass_filter:start

________________________________________________________________
On 19/02/2017 10:11, Angela K wrote:

> I am new to ImageJ and trying to figure out the best way to analyze
> fish otoliths (same idea as tree rings). I have downloaded the
> ObjectJ plugin, but can't get it to automatically mark rings, or
> record the lengths into columns. From the short tutorial video it
> seemed like these options would be ready to go when I downloaded the
> plugin, but it seems like I have to set everything up and I am
> struggling to do so.
>
> Here's the general procedure I wish to accomplish using ImageJ - I
> have attached one of my images so you may have a better idea of what
> I'm working with: I will select a center point (if someone knows a
> good way to automatically select this, compared to just visually
> selecting the center, please share your thoughts), then mark points
> at each growth ring (along some sort of axis or straight line). I
> want to measure the increment lengths between each ring, as well as
> the total radius. I will be analyzing thousands of these otolith
> images so I would like to automate the process as much as possible to
> be consistent and save time.
>
> It seems like the plot profile tool may be useful (that measures
> intensity along a line). If I could somehow select points along the
> intensity profile and have the points be applied to the image that
> would be GREAT. Another option could be the gel tool because the
> growth bands slightly resemble bands on a western blot.
>
> Any suggestions will be greatly appreciated! Thanks! Angela
>
> -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

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