Validation of ImageJ

imagej

#1

Hello,
I’m quietly new in ImageJ, and I was wondering if ImageJ has ever passed any validation process.
I’m required to proof that this tool is valide.
My use was restricted to only pixel grey value measurement, no more than that. No image processing or other manipulation.
Any help will be appreciated.

Thanks,
Raphael


#2

Good day Raphael,

This question has been discussed several times before and mainly on the ImageJ-list. Just do a search in: https://list.nih.gov/cgi-bin/wa.exe?A0=IMAGEJ

The problem is that ImageJ is open source and as such it can be arbitraily modified and augmented. Except for a rather small core, ImageJ is a collection of plugins and although there is a basic collection of plugins which makes plain ImageJ, you can load as many other plugins coded by who-so-ever. Of course nobody knows what configuration of ImageJ is in use on a certain computer and this makes it difficult to validate ImageJ.

The good news is, that very many people use ImageJ and severe flaws are rapidly detected and corrected.

HTH

Herbie


#3

Thanks a lot for your response.
The point is that my use is very basic, using only grey value picker…

Thanks anyhow,
Raphael


#4

I understand but how do you expect someone to validate the very function.

Who should look at what you are really doing with exactly which operations performed by which configurations of ImageJ and validate that it is absolutely correct?

I guess you won’t find somebody who is able and willing to do this.

Of course you can be pretty sure that such basic operations are correctly performed. But that’s only “pretty sure”, not a validation.

Regards

Herbie


#5

just a generic question, for my pure curiousity, how are other software validated?


#6

Software-validation can be a pretty costly affair …

You need to show at least that a software tool does what it is expected to do when handled correctly and sometimes even to show that it can’t do any harm if handled incorrectly …

There are companies who do this job.

Best

Herbie


#7

Hello Herbie, Again.

Now I need to validate ImageJ. For this I’m looking for standard reference grayscale images , on which the gray value is known. Can you help me to find such images? Have you ever heard about such images?

Thanks in advance,
Raphael


#8

Good day Raphael!

Now I need to validate ImageJ.

How do you think to manage this?
Are you qualified to do so?

[…] standard reference grayscale images , on which the gray value is known.

In which units?

I think this doesn’t help for validation or at least only for a confirmation of certain i/o-properties of ImageJ. If you think of the TIF image format: It allows a very large variety of image definitions that I doubt you will be able to test all and actually there are quite some TIF-styles that are not recognized by ImageJ although BioFormats extends to many more but not all.

It would be helpful to know what you are going to do and what your input images will look like in the end.

Regards

Herbie


#9

I’m not qualified for this specifically. I’m asked to try to…

My use of ImageJ is very limited. I only measured pixel gray value of seleted area on an image. (.png image 8-bit with value from 0 to 255)

Then, the purpose is to confirm that the measured gray value by ImageJ is correct. For that I expected to find images with known values (I found http://www.color-hex.com/color-palette/3046 but I prefer to use a standard one)

Thanks,
Raphael


#10

Raphael,

you need achromatic test images in PNG-format and you need to confirm that they are imported correctly and that the gray values that you get with what operation ? in ImageJ are those specified by the maker of the test images.

I’m not aware of such specified images that are themselves validated.

Furthermore, if you manage to perform the validation, you are responsible for the result. Are you sure you want to be bear the responsibility?

Regards

Herbie


#11

The operation is ctrl + m (measure),

How do I convince myself that ImageJ is ok?


#12

Raphael,

who is asking you for validation?

As I’ve written before, if you need to be pretty sure that the measurements are ok, then go ahead. If you need a validation then you are entering a complicated and perhaps costly process. I just had a look at the link you’ve provided and the images are surely not validated and only a few gray values are provided.

Which images are you importing to ImageJ for measurement. Where do they come from?

How do I convince myself that ImageJ is ok?

I have no idea but I never encountered any problems with such basic operations.

Regards

Herbie


#13

What do you mean by go ahead?

Thanks,
Raphael


#14

dear Rapahel,
I really think this is the main question.


#15

What do you mean by go ahead?

Just do your measurements with ImageJ and make sure all steps of processing/measuring are precisely protocolled:

What images (formats etc.) are opened, what settings are use for ImageJ in general and for the measurements. What measurements are performed. How are the results saved etc.

Regards

Herbie


#16

Internal (company) requirments…


#17

Raphael,

if you really need validation in the strict sense, then I would use a validated application, not ImageJ. (I assume that your company is not willing to pay the costs for validation by an external institution.)

Regards

Herbie


#18

Thanks for your advice. Can you tell me some examples of validated application?

Raphael


#19

You may ask Adobe if Photoshop is validated for the processes you need.
Have a look at Bio/Medical applications that are sometimes validated.

Regards

Herbie


#20

Thanks for your help.

Raphael