Reliability of Artificial Intelligence-Based Cone Beam Computed Tomography Integration with Digital Dental Images
February 23rd, 2024
A process of registering cone-beam computed tomography scans and digital dental images has been presented using artificial intelligence (AI) -assisted identification of landmarks and merging. A comparison with surface-based registration shows that AI-based digitization and integration are reliable and reproducible.
Our research is focused on integration of CBCT and antiscatter images, which is inevitable in creating a virtual hat. So far, those three images are markedly using the best film method, which is surface-based. This research is to introduce a new integration method based on AI-assisted digitization and to evaluate its accuracy.
Artificial intelligence has been used to predict treatment outcomes and digitize landmarks in supplementary radiographs or CBCT images. There are some commercial softwares available. This program adopts AI-assisted machine learned automation in digitizing landmarks in CBCT and also calibrating manually picked landmarks in dental images.
The interop job of the lab team showed a significant and almost perfect ICC in each method. The main difference showed no significant between the first and second registrations in each ABR and SBR in between both methods. However, their ranges were lower with ABR than with the SBR method.
The ABR protocol not only improved the accuracy, but also significantly reduced the merging time. While the SBR method took three to four minutes, the ABR program only required about 50 seconds for landmark selection, 40 seconds for DDI landmarks and two to three seconds for CBCT and DDI merging. To begin, launch the ON3D software on the computer, click on the Load DICOM File icon to import the files, select any DICOM file and click Open.
Click on the Reorientation icon in the Landmark panel. Then click the V notch of the frontal bone in the 3D view. Scroll up and down the Sagittal view to find the anterior-most point, then click the point where the frontal nasal suture meets the nasal and frontal bones.
Similarly, scroll through the Coronal view and click to determine the horizontal position of the nasion. Next, click on the most inferior point on the margin of the right orbital contour. Then scroll through the Coronal view to find the lowest point on the inferior margin of the right orbit and click.
Now in the Sagittal view, click on the most superior point of the right maxilla that constitutes the lower boundary of the orbit. Scroll through the Axial view, then click to align the red cross against the point where the eye orbit rims meet. Click the most inferior point on the margin of the left orbital contour, then press the most superior point of the outline of the right external auditory meatus.
Click the lowest point of the right temporal bone in the Coronal view to determine the horizontal and vertical positions, adjust the vertical and anterior posterior positions by clicking on the most superior point of the outline of the right external auditory meatus in the Sagittal view. Next, scroll through the Axial view to where the external auditory canal appears and the line of the temporal bones disappears, then click. Choose the most superior point of the outline of the left external auditory meatus.
Now click on the Done icon to finish the reorientation of the reconstructed craniofacial model. Then press the Preliminary Landmark Picking icon in the Landmark panel. Press the Execute icon in the Preliminary Landmark Picking panel to automatically choose preliminary landmarks.
To modify the landmarks, press the Manual Landmark Picking icon in the Volume tab, then make the necessary adjustments and press Done to confirm the placements of the landmarks. Press the Registration of Dentition Scan icon in the Tools panel. Then select the maxilla dentition and click on the Load icon in the Dentition Registration panel.
Now choose the STL files of the same patient with the CBCT model in the folder to load the maxilla dentition STL files. Switch the blue triangular arrows back and forth to pick the mesial buccal cusps of the right left maxillary first molars, and the right maxillary central incisor midpoint as registration landmarks. Finally, press the Done icon in the Dentition Registration panel and click on the Yes icon to confirm the automatic registration.
To merge the mandibular dentition, choose the mandible dentition and click on the Load button in the Dentition Registration panel. After generating the STL files, choose the mesial buccal cusp of the right left lower first molar and the right lower first incisor midpoint as registration landmarks. Now click on the Pick Registration Landmark icon in the Dentition Registration panel and press Confirm to modify the merged model.
To obtain the 3D coordinates of the landmarks, click on the Manual Landmark Picking icon in the Volume tab, or click on the Analysis tab to obtain the 3D coordinates of the landmarks. Lastly, click the Analysis tab and press on the Data Export panel followed by the Landmark icon to export the data as a file.
Overview
This study presents a novel method for integrating cone-beam computed tomography (CBCT) scans with digital dental images using artificial intelligence (AI). The AI-assisted approach focuses on landmark identification and merging, demonstrating reliability and reproducibility compared to traditional surface-based methods.
Key Study Components
Area of Science
- Neuroscience
- Dental Imaging
- Artificial Intelligence
Background
- Integration of CBCT and digital images is crucial for accurate dental assessments.
- Traditional methods rely heavily on surface-based registration techniques.
- AI has the potential to enhance the accuracy of landmark digitization.
- Commercial software exists that utilizes AI for similar applications.
Purpose of Study
- To introduce a new integration method based on AI-assisted digitization.
- To evaluate the accuracy of this new method compared to existing techniques.
- To improve the reliability of landmark identification in dental imaging.
Methods Used
- AI-assisted machine learning for digitizing landmarks in CBCT images.
- Manual calibration of landmarks in digital dental images.
- Comparison of AI-based integration with traditional surface-based methods.
- Evaluation of treatment outcome predictions using AI.
Main Results
- AI-based digitization shows high reliability and reproducibility.
- Integration method outperforms traditional surface-based registration.
- AI-assisted techniques can enhance the accuracy of dental imaging.
- Commercial software demonstrates practical applications of the research.
Conclusions
- AI-assisted methods provide a promising alternative for dental image integration.
- Future research should focus on refining these techniques for broader applications.
- The study highlights the potential of AI in improving dental imaging practices.
Frequently Asked Questions
Executive Industry Relevance
Reliable integration of CBCT and digital dental images using AI-based registration enhances quantitative imaging workflows in dental and craniofacial research. Improved reproducibility and reduced operator variability support higher predictive confidence at the imaging-to-analysis inflection point. This capability enables more robust data for downstream translational and preclinical model development in biopharma R&D.
Strategic Applications in Biopharma R&D
Early Discovery & Target Validation
- Enables precise anatomical mapping for hypothesis-driven target validation in craniofacial and dental research.
- Reduces manual error in landmark identification, supporting mechanistic de-risking of imaging-based endpoints.
- Improves confidence in quantitative imaging data for early-stage portfolio triage.
Screening & Assay Development
- Facilitates preparation of standardized, reproducible imaging datasets for downstream analysis.
- Supports assay development by providing consistent, quantitative 3D coordinate outputs.
- Enables scalable integration of imaging data for screening and comparative studies.
Translational & Preclinical Research
- Aligns imaging outputs with translational biomarker strategies in dental and craniofacial models.
- Ensures continuity of quantitative imaging data from discovery through preclinical validation.
- Supports risk-adjusted advancement decisions by reducing variability in imaging endpoints.
Pipeline & Workflow Integration
This AI-based CBCT and DDI integration method fits at the interface of imaging data acquisition and quantitative analysis, supporting workflows from early discovery through preclinical research.
- Discovery Biology: Provides reproducible, quantitative imaging data for hypothesis testing and pathway clarification.
- Screening: Delivers standardized 3D coordinate outputs for assay readiness and reproducibility.
- Analytics: Enables statistical comparison of imaging conditions using ICC and coordinate differences.
- Translational Research: Bridges imaging data to preclinical model validation when anatomical precision is required.
- Enterprise Reuse: Offers a scalable, operator-independent workflow for repeated imaging integration tasks.
Operational & Enterprise Impact
- Scientific Value: Increases predictive confidence and reduces mechanistic ambiguity in imaging-based studies.
- Operational Value: Standardizes and accelerates imaging integration, reducing manual workload and variability.
- Strategic Value: Supports better go/no-go decisions by providing robust, reproducible imaging data.
- Portfolio Impact: Enables risk-adjusted prioritization of imaging-dependent research programs.
Implementation Considerations
- Requires expertise in CBCT imaging and digital dental image processing.
- Depends on access to AI-enabled registration software and compatible analytical infrastructure.
- Demands cross-team standardization of landmark selection and data export protocols.
- May require adaptation for different anatomical regions or imaging modalities.
- Performance is contingent on the quality of input images and landmark visibility.
