Image Properties (pixels to microns) ISSUE

Hi All,

I have an issue with my images when I change the properties from pixels to microns. I have a 3D image of a red blood cell (stack of 96 slides), I open it with ImageJ and it automatically reads it in pixels (512x512). If I open the image with the 3D viewer WITHOUT changing its properties, I have a nice red blood cell! with the expected shape. But when I change the image properties to microns with the following values (as reported from the microscope manufacturer), [Pixel width: 0.182, Pixel height: 0.182, Voxel depth: 0.372] the image of my red blood cell gets distorted.

Would any of you be able to explain to me why this is the case and if this distortion can influence the values I then measure for volume and surface area using MorphoLibJ plugin and 3D manager?

I have attached two example images:

  1. red blood cell with normal shape (unchanged properties)

  2. red blood cell with distorted shape (much thicker than the other one)

I have been struggling with this issue for the last two days and cannot seem to find an answer anywhere.

Thank you in advance!

Best,
Talia

@tal

Wait. How exactly are you opening your images in ImageJ?

We recommend using Bio-Formats to open your images… File > Import > Bio-Formats This will ensure that your metadata (as long as it is present and correct) will get read in correctly when opening the image. The Drag&Drop method isn’t so reliable when it comes to metadata. So use Bio-Formats - for sure - try this first.

When you adjust the image properties to microns… are you doing the correct conversion? So if you go to Set Scale - make sure you have a ‘Known Distance’ and that the ‘Pixel Axis Ratio’ is correct.

Do all of this before opening with the 3D Viewer… double-check those Image Properties before/after conversions.

And if you want - you can share an original image with us here so we can try to see if we get the same error, etc.

@etadobson

Thank you for your suggestions!

I tried both points but the image still get distorted once I change the properties. The conversion values were given to me by the company producing the microscope so I’m guessing they are correct.

I’m attaching an original image as you suggested so it may be easier for you to help me out.
https://www.dropbox.com/s/ncy2k87krk9twbh/Example_RBC.tiff?dl=0

Thanks!
Talia

@tal

I honestly don’t know what’s going on here… The conversion values you are getting from the manufacturer seem ‘correct’ (checking diameter of your RBC in the image provided was ~44pixels - so converting would give ~8pixels) - so your width/height conversions are fine… in theory. Maybe the voxel depth is ‘off’?? What step size did you use during acquisition?

Do others here have some tips? I don’t do too much 3D/volumetric work myself… so my experience in this is a bit limited. @kephale or @skalarproduktraum ??

@etadobson

The step size is 0.3125 um (depth of field is 30 um & a z-stack of 96 images)

I am also very confused by this and the manufacturer is not giving me any inputs on this issue.
Thanks for your help trying to figure this out!

@tal
I get it. It’s mixing me up too… but perhaps this older thread has the answer:

I think that’s why your image is ‘stretched’ in the z dimension… the difference between 0.3125um step size and then this 0.372 voxel depth given to you by the company. Can you try using your step size instead? 0.3125um? Do you see that stretching issue again in 3D viewer then?

1 Like

@etadobson

Unfortunately I still see the z distortion. I think the stretching is better but there is still a difference with the original image (unchanged properties)

The only way I don’t seem to get the distortion is if I put 0.182 as the value for voxel depth as well.
But I am don’t really have a reason for doing so.

@tal

Yeah - you don’t really want to be putting arbitrary numbers in there… Perhaps @imagejan or @bnorthan have some thoughts on this? I’m sorry I’m not so much help in this… we’ll get it figured out though!

@etadobson

Thanks a lot anyways for all your inputs!
Let’s see if anyone else has any suggestions and in the meantime, I will keep pressuring the manufacturer for their opinion on this.

1 Like

Hi @tal

One possibility is that there is axial blur resulting in a loss of depth resolution.

First what shape do you expect?? It looks to me that with unchanged properties the shape of the cell is more ‘pancake’ like. After changing properties it is elongated in the z direction.

Is the true shape spherical??

If so the second one could be correct, depending on the amount of axial blur. A spherical object would look slightly elongated in the z (depth) direction.

What modality are these images ?? Are you able to share the raw data?? If so that could be informative as we could look projections along different dimensions.

1 Like

That’s a great point @bnorthan !!! RBCs are kind of donut-shaped… so not completely spherical: https://www.quora.com/What-is-the-shape-of-a-red-blood-cell

Hi @bnorthan,

Thanks for your inputs.
Regarding the shape of the red blood cell, it is kind of a donut-shape as @etadobson mentioned. So the image with the unchanged properties is the more correct one.

I think you are right in saying that there is axial blur. The thing I don’t understand is that my volumetric and surface area measurements remain the same regardless of whether I change the image properties or not. However, the Feret measurement changes and this is crucial to me as I have to measure the diameter.

Unfortunately the raw images are in a unique format to the software of the microscope. I can only share the .tiff images as I did in a previous post.

Let me know if there is any other information I can provide you with to help solve this issue!

1 Like

Have you tried any image repair software? such as Stellar Repair for Photo, Recuva, or any other tool to fix the distorted images. I think you must to try at least one if the image distorted or corrupted!

@jason_william

I’m not familiar with image repair softwares. I tried to download Stellar Repair for Photo but it’s not a free software, and Recuva it seems to only recover files. These tools seem to refer to corrupted images only, so I don’t know if they would actually resolve the distortion issues.

But again, I am definitely not an expert so any further suggestion is greatly appreciated!