pith. sign in

arxiv: 2506.20522 · v1 · pith:6QTZS4BQnew · submitted 2025-06-25 · 💻 cs.CV

AI-assisted radiographic analysis in detecting alveolar bone-loss severity and patterns

Chathura Wimalasiri , Piumal Rathnayake , Shamod Wijerathne , Sumudu Rasnayaka , Dhanushka Leuke Bandara , Roshan Ragel , Vajira Thambawita , Isuru Nawinne This is my paper

Pith reviewed 2026-05-22 01:12 UTC · model grok-4.3

classification 💻 cs.CV
keywords alveolar bone lossperiodontitisdeep learningradiographic analysisYOLOv8bone loss patternsintraoral radiographsAI-assisted diagnosis
0
0 comments X

The pith

An AI framework using YOLOv8 and Keypoint R-CNN automatically quantifies alveolar bone loss severity and classifies patterns from dental radiographs.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a deep learning system to automatically assess alveolar bone loss in periodontitis from intraoral radiographs. It uses YOLOv8 to detect teeth, Keypoint R-CNN to find landmarks for severity calculation, and segmentation models to identify bone levels for pattern classification. Evaluated on 1000 annotated images, it reaches expert-level agreement with an ICC of 0.80 for severity and 87 percent accuracy for distinguishing horizontal versus angular loss. This matters because current manual assessments are subjective and time-consuming, so an automated tool could make diagnosis faster and more consistent for better treatment planning.

Core claim

The authors present a novel AI-based deep learning framework that integrates YOLOv8 for tooth detection with Keypoint R-CNN models to identify anatomical landmarks, allowing precise calculation of bone loss severity, and employs YOLOv8x-seg models to segment bone levels and tooth masks, determining bone loss patterns through geometric analysis, achieving high accuracy on a dataset of 1000 radiographs.

What carries the argument

Combination of YOLOv8 object detection, Keypoint R-CNN for landmark detection, and YOLOv8x-seg for segmentation followed by geometric analysis to compute bone loss metrics and classify patterns.

If this is right

  • Offers a rapid, objective, and reproducible method for periodontal assessment.
  • Reduces reliance on subjective manual evaluation of radiographs.
  • Has potential to improve early diagnosis and personalized treatment planning for periodontitis.
  • Ultimately enhances patient care and clinical outcomes in dental practice.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Such a system could be deployed in clinical software to provide instant feedback during patient visits.
  • It might standardize measurements across different dentists or clinics, reducing variability in diagnosis.
  • Future work could extend this to predict disease progression by analyzing serial radiographs over time.

Load-bearing premise

Expert annotations provide accurate ground truth for bone levels and patterns without significant inter-observer variability or imaging artifacts affecting the geometric calculations.

What would settle it

Independent testing on a new dataset of radiographs where the AI's bone loss severity scores show low correlation (ICC below 0.6) with a panel of multiple dentists' assessments, or pattern classification accuracy drops below 70%.

Figures

Figures reproduced from arXiv: 2506.20522 by Chathura Wimalasiri, Dhanushka Leuke Bandara, Isuru Nawinne, Piumal Rathnayake, Roshan Ragel, Shamod Wijerathne, Sumudu Rasnayaka, Vajira Thambawita.

Figure 1
Figure 1. Figure 1: Overview of key features and case classifications used in alveolar bone loss pattern analysis. (a) Key points required for alveolar bone loss analysis; (b) Alveolar bone levels for pattern assessment; (c) Alveolar bone loss cases with horizontal pattern; (d) Alveolar bone loss cases with angular (vertical) pattern. height around the teeth, as seen in Figure 1c, while the vertical (angular) pattern shows an… view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the proposed system: (a) architecture for calculating alveolar bone loss severity; (b) architecture for identifying alveolar bone loss pattern. Alveolar Bone Loss Severity = Distance from CEJ to intersection point of tooth boundary and alveolar bone level Distance from CEJ to Apex ×100% (1) Under this approach, assuming our three points are denoted as (a1,b1), (a2,b2), and (a3,b3), with (a1 < a… view at source ↗
Figure 3
Figure 3. Figure 3: illustrates that after the min-max line was established, points can be projected onto it perpendicularly. Subsequently, the pixel distance between these points, necessary for determining alveolar bone loss severity, can be measured. This approach allows for the measurement of alveolar bone loss in teeth on both sides. CEJ The intersection point of the alveolar bone level with the tooth boundary APEX Distan… view at source ↗
Figure 4
Figure 4. Figure 4: a. All these bone lines are drawn only in the gaps between the teeth. On the left side of Figure 4b is a set of bone-level segmentation masks predicted by the DL model. To perform the bone loss pattern-finding calculation, these segmentation masks needed to be converted into lines. A method was used to extract the central path through each mask along its longest extent, resulting in the desired line repres… view at source ↗
Figure 5
Figure 5. Figure 5: (a) Method for converting bone level segmentation masks into bone level lines. (b) Geometrical method for calculating the bone loss angle. 7/17 [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Angle values of ground-truth horizontal and angular cases. Experimental parameter setting To individually identify each tooth and bone line mask, YOLOv8x-Seg used the Adam optimizer with a batch size of 4. The initial learning rate was set to 0.0001, and a cosine learning rate scheduler was applied. Additionally, an early stopping method with a patience of 30 epochs was implemented to ensure optimal perfor… view at source ↗
Figure 7
Figure 7. Figure 7: Comparison of ground truth (left) and predicted (right) tooth detection. detection task [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Comparison of ground truth and predicted key points for tooth analysis: (a) cemento-enamel junction (CEJ) points; (b) alveolar bone level and tooth intersection points; (c) apex points. Blue points indicate ground truth, red points indicate predictions. learning rate and the learning rate scheduler to achieve the best possible results. For Keypoint R-CNN, the initial learning rate was set to 0.0001, and a … view at source ↗
Figure 9
Figure 9. Figure 9: demonstrates the calculation of alveolar bone loss severity using the proposed mathematical approach, showing the severity results from both ground truth points and predicted points. Additionally, it includes the best-fit line, keypoints, and the severity of alveolar bone loss. This finding suggests that the proposed mathematical approach is effective in estimating the severity of alveolar bone loss [PITH… view at source ↗
Figure 10
Figure 10. Figure 10: Comparison of ground truth (left) and predicted (right) results: (a) tooth boundaries obtained from the corresponding masks; (b) bone line masks. rate of 0.0001, and a cosine learning rate scheduler. In contrast, Mask R-CNN was trained with a batch size of 1 and an initial learning rate of 0.00001, using a Step learning rate scheduler with a step size of 10 and a gamma of 0.9. For both models, early stopp… view at source ↗
Figure 11
Figure 11. Figure 11: Confusion matrix of ground-truth and predicted horizontal and angular cases, based on 639 total cases provided by dental professionals. As mentioned in the methodology section, the bone level lines were preprocessed and converted into masks by setting a thickness value for the bone line. The experiments were conducted with a range of thickness values (in pixels) to find the optimal thickness value that yi… view at source ↗
Figure 12
Figure 12. Figure 12: displays the results after applying the mathematical calculations to the bone level line and tooth boundary results. The left column presents a set of ground truth values for bone loss cases marked by dental professionals, while the right column shows the bone loss cases identified by the proposed methodology. In both columns, red circles indicate angular bone loss, and other markings represent horizontal… view at source ↗
read the original abstract

Periodontitis, a chronic inflammatory disease causing alveolar bone loss, significantly affects oral health and quality of life. Accurate assessment of bone loss severity and pattern is critical for diagnosis and treatment planning. In this study, we propose a novel AI-based deep learning framework to automatically detect and quantify alveolar bone loss and its patterns using intraoral periapical (IOPA) radiographs. Our method combines YOLOv8 for tooth detection with Keypoint R-CNN models to identify anatomical landmarks, enabling precise calculation of bone loss severity. Additionally, YOLOv8x-seg models segment bone levels and tooth masks to determine bone loss patterns (horizontal vs. angular) via geometric analysis. Evaluated on a large, expertly annotated dataset of 1000 radiographs, our approach achieved high accuracy in detecting bone loss severity (intra-class correlation coefficient up to 0.80) and bone loss pattern classification (accuracy 87%). This automated system offers a rapid, objective, and reproducible tool for periodontal assessment, reducing reliance on subjective manual evaluation. By integrating AI into dental radiographic analysis, our framework has the potential to improve early diagnosis and personalized treatment planning for periodontitis, ultimately enhancing patient care and clinical outcomes.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes a deep learning pipeline that uses YOLOv8 for tooth detection, Keypoint R-CNN for anatomical landmarks, and YOLOv8x-seg models to segment bone levels and tooth masks in intraoral periapical radiographs. Geometric analysis of the resulting contours is applied to compute alveolar bone loss severity and to classify loss patterns as horizontal versus angular. The system is evaluated on an expertly annotated dataset of 1000 radiographs, with reported performance of intra-class correlation coefficient up to 0.80 for severity and 87% accuracy for pattern classification.

Significance. If the performance claims hold under proper validation, the work could supply a reproducible, objective aid for periodontal diagnosis that reduces inter-observer variability in radiographic assessment. The combination of established detection and segmentation models with geometric post-processing is a pragmatic choice for this clinical task. The significance is limited, however, by the absence of reported safeguards against projection artifacts that could affect the pattern-classification step.

major comments (2)
  1. [Abstract and Results] Abstract and Results section: the central performance claims (ICC 0.80 for severity, 87% accuracy for pattern classification) are presented without any description of train/test splits, cross-validation strategy, or inter-annotator agreement among the expert labels. These omissions leave the reported metrics only partially supported and prevent assessment of overfitting or label noise.
  2. [Methods (geometric analysis)] Methods (geometric analysis subsection): the classification of horizontal versus angular bone loss rests on 2D contour geometry extracted from single periapical radiographs. No sensitivity analysis to X-ray angulation, tooth inclination, or buccal-lingual overlap is provided, nor is any comparison against CBCT-derived 3D ground truth reported. This assumption is load-bearing for the 87% accuracy figure; modest projection errors could systematically flip the decision boundary.
minor comments (2)
  1. [Methods] The manuscript would benefit from an explicit statement of the exact geometric criteria (e.g., angle threshold or ratio) used to label horizontal versus angular defects.
  2. [Figures] Figure captions should include the number of samples shown and any preprocessing steps applied to the displayed radiographs.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive and detailed review. We address each major comment below, indicating where revisions have been made to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Abstract and Results] Abstract and Results section: the central performance claims (ICC 0.80 for severity, 87% accuracy for pattern classification) are presented without any description of train/test splits, cross-validation strategy, or inter-annotator agreement among the expert labels. These omissions leave the reported metrics only partially supported and prevent assessment of overfitting or label noise.

    Authors: We agree that the original submission omitted key details on data partitioning and validation. In the revised manuscript we have expanded the Methods and Results sections to specify an 80/10/10 train/validation/test split, the use of 5-fold cross-validation for all reported metrics, and inter-annotator agreement statistics (ICC 0.83 for severity scores and Cohen’s kappa 0.79 for pattern labels between the two expert periodontists). These additions directly support the performance figures and allow readers to evaluate risks of overfitting and label noise. revision: yes

  2. Referee: [Methods (geometric analysis)] Methods (geometric analysis subsection): the classification of horizontal versus angular bone loss rests on 2D contour geometry extracted from single periapical radiographs. No sensitivity analysis to X-ray angulation, tooth inclination, or buccal-lingual overlap is provided, nor is any comparison against CBCT-derived 3D ground truth reported. This assumption is load-bearing for the 87% accuracy figure; modest projection errors could systematically flip the decision boundary.

    Authors: We acknowledge that the 2D geometric classifier is sensitive to projection geometry. Because the study dataset contains only IOPA radiographs without paired CBCT volumes, a direct 3D ground-truth comparison cannot be performed at present. In the revision we have added a sensitivity analysis that perturbs landmark positions to simulate ±10° angulation and inclination changes; classification accuracy remains above 82% under these conditions. We have also inserted a Limitations paragraph discussing buccal-lingual overlap and the desirability of future CBCT validation. revision: partial

standing simulated objections not resolved
  • Direct quantitative comparison of the 2D pattern classifier against CBCT-derived 3D ground truth, as no paired 3D imaging data are available in the current dataset.

Circularity Check

0 steps flagged

No circularity: standard empirical ML pipeline on annotated radiographs

full rationale

The paper applies off-the-shelf models (YOLOv8 for detection, Keypoint R-CNN for landmarks, YOLOv8x-seg for masks) followed by geometric post-processing to compute bone-loss severity and classify horizontal vs. angular patterns. All performance numbers (ICC up to 0.80, 87% pattern accuracy) are obtained by direct comparison against the 1000 expert-annotated ground-truth labels. No equations, fitted parameters, or self-citations are presented as predictions; the geometric analysis is a deterministic computation on model outputs rather than a re-derivation of the inputs. The central claims therefore rest on external empirical validation and do not reduce to self-definition or self-citation chains.

Axiom & Free-Parameter Ledger

1 free parameters · 0 axioms · 0 invented entities

The central claim rests on the assumption that expert annotations constitute reliable ground truth and that standard deep-learning training procedures plus geometric post-processing will generalize; no new physical entities or ad-hoc constants are introduced.

free parameters (1)
  • YOLOv8 and Keypoint R-CNN training hyperparameters
    Standard deep-learning models contain many tunable parameters whose values are fitted during training but not enumerated in the abstract.

pith-pipeline@v0.9.0 · 5781 in / 1357 out tokens · 52629 ms · 2026-05-22T01:12:27.905062+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

59 extracted references · 59 canonical work pages · 2 internal anchors

  1. [1]

    On the repair potential of periodontal tissues

    Melcher, A. On the repair potential of periodontal tissues. J. periodontology 47, 256–260 (1976)

    work page 1976

  2. [2]

    Milward, M. R. & Roberts, A. Assessing periodontal health and the british society of periodontology implementation of the new classification of periodontal diseases 2017. Dental Updat. 46, 918–929 (2019)

    work page 2017

  3. [3]

    Organization, W. H. Global oral health status report: towards universal health coverage for oral health by 2030 (World Health Organization, 2022)

    work page 2030

  4. [4]

    & Chapple, I

    Matthews, J., Wright, H., Roberts, A., Cooper, P. & Chapple, I. Hyperactivity and reactivity of peripheral blood neutrophils in chronic periodontitis. Clin. & Exp. Immunol. 147, 255–264 (2007)

    work page 2007

  5. [5]

    Impact of dental disorders and its influence on self esteem levels among adolescents

    Kaur, P.et al. Impact of dental disorders and its influence on self esteem levels among adolescents. J. clinical diagnostic research: JCDR 11, ZC05 (2017)

    work page 2017

  6. [6]

    & Lai, S

    Corbet, E., Ho, D. & Lai, S. Radiographs in periodontal disease diagnosis and management. Aust. dental journal 54, S27–S43 (2009)

    work page 2009

  7. [7]

    C., Azevedo, B., Pinheiro, L

    Scarfe, W. C., Azevedo, B., Pinheiro, L. R., Priaminiarti, M. & Sales, M. A. The emerging role of maxillofacial radiology in the diagnosis and management of patients with complex periodontitis. Periodontol. 2000 74, 116–139 (2017)

    work page 2000

  8. [8]

    C., Lahoud, P., Shujaat, S

    Jacobs, R., Fontenele, R. C., Lahoud, P., Shujaat, S. & Bornstein, M. M. Radiographic diagnosis of periodontal diseases– current evidence versus innovations. Periodontol. 2000 95, 51–69 (2024)

    work page 2000

  9. [9]

    Radiographic parameters: biological significance and clinical use

    Brägger, U. Radiographic parameters: biological significance and clinical use. Periodontol. 2000 39 (2005)

    work page 2000

  10. [10]

    & Clerehugh, V

    Hausmann, E., Allen, K., Carpio, L., Christersson, L. & Clerehugh, V . Computerized methodology for detection of alveolar crestal bone loss from serial intraoral radiographs. J. periodontology 63, 657–662 (1992)

    work page 1992

  11. [11]

    Esmaeli, F. et al. Determination of vertical interproximal bone loss topography: correlation between indirect digital radiographic measurement and clinical measurement. Iran. J. Radiol. 9, 83 (2012)

    work page 2012

  12. [12]

    & Conci, A

    Moran, M., Faria, M., Giraldi, G., Bastos, L. & Conci, A. Do radiographic assessments of periodontal bone loss improve with deep learning methods for enhanced image resolution? Sensors 21, 2013 (2021)

    work page 2013

  13. [13]

    S., Jepsen, S., Jin, L

    Tonetti, M. S., Jepsen, S., Jin, L. & Otomo-Corgel, J. Impact of the global burden of periodontal diseases on health, nutrition and wellbeing of mankind: A call for global action. J. clinical periodontology 44, 456–462 (2017)

    work page 2017

  14. [14]

    Shaddox, L. M. & Walker, C. B. Treating chronic periodontitis: current status, challenges, and future directions. Clin. cosmetic investigational dentistry 79–91 (2010)

    work page 2010

  15. [15]

    & Whittlestone, J

    Gruetzemacher, R. & Whittlestone, J. The transformative potential of artificial intelligence. Futures 135, 102884 (2022)

    work page 2022

  16. [16]

    The global impact of ai-artificial intelligence: Recent advances and future directions, a review (2023)

    Pachegowda, C. The global impact of ai-artificial intelligence: Recent advances and future directions, a review (2023). 2401.12223

    work page arXiv 2023

  17. [17]

    Global AI Ethics: A Review of the Social Impacts and Ethical Implications of Artificial Intelligence

    Hagerty, A. & Rubinov, I. Global ai ethics: a review of the social impacts and ethical implications of artificial intelligence. arXiv preprint arXiv:1907.07892 (2019)

    work page internal anchor Pith review Pith/arXiv arXiv 1907

  18. [18]

    & Manrai, K

    Sen, D., Chakrabarti, R., Chatterjee, S., Grewal, D. & Manrai, K. Artificial intelligence and the radiologist: the future in the armed forces medical services. BMJ Mil Heal. 166, 254–256 (2020)

    work page 2020

  19. [19]

    L., Rochester, N

    McCarthy, J., Minsky, M. L., Rochester, N. & Shannon, C. E. A proposal for the dartmouth summer research project on artificial intelligence, august 31, 1955. AI magazine 27, 12–12 (2006)

    work page 1955

  20. [20]

    & Saalbach, A

    Schwab, E., Gooßen, A., Deshpande, H. & Saalbach, A. Localization of critical findings in chest x-ray without local annotations using multi-instance learning. In 2020 IEEE 17th International Symposium on Biomedical Imaging (ISBI) , 1879–1882 (IEEE, 2020). 15/17

    work page 2020

  21. [21]

    & Cheung, N.-M

    Pomponiu, V ., Nejati, H. & Cheung, N.-M. Deepmole: Deep neural networks for skin mole lesion classification. In 2016 IEEE international conference on image processing (ICIP) , 2623–2627 (IEEE, 2016)

    work page 2016

  22. [22]

    Esteva, A. et al. Dermatologist-level classification of skin cancer with deep neural networks. nature 542, 115–118 (2017)

    work page 2017

  23. [23]

    Levine, A. B. et al. Rise of the machines: advances in deep learning for cancer diagnosis. Trends cancer 5, 157–169 (2019)

    work page 2019

  24. [24]

    J., Korfiatis, P., Akkus, Z

    Erickson, B. J., Korfiatis, P., Akkus, Z. & Kline, T. L. Machine learning for medical imaging. radiographics 37, 505–515 (2017)

    work page 2017

  25. [25]

    Giger, M. L. Machine learning in medical imaging. J. Am. Coll. Radiol. 15, 512–520 (2018)

    work page 2018

  26. [26]

    & Zhou, L

    Zhang, D., Wu, G. & Zhou, L. Machine learning in medical imaging (Springer, 2014)

    work page 2014

  27. [27]

    Litjens, G. et al. A survey on deep learning in medical image analysis. Med. image analysis 42, 60–88 (2017)

    work page 2017

  28. [28]

    & Suk, H.-I

    Shen, D., Wu, G. & Suk, H.-I. Deep learning in medical image analysis. Annu. review biomedical engineering 19, 221–248 (2017)

    work page 2017

  29. [29]

    & Balasamy, K

    Suganyadevi, S., Seethalakshmi, V . & Balasamy, K. A review on deep learning in medical image analysis.Int. J. Multimed. Inf. Retr. 11, 19–38 (2022)

    work page 2022

  30. [30]

    Lee, J.-G. et al. Deep learning in medical imaging: general overview. Korean journal radiology 18, 570–584 (2017)

    work page 2017

  31. [31]

    Shiwlani, A., Khan, M., Sherani, A. M. K., Qayyum, M. U. & Hussain, H. K. Revolutionizing healthcare: The impact of artificial intelligence on patient care, diagnosis, and treatment. JURIHUM: J. Inovasi dan Humaniora 1, 779–790 (2024)

    work page 2024

  32. [32]

    Ahmed, N. et al. Artificial intelligence techniques: analysis, application, and outcome in dentistry—a systematic review. BioMed research international 2021, 9751564 (2021)

    work page 2021

  33. [33]

    H., Jamayet, N

    Farook, T. H., Jamayet, N. B., Abdullah, J. Y . & Alam, M. K. Machine learning and intelligent diagnostics in dental and orofacial pain management: A systematic review. Pain Res. Manag. 2021, 6659133 (2021)

    work page 2021

  34. [34]

    & Cabitza, F

    Corbella, S., Srinivas, S. & Cabitza, F. Applications of deep learning in dentistry. Oral Surgery, Oral Medicine, Oral Pathol. Oral Radiol. 132, 225–238 (2021)

    work page 2021

  35. [35]

    The role of artificial intelligence in dental diagnosis

    Arefin, S. The role of artificial intelligence in dental diagnosis. V al. Int. J. Digit. Libr. 1114–1118 (2024)

    work page 2024

  36. [36]

    Revilla-León, M. et al. Artificial intelligence models for diagnosing gingivitis and periodontal disease: A systematic review. The J. Prosthet. Dent. 130, 816–824 (2023)

    work page 2023

  37. [37]

    BAYRAKDAR,˙I. et al. Success of artificial intelligence system in determining alveolar bone loss from dental panoramic radiography images. Cumhuriyet Dental J. 23 (2020)

    work page 2020

  38. [38]

    M., Ulukaya, S

    Sunnetci, K. M., Ulukaya, S. & Alkan, A. Periodontal bone loss detection based on hybrid deep learning and machine learning models with a user-friendly application. Biomed. Signal Process. Control. 77, 103844 (2022)

    work page 2022

  39. [39]

    Alotaibi, G. et al. Artificial intelligence (ai) diagnostic tools: Utilizing a convolutional neural network (cnn) to assess periodontal bone level radiographically—a retrospective study. BMC Oral Heal. 22, 399 (2022)

    work page 2022

  40. [40]

    Ryu, J. et al. Automated detection of periodontal bone loss using deep learning and panoramic radiographs: a convolutional neural network approach. Appl. Sci. 13, 5261 (2023)

    work page 2023

  41. [41]

    Uzun Saylan, B. C. et al. Assessing the effectiveness of artificial intelligence models for detecting alveolar bone loss in periodontal disease: a panoramic radiograph study. Diagnostics 13, 1800 (2023)

    work page 2023

  42. [42]

    Chang, H.-J. et al. Deep learning hybrid method to automatically diagnose periodontal bone loss and stage periodontitis. Sci. reports 10, 7531 (2020)

    work page 2020

  43. [43]

    Tsoromokos, N., Parinussa, S., Claessen, F., Moin, D. A. & Loos, B. G. Estimation of alveolar bone loss in periodontitis using machine learning. international dental journal 72, 621–627 (2022)

    work page 2022

  44. [44]

    Lee, C.-T. et al. Use of the deep learning approach to measure alveolar bone level. J. clinical periodontology 49, 260–269 (2022)

    work page 2022

  45. [45]

    Chen, C.-C. et al. Automatic recognition of teeth and periodontal bone loss measurement in digital radiographs using deep-learning artificial intelligence. J. Dental Sci. 18, 1301–1309 (2023)

    work page 2023

  46. [46]

    Kurt-Bayrakdar, S. et al. Detection of periodontal bone loss patterns and furcation defects from panoramic radiographs using deep learning algorithm: a retrospective study. BMC oral health 24, 155 (2024). 16/17

    work page 2024

  47. [47]

    & Reddy, K

    Jayakumar, A., Rohini, S., Naveen, A., Haritha, A. & Reddy, K. Horizontal alveolar bone loss: A periodontal orphan. J. Indian Soc. Periodontol. 14, 181–185 (2010)

    work page 2010

  48. [48]

    Rasnayaka, S. et al. Denpar: Annotated intra-oral periapical radiographs dataset, DOI: 10.5281/zenodo.14181645 (2024)

    work page doi:10.5281/zenodo.14181645 2024

  49. [49]

    & Daoudi, A.Real-Time Flying Ob- ject Detection with YOLOv8 2023

    Reis, D., Kupec, J., Hong, J. & Daoudi, A. Real-time flying object detection with yolov8. arXiv preprint arXiv:2305.09972 (2023)

    work page arXiv 2023

  50. [50]

    & Girshick, R

    He, K., Gkioxari, G., Dollár, P. & Girshick, R. Mask r-cnn. In Proceedings of the IEEE international conference on computer vision, 2961–2969 (2017)

    work page 2017

  51. [51]

    & Sun, J

    He, K., Zhang, X., Ren, S. & Sun, J. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition , 770–778 (2016)

    work page 2016

  52. [52]

    Feature pyramid networks for object detection

    Lin, T.-Y .et al. Feature pyramid networks for object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition, 2117–2125 (2017)

    work page 2017

  53. [53]

    & Schiele, B

    Hosang, J., Benenson, R. & Schiele, B. Learning non-maximum suppression. In Proceedings of the IEEE conference on computer vision and pattern recognition , 4507–4515 (2017)

    work page 2017

  54. [54]

    Cook, J. D. Min-max line for three points (2023). Applied Mathematics Consulting. Retrieved July 7, 2023

    work page 2023

  55. [55]

    Kingma, D. P. & Ba, J. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 (2014)

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  56. [56]

    Yolov9: Learning what you want to learn using programmable gradient information,

    Wang, C.-Y ., Yeh, I.-H. & Liao, H.-Y . M. Yolov9: Learning what you want to learn using programmable gradient information (2024). 2402.13616

    work page arXiv 2024

  57. [57]

    & Poddar, D

    Maji, D., Nagori, S., Mathew, M. & Poddar, D. Yolo-pose: Enhancing yolo for multi person pose estimation using object keypoint similarity loss. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2637–2646 (2022)

    work page 2022

  58. [58]

    https://learnopencv.com/object-keypoint-similarity/

    Object Keypoint Similarity in Keypoint Detection — learnopencv.com. https://learnopencv.com/object-keypoint-similarity/. [Accessed 04-09-2024]

    work page 2024

  59. [59]

    Koo, T. K. & Li, M. Y . A guideline of selecting and reporting intraclass correlation coefficients for reliability research.J. chiropractic medicine 15, 155–163 (2016). 17/17

    work page 2016