ROI statistics plugin: Surface area and Volume calculations

[r.g.r...@gmail.com]

Hello everyone,

I have been using the "intensity threshold to ROI (batch)" protocol to extract ROIs from my Z-stacks. The protocol works nicely and I get all the values I need in multiple excel files. I can then consolidate the individual excel files into one big master file. The three measurements I would like to get are maximum ferret diameter (in um), surface area of ROI (in um^2), and volume (um^3). 

 

The ROI statistics block, however uses units of pixels. To get maximum ferret diameter, I have just multiplied the pixel value given in the 'max ferret diam.' by the pixel size (0.1um in my case). Similarly, I thought that I could convert the surface area and volume values, by multiplying the 'contour' column by (0.1um)^2 and the 'interior' column by (0.1um)^3 (though I'm not sure about the 3D conversion, since my voxel size is actually 0.1x0.1x0.4um).

 

However, the values I calculate in excel differ from the surface area and volume values that show up when I open the image file in Icy (which now has the ROIs added in the metadata by the batch protocol) and pull up the ROI inspector (which provides surface area and volume in um^2 and um^3 respectively).

 

I am not sure if I have done something wrong in my calculations and whether to trust the ROI inspector or ROI statistics values.

 

My question is basically the following: How can I get measurements of ROI surface area and volume in um^2 and um^3 using the ROI statistics tool? 

Although I understand that I could export the data given in the ROI inspector (which already calculates these values for you), I would prefer to just do the conversion from the pixel values of contour and interior in my excel master file.

 

I am attaching a snapshot of the table I get when I pull up the ROI inspector window.

 

Thanks a lot for any insight, I'm still in the learning phase of bioimage analysis and Icy in general!

Roland    

 

Postdoctoral Researcher

Mark Harris Laboratory

University of Leeds

 

 

P.S. I've also noted that the ROI statistics plugin contains a typo ('roudness' instead of 'roundness')

[Stephane]

Hi Roland,

Did you checked the pixel size information are correctly filled in the Sequence ? You can access to them at the bottom of the Sequence tab information panel.
But given your numbers it seems that result are corrects :
contour = 46, interior = 52
surface area = contour * sqrt(psX * psY * psZ) = 46 * sqrt(0.004) = 46 * 0.063 ~ 2.9 um2
volume = interior * (psX * psY * psZ) = 52 * 0.004 = 0.208 um3

The surface area has to take the pixel size Z in account as well as this is an information computed on a 3D object. The result still is an estimation (which can be more accurate depending the implementation of the ROI object) but a better one that if we were using surface area = contour * (psX * psY)

Best,

- Stephane

[r.g.r...@gmail.com]

Thanks Stephane,

it really helped seeing it written out like that, thanks for taking the time!

Best,

Roland