1. Intro

The department of dentistry at the University of Gent conducted a study involving the production of numerous images through micro-CT scans. Due to the sheer quantity of images and the potential for human error during manual processing, it was not feasible to process them manually.

Invisto created a tool to automate the processing of these images.

‍

‍

2. The study performed by UGent

Achieving complete disinfection of the root canal space is the objective of root canal treatment. However, due to the limitations of the mechanical aspect of root canal cleaning, irrigation plays a crucial role in endodontic treatment. Conventional irrigation methods have been found to be inadequate, prompting a surge in research on the effectiveness of irrigation. The efficacy of the EddyTM (VDW), a new sonic irrigation device, in cleaning root canals and accumulated hard tissue debris (AHTD) had not yet been examined.

‍

2.1 Objective

The objective of the study was to examine the effectiveness of a new sonic irrigant activation device in eliminating accumulated hard tissue debris (AHTD) from the mesial root canal system of mandibular molars, using micro-computed tomography analysis.

‍

2.2 Methodology

Mandibular molars from human subjects were extracted and selected based on micro-CT scans (12 µm). The mesial roots were mechanically instrumented to an apical diameter of ISO 30. The teeth were then randomly assigned to one of three final irrigation groups:

• Passive Sonic Irrigation using Eddy^TM (VDW)
• Manual Irrigation
• Ultrasonic Activated Irrigation

Following instrumentation and final irrigation, micro-CT scans (12 µm) were performed, and the volume of the isthmus filled with hard tissue debris was calculated before and after irrigant activation through reconstruction and image processing.

‍

‍

3. Micro-CT

The teeth underwent micro-CT scans at three different stages - before preparation, after preparation, and after irrigation. A single scan produced 1500 images, resulting in a total of 4500 images per tooth that required processing.

‍

‍

4. Image analysis

4.1 Co-Registration

The software utilized to align and overlay images of the teeth before and after instrumentation was Bruker microCT 3D registration, with reference to the outer root contours.

‍

4.2 OpenCV image analysis of the root canal

Until now, this post has concentrated on presenting a summary of the research and its background. The remainder of the post will explore the automation tool that was created to perform the analysis.

As many are aware, at Invisto, we often prefer to use Qt for our software development. Qt proved to be an excellent choice for this particular project:

• Efficient memory allocation and usage. The tool needed to load 1500 images twice, once before and once after, into the memory.
• Intensive processing. Each slice went through a range of operations (OpenCV).
• Multithreaded.  The tool employed a multithreaded approach utilizing QThreadPool to maximize processing speed across the slices. QThreadPool manages and recycles individual QThread objects to help reduce thread creation costs.
• QML: We have used QML in favour of QWidgets to create a nice, modern user interface. The user interface runs on a separate thread, ensuring it remains smooth and responsive while performing all processing tasks.

We utilized an Intel® Core™ i9-9900K CPU @ 3.60GHz × 16 as our hardware, which features 8 cores and 16 threads.

‍

Step 1: selecting the images

The initial step is to choose the folders that contain the reference and post-instrumentation images

‍

Step 2: Slice selection

During the slice selection stage, it is possible to choose a subset of slices. In this study, the focus was solely on the slices that defined the mesial root canal space. The first slice was situated below the pulp chamber, where the distal and mesial roots separate to form distinct canals. The final slice was located at the tooth apex.

Step 3: Region of interest

The third step entailed identifying the region of interest by indicating the location of the mesial root canal on the image.

Step 4: Threshold

OpenCV's thresholding technique is employed to segment images by distinguishing the object or region of interest, in our case the root canal and debris within it, from the background. This process entails comparing each pixel value of the image with a threshold value and converting it to a binary value.

Step 5: Opening

Opening in OpenCV is a morphological operation that involves applying erosion followed by dilation to an image. The erosion operation involves the removal of small foreground objects from an image while the dilation operation restores the size and shape of the remaining objects. Opening is used to remove small noise from the image while preserving the shape and size of the larger objects in the image.

Step 6: Result

The result generated by the tool was presented in the form of a CSV table, which displays the quantity of debris eliminated for each irrigation method utilized.

5. Conclusion

We were happy to be able to help out the researchers with this tool. The researchers could focus on their expertise while the tool handled the massive amount of images that were generated from the 30 teeth that were used during this study.

Leveraging Qt and OpenCV for this undertaking enabled us to create a high-performing tool in a short period.

‍