TUTORIALS

Before to proceed with quantitative analysis of microCT images with our software VDL it is important to make an image pre-processing in order to increase the quality and reduce the quantitative analysis that otherwise would be more complex and time consuming.
This handbook takes  as its reference ImageJ, open source software developed by NIH (http://imagej.nih.gov/ij/download.html).
ImageJ allows to perform operations in parallel and supports the so-called “stacks”, that is series of images that share the same window.

To open a stack of images with ImageJ click File ⊳ Import ⊳ Image Sequence and select the first image in the folder, then follow  the instructions and select the desired set of images.

MicroCT data analysis with ImageJ
The steps suggested are the following:
• Check that the sample did not move during the measurements. To do this you need to compare multiple images of the same measure and make sure that the sample edges between successive images are similar and that the images do not contain artifacts that substantially alter the shape of the object studied.
• Delete and remove from your stack images without the sample, or rather those in which there is only the background.
• In the case of multiple scan along vertical axe be sure that during microCT experiments motor rotation was always set correctly. In fact, if between a first measurement and a second one the motor was not brought back to its initial position, the second image stack will be rotate by 180° in respect to the first stack.

In this case use the ImageJ tool Image ⊳ Transform ⊳ Rotate … to turn the stack.

Image Crop
It is convenient to reduce the image background by resizing with the Crop tool. First you have to locate a Region of Interest using the rectangular selection tool on the first stack, this selection is applied automatically to all images in the stack.
To make the image crop you must position on the first stack and select:  Image ⊳ Crop  and select the option to apply crop to the whole stack.
To preserve the original images new resized images must be saved in a working folder (File ⊳ Save As ⊳ Image Sequence).

If you collected different portion of the same sample, the crop has to be applied to all of the stacks related to the same sample. To do this, you must select the first stack where was positioned the ROI and select Edit ⊳ Copy, select the second stack and select Edit ⊳ Paste and Edit ⊳ Undo, this procedure must be performed on all of the same sample stacks.
Increase the image contrast 
Quantitative analyses depend very much on contrast and how the edges are detectable. Images can be improved in this respect by increasing the contrast through Process ⊳ Enhance Contrast
Note that the value of “Saturated pixels” is set to 0.4%, then it can be useful to change it as explained in the next paragraphs.

Figure on the left shows an image of tooth before the contrast enhancement, on the right after the contrast enhancement. You may notice that in the second image the distinction between the phases of the tooth (background, dentin and enamel) is more pronounced.

Image conversion to 8 bits
Select the command Image ⊳ Type ⊳ 8-bit. Save 8 bits images to reduce disk space and reduce time needed to data analysis and visualization. In 8 bit format you can loose some information  about the differences between very similar tissues, for example if you study regenerated bone, but  in the study of tooth it is very convenient format.

Creating histograms
Select the tool Analyze ⊳ Histogram, including all images  to  have quantitative feeling with your data.
The histogram contains Gaussian-like curves, on the x-axis are reported the pixel gray values which can vary between 0 and 255 (8 bit = 256 bin, being 1 byte per pixel): in the images above the value 0 indicates a black pixel, the value 255 corresponds to white pixels, intermediate values represent different level of gray. It is possible to have images where this encoding is reversed 0 white 255 black, depending on the TAG of the analyzed image which can be MIN-IS-BLACK or MIN-IS-WHITE (the image TAGS you can access with the tool Image ⊳  Show Info). On the y-axis is reported the number of pixels corresponding to a certain tone of gray (intensity). Each peak corresponds to a different phase of the sample. To know the position of the peaks relating to the different phases of the sample you must move with the cursor on the top of the curve and next to the notice “value” it will appear the relative value. In this case, for example, the first peak corresponds to the empty space and has a value of 64, the second peak corresponds to dentin and has a value of 151 and the third peak corresponds to the enamel and has a value of 212.

The figure shows the comparison between two histograms: to the left the histogram obtained before the contrast enhancement, to the right it is shown the histogram after contrast enhancement. You may notice that the contrast enhancement increases the distance between the peaks corresponding to different material phases embedded in the sample.

At this point it is appropriate to create an Excel table where, for each stack, or rather. for each measurement by microCT, to mark the peak values of the different phases.
Histogram data can be saved and later plotted with other software such as Excel or Gnuplot; to do this you must click the button below the histogram List ⊳ File ⊳ Save As.
Save histograms can be useful for quantitative analysis and the choice of thresholds to distinguish between the different phases.

Correct Bias on the histogram value (very expert user)
Comparing data in tables related to the different stacks of the same sample you will notice that the peaks corresponding to the different phases are not aligned, this can be explained by the fact that the intensity of the X-ray beam is not constant. To make quantitative comparisons on local mass density in different samples or in the same sample following a physical-chemical or biological treatment may be useful to proceed with a “manual alignment of the peaks” by following the steps below.
Repeat the contrast enhancement at different values of saturated pixels and 8 bit conversion so that in the histograms the first peak (corresponding to the empty space) has a similar value.

Create image difference
In an experiment by Synchrotron microCT often happens that due to the limited Field of View of the detector the sample has to be acquired through two or more successive scans shifting the sample along the z-axis; therefore, after processing the images as described above, the various stacks related to the same sample must be connected. In particular between one section and the next of the same sample can be present several overlapping images. It is therefore necessary to identify those overlapping images and remove them from one of the stacks

Once visually identified images that overlap it is necessary to verify that actually are the right ones.
Open two images of successive stacks you believe coincident as those that follow.

 

Subtract the two images through the command Process ⊳ Image Calculator.

In the pictures below are reported the subtraction between two corresponding slices (on the left side) and the subtraction between two slices belonging to the sample but they are about 50 micron apart (right side)

In the graphs below are reported the pixel value of the two images along a straight line, to obtain the graph use the tool Analyze  ⊳  Plot Profile

If the subtraction is performed on two corresponding images the plot of pixels intensity along the line should be contain only noise. The presence of different peaks in the graph on the right side indicates that the two original pictures contain different information.

Files obtained from the vertical overlap of the stack are the starting files that can be used with the applications VDL. The software allow to import the whole image stack for the analyzed sample or more than one stack by automatically rename of the file of the consecutive stacks. This tool creates a new sequence of images representing the whole object and gives to the images a new progressive name.

For the upload procedure you can upload the data in the user area  as separate images or in a single zip file. Zip file has to be preferred to save time and disk space.

Micro-CT [1,2] allows studying the presence of cracks within the dentin and verifying whether these have originated before or after different dental treatments [3,4].

In this tutorial is analyzed the formation and propagation of micro cracks generated by endodontic treatment. The endodontic treatment was applied two times consecutively on the same tooth to compare the effects of minimally invasive approach and conventional more extended cavity access preparation.

The experiments were performed as follows:

A first SRmCT scan on intact extracted teeth (Pre)
Endodontic treatment with a minimally invasive access procedure for crown cavity preparation. A second SRmCT scan on samples treated with the minimally invasive approach (Minimal) Endodontic treatment with a conventional approach for cavity crown preparation
A third SRmCT scan after the extended cavity outline opening (Extended).

Our analysis allows the characterization of the size of a crack and a structural modification caused by the two consecutive treatments. The first step needed to obtain a morphometric evolution of cracks is the co-registration of the image stacks obtained through micro computed tomography (micro-CT) before and after the two treatments. The developed co-registration algorithm allows finding correct superposition of the three data sets, despite the pre and post-structures can be markedly different, in fact part of the pre-existing crack is destroyed by endodontic treatment .

Many of the features of VDL can be used to obtain this information [4,5]. In particular, in our analysis we used the RestorativeLab software (http://www.multimodal3d.com/restorativelab), this software processes simultaneously the pre and post treatment structures and analytically calculates the best superposition between the pre and post treatment structures.
Figure 1 and 2 show some original images obtained with SRmCT, belonging to the crown and the root respectively, that correspond at almost the same section of the tooth. After minimally invasive treatment (Figure 1B) the tooth crown was very similar to the native structure (Figure 1A) but crown pulp chamber was enlarged by the conventional treatment (Figure 1C). On the other hand root canals were enlarged by the minimal treatment (Figure 2B) and after the second treatment (Figure 2C) they were unchanged in respect to the first treatment (Figure 2A).

Figure 1. Slices of a dental crown, obtained with SRmCT. In (A) dental crown before treatments. In (B) dental crown after minimally invasive treatment. In (C) dental crown after conventional treatment.

Figure 2. Slices of a dental root, obtained with SRmCT. In (A) dental root before treatments. In (B) dental root after minimally invasive treatment. In (C) dental root after conventional treatment.

In figure 3 and 4 are shown some images co-registered with VDL where it is possible to see the difference between the three different datasets: Pre, Minimal and Extended. In figure 3 is shown to the left the co-registration between the images of pre- treatment and after the first treatment and to the right the co-registration between the first and the second treatment both at the level of the crown. In figure 4 is shown the co-registration between the images of pre-treatment and after the first treatment (left) and the co-registration between the first and the second treatment (right) both at the level of the roots. According to our analysis endodontic treatment causes the appearance or the propagation of micro-cracks in the crown, while at the root level the size of cracks and their length would remain unchanged.

Figure 3. (A) superposition between images of pre-treatment and after the first treatment. (B) superposition between the first and the second treatment. Both at the level of the crown. Dark grey corresponds to areas where micro-cracks were formed or evolved or the dentin was removed. White represent area where dentin was present in both structure.

Figure 4. (A) superposition between images of pre-treatment and after the first treatment. (B) superposition between the first and the second treatment. Both at the level of the roots. The colours are explained in the caption of figure 3.

Quantitative evaluation of crack dimensions

Once found the best superposition between the three different tooth structures it is possible to perform a quantitative morphometric analysis.
The morphometric analysis was performed using the ROI Manager of ImageJ software and other plugin embedded in ImageJ and BonJ []. These plug-in allow varying the ROI depending on the slice and merges the ROIs of selected in different slices in a unique 3D ROI. In fact as you scroll through the images the location of the crack does not remain the same, so you must choose the ROIs that include the crack following the the crack pattern.

The ROIs are selected from the Extended stack after co-registration and then the same ROIs are used on Minimal and Pre, this is because the cavity on the Extended is enlarged compared to the other stacks and the portion of the crack existing in that area on the Minimal and Pre stacks no longer exists on the Extended stack.

How to proceed:

Threshold

With the Threshold tool you can select parts of the tooth corresponding to empty spaces and therefore also the cracks by choosing a Threshold higher than the pixel gray value belonging to the cracks.

We start from the Extended: open the stack of images related to the Extended treatment and select Image ⊳ Adjust ⊳ Threshold

ROI Manager                                                           Click on Apply                                             Click on Ok

Open the ROI Manager (Analyze ⊳ Tools ⊳ ROI Manager) and select the ROI around the crack. Add the different ROIs on the ROI Manager by means of Add.

Select all the ROIs on the ROI Manager and click More and then Interpolated ROIs

Crack thickness analysis with BoneJ

Select Plugins ⊳ BoneJ ⊳ Thickness
This plugin calculates the thickness at each point of the crack as the diameter of the greatest sphere that fits within the structure and which contains the point. The plugin calculates mean and standard deviation of the trabecular thickness.

Convert images

Image ⊳ Type ⊳ 8-bit
If you want to display the image in grayscale, select Image ⊳ Lookup Tables ⊳ Grays

Threshold of the crack image

This second threshold is used to obtain a black and white image that is then processed with the tool Analyze Particles. Select window Threshold between 2 and 255.

Analyze Particles

Analyze Particles calculates the 3D volume of the crack, the volume is then measured in number of voxels. Analyze ⊳ Analyze particles

On the Size bar put a minimum of 2-3 to filter some noise. The crack can appear fragmented if images are not good quality.

Click Yes.

Save data files Results and Summary, in which is reported the crack area for each slice.

Repeat the procedure on the Minimal and Pre treatments stacks.

Open the stack with images related to the tooth after the Minimal treatment and make the first Threshold. The ROIs selected on the basis of Extended you can use for the other two stacks, to select the same ROIs position on the previous ROI manager and holding open the second stack select the “ROI previously interpolated”.
Thus it is not necessary to repeat interpolate ROIs.

Redo all steps starting with the thickness calculation performed with BoneJ.

Open the stack with images related to the Pre treatment and redo all steps starting with the first Threshold.

Summary of morphometric analysis

Evaluation of cracks modification performed on the Co-registered 3D structures can then be summarized in tables containing morphometric parameters. In our example 6 different cracks were selected 3 belonging to the crown and three to the roots.

TAB Tutorial2

References
[1] F. Paqué, F. Barbakow and O. A. Peters. Root canal preparation with Endo-Eze AET: changes in root canal shape assessed by micro-computed tomography. International Endodontic Journal, 38, 456–464, 2005.
[2] M.A. Marciano, M.A.H. Duarte, R. Ordinola-Zapata, A. Del Carpio Perochena, B.C. Cavenago, M.H. Villas-Bôas, P.G. Minotti, C.M. Bramante and I.G. Moraes. Applications of micro-computed tomography in endodontic research. Current Microscopy Contributions to Advances in Science and Technology (A. Méndez-Vilas, Ed.), 2012.
[3] I. Pop, A. Manoharan, F. Zanini, G. Tromba, S. Patel, F. Foschi. Synchrotron light-based μCT to analyse the presence of dentinal microcracks post-rotary and reciprocating NiTi instrumentation. Clin Oral Invest, 2014.
[4] R. Sinibaldi, A. Conti, R. Pecci, G. Plotino, R. Guidotti, N.M. Grande, M.G. Ortore, C. Becce, R. Bedini and S. Della Penna. Software tools for the quantitative evaluation of dental treatment effects from μCT scans. Journal of Biomedical Graphics and Computing, 2013, Vol. 3, No. 4.
[5] R. Sinibaldi, R. Pecci, F. Somma, S. Della Penna, R. Bedini. A new software for dimensional measurements in 3D endodontic root canal instrumenta tion. Ann. Ist. Super. Sanità vol.48 n.1 Roma Jan. 2012
[6] Doube M, Kłosowski MM, Arganda-Carreras I, Cordeliéres F, Dougherty RP, Jackson J, Schmid B, Hutchinson JR, Shefelbine SJ. BoneJ: free and extensible bone image analysis in ImageJ. Bone 2010 47:1076-9