Confocal Laser Scanning Microscope HELP!




I have taken some confocal laser scanning microscope images of bovine enamel by dying with a fluorescent dye. This will cover all the surface and fill small porosity into the enamel. I would love to quantify some volumes or surface areas of these pores through the space filled by the dye. I this possible firstly? I have been scouring the web and this forum and to no avail through my lack of knowledge on the subject or there not being a clear method. I am a new researcher to the field of image processing and would love any assistance any one can give.

I have attached the data into a google drive folder below:

Thanks in advance


Yes it’s possible but you have to be careful that your imaging conditions give you sufficient resolution in all the dimensions that you’ll need to do the measurements. Confocal has relatively much worse resolution in z than in xy (especially at low magnification) which makes volume measurements problematic. If the enamel pores are basically linear, you may be better off estimating pore quantity in 2D, e.g. by length of pore per area, rather than by volume.

Dental enamel is also highly scattering (that’s why your teeth are white), so it would help your protocol a lot if you can work out a clearing method to reduce scatter. A glycerol/ethanol/water combo might do it, or the dental histology literature may recommend something suitable.


Thanks for replying Michael,

I have been looking into the boneJ plugin as tooth and bone are similar. I initially looked to use the volume fraction analyser as the fluorescence is rhodamine B should only cover the surface and pore surface area within the porosity between the crystal rods of enamel, the problem i have with this is knowing the results im producing are in fact reliable as i’m unsure how to threshold and process the samples images correctly.

I have also looked into the 3D object counter for volume but seem to not produce any numbers probably again my lack of knowing how to pre process the images for such analysis.

Another way i have been looking into is counting the voxels of each stack and calculating the volume that way but haven’t been able to figure out what plugin or command i should use.

Any help you can provide or even links to some basics on what thresholding and processing images before running quantitative analysis would be greatly appreciated. I feel little like i’m running around in a circle at present.

In regards to the clearing method how can i determine if i have severe scattering as my images look pretty clear to my eye but i could be wrong. Also how would the combo you have suggested (or any other) reduce the scattering? i’ve not come across this and would love to know more about this as the protocol i use was developed by a previous PhD student, and i aim to develop it further and produce some numbers for analysis hence the volume and possible surface area being of particular interest.

Kind regards,


This is the quintessential image analysis problem. One way to proceed is to do a sensitivity analysis, which is where you systematically test slightly different processing settings’ effects on your final result, and determine where the result is most stable and/or closest to a ground truth or gold standard value (e.g. values measured using images from a higher resolution instrument like SEM).

Ways to reduce that uncertainty, or at least inter-sample error/noise, include standardising your imaging conditions and including some well-defined test objects, either commercial or of your own design. Because researchers are often doing some quirky thing that no-one has done before, we often have to devise our own standard objects.

Once you have got a thing that is a reasonable test object, you can test your processing steps with a sensitivity analysis, and see how that translates to the sensitivity analysis on your experimental objects.

Simple measurements of pore volume are completely doable, and BoneJ’s Volume Fraction plugin can do a simple volumetric pixel count for you (the ‘BV’ value is most useful for your purposes probably), provided a segmented (rhodamine as foreground) input image.


How will i segment the rhodamine dye? I’m struggling to figure this out for my image without causing the image to change drastically.


Segmentation by nature causes your image to change drastically: from original pixel values to pixel values that represent the determination of ‘foreground’ and ‘background’. You will need to try some different preprocessing and segmentation approaches that work for your images and your question, and choose the one that works the best (ideally with a clear definition of what ‘best’ means).


what would be the best route of segmentation for my image stack in the google drive? I have tried a few ways and gotten numbers but unsure on what im meant to be doing or watching out for. I’m a total novice at this. I tried binary erode and dilating then running the analysis.


I have been playing with pre-processing techniques (sharpen,smooth, enhance contrast) I was wondering if the enhance contrast should be used? It seems to make auto threshold near impossible adding lots of noise at the deeper stacks. I’m unsure on the best way to get a good image for analysis still. I have used find edges also which gives I’m assuming where the void is on the edges but the intensity in the voids from the dye has been lost. I’m assuming it will only show the intensity at the edges?

How can I prepare the image best for a volume fraction measurement?