Deep learning for detection and 3D segmentation of maxillofacial bone lesions in cone beam CT

Yeshua, Talia; Ladyzhensky, Shmuel; Abu-Nasser, Amal; Abdalla‐Aslan, Ragda; Boharon, Tami; Itzhak-Pur, Avital; Alexander, Asher; Chaurasia, Akhilanand; Cohen, Adir; Leichter, Isaac; Nadler, Chen

doi:10.1007/s00330-023-09726-6

Cited by 7 publications

(3 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thanks to the open source nature of the nnU-Net framework, our model could easily be finetuned with more (CB)CT data to meet specific needs. Detection of specific pathologies like periapical lesions or bone lesions could be added to the method (Yeshua et al 2023;Fu et al 2024). In the medium term, it is likely that other DL methods will exceed the classical 3D U-Net used in the study.…”

Section: Discussionmentioning

confidence: 99%

DentalSegmentator: robust deep learning-based CBCT image segmentation

Dot,

Chaurasia,

Dubois

et al. 2024

Preprint

Self Cite

View full text Add to dashboard Cite

Delineation of anatomical structures on dento-maxillo-facial (DMF) computed tomography (CT) or cone beam computed tomography (CBCT) scans is greatly needed for an increasing number of digital dentistry tasks. Following this process, called segmentation, three-dimensional (3D) patient-specific models can be exported for visualization, treatment planning, intervention, and follow-up purposes. Although several methods based on deep learning (DL) have been proposed for automating this task, there is no thoroughly evaluated publicly available tool offering segmentation of the anatomical structures needed for digital dentistry workflows. In this work, we propose and evaluate DentalSegmentator, a tool based on the nnU-Net deep learning framework, for fully automatic segmentation of 5 anatomic structures on DMF CT and CBCT scans: maxilla and upper skull, mandible, upper teeth, lower teeth and mandibular canal. A retrospective sample of 470 CT and CBCT scans was used as a training/validation set. The performance and generalizability of the tool was evaluated by comparing segmentations provided by experts and automatic segmentations on 2 hold-out test datasets: an internal dataset of 133 CT and CBCT scans acquired before orthognathic surgery and an external dataset of 123 CBCT scans randomly sampled from routine examinations in 5 institutions. In our internal test dataset (n = 133), the mean overall results were a Dice similarity coefficient (DSC) of 92.2 ± 6.3% and a normalized surface distance (NSD) of 98.2 ± 2.2%. In our external test dataset (n = 123), the mean overall results were a DSC of 94.2 ± 7.4% and a NSD of 98.4 ± 3.6%. The results obtained on this highly diversified dataset demonstrate that our tool can provide fully automatic and robust multiclass segmentation for DMF (CB)CT scans. To encourage the clinical deployment of DentalSegmentator, our pretrained nnU-Net model is made publicly available along with an extension for the 3D Slicer software.

show abstract

Section: Discussionmentioning

confidence: 99%

DentalSegmentator: robust deep learning-based CBCT image segmentation

Dot,

Chaurasia,

Dubois

et al. 2024

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…An alternative method is the manual segmentation for each anatomical region, but this is a very time consuming procedure that is still prone to error 37 . More recent, sophisticated approaches involving deep learning and artificial intelligence applications manage to reduce time considerably, but the segmentation accuracy levels are often not properly assessed or not substantially enhanced 38 , 39 . Due to its straightforward application, the single threshold segmentation remains the standard process for many imaging software, despite its weaknesses, especially in CBCT images.…”

Section: Discussionmentioning

confidence: 99%

Accuracy of facial skeletal surfaces segmented from CT and CBCT radiographs

Ghamri,

Dritsas,

Probst

et al. 2023

Sci Rep

View full text Add to dashboard Cite

The accuracy of three-dimensional (3D) facial skeletal surface models derived from radiographic volumes has not been extensively investigated yet. For this, ten human dry skulls were scanned with two Cone Beam Computed Tomography (CBCT) units, a CT unit, and a highly accurate optical surface scanner that provided the true reference models. Water-filled head shells were used for soft tissue simulation during radiographic imaging. The 3D surface models that were repeatedly segmented from the radiographic volumes through a single-threshold approach were used for reproducibility testing. Additionally, they were compared to the true reference model for trueness measurement. Comparisons were performed through 3D surface approximation techniques, using an iterative closest point algorithm. Differences between surface models were assessed through the calculation of mean absolute distances (MAD) between corresponding surfaces and through visual inspection of facial surface colour-coded distance maps. There was very high reproducibility (approximately 0.07 mm) and trueness (0.12 mm on average, with deviations extending locally to 0.5 mm), and no difference between radiographic scanners or settings. The present findings establish the validity of lower radiation CBCT imaging protocols at a similar level to the conventional CT images, when 3D surface models are required for the assessment of facial morphology.

show abstract

“…Chen Nadler presented on the use of AI in oral and maxillofacial imaging at Hadassah Medical Center, Israel. The study included 41 dental cone beam CT (CBCT) scans with, and 41 without histologically confirmed benign jaw lesions from 3 different CBCT machines (Yeshua et al, 2023). The team utilized MASK-RCNN to detect the bone lesion in each axial section.…”

Section: Medical Applications Of Aimentioning

confidence: 99%

Artificial intelligence in biology and medicine, and radioprotection research: perspectives from Jerusalem

Socol,

Richardson,

Garali-Zineddine

et al. 2024

Front. Artif. Intell.

View full text Add to dashboard Cite

While AI is widely used in biomedical research and medical practice, its use is constrained to few specific practical areas, e.g., radiomics. Participants of the workshop on “Artificial Intelligence in Biology and Medicine” (Jerusalem, Feb 14–15, 2023), both researchers and practitioners, aimed to build a holistic picture by exploring AI advancements, challenges and perspectives, as well as to suggest new fields for AI applications. Presentations showcased the potential of large language models (LLMs) in generating molecular structures, predicting protein-ligand interactions, and promoting democratization of AI development. Ethical concerns in medical decision making were also addressed. In biological applications, AI integration of multi-omics and clinical data elucidated the health relevant effects of low doses of ionizing radiation. Bayesian latent modeling identified statistical associations between unobserved variables. Medical applications highlighted liquid biopsy methods for non-invasive diagnostics, routine laboratory tests to identify overlooked illnesses, and AI's role in oral and maxillofacial imaging. Explainable AI and diverse image processing tools improved diagnostics, while text classification detected anorexic behavior in blog posts. The workshop fostered knowledge sharing, discussions, and emphasized the need for further AI development in radioprotection research in support of emerging public health issues. The organizers plan to continue the initiative as an annual event, promoting collaboration and addressing issues and perspectives in AI applications with a focus on low-dose radioprotection research. Researchers involved in radioprotection research and experts in relevant public policy domains are invited to explore the utility of AI in low-dose radiation research at the next workshop.

show abstract

Deep learning for detection and 3D segmentation of maxillofacial bone lesions in cone beam CT

Cited by 7 publications

References 43 publications

DentalSegmentator: robust deep learning-based CBCT image segmentation

DentalSegmentator: robust deep learning-based CBCT image segmentation

Accuracy of facial skeletal surfaces segmented from CT and CBCT radiographs

Artificial intelligence in biology and medicine, and radioprotection research: perspectives from Jerusalem

Contact Info

Product

Resources

About