國立虎尾科技大學 |

Applying Convolutional Neural Networks for Radiographic Diagnosis of Periodontally Involved Teeth.

紀錄類型:	書目-語言資料,手稿 : Monograph/item
正題名/作者:	Applying Convolutional Neural Networks for Radiographic Diagnosis of Periodontally Involved Teeth./
作者:	Chang, Kearny.
面頁冊數:	1 online resource (81 pages)
附註:	Source: Masters Abstracts International, Volume: 85-01.
Contained By:	Masters Abstracts International85-01.
標題:	Dentistry. -
電子資源:	click for full text (PQDT)
ISBN:	9798379765347

Applying Convolutional Neural Networks for Radiographic Diagnosis of Periodontally Involved Teeth.
Chang, Kearny.

Applying Convolutional Neural Networks for Radiographic Diagnosis of Periodontally Involved Teeth. - 1 online resource (81 pages)

Source: Masters Abstracts International, Volume: 85-01.

Thesis (M.Sc.D.)--The University of Texas School of Dentistry at Houston, 2023.

Includes bibliographical references

Introduction: Convolutional neural networks (CNNs) are a type of deep machine learning, a branch of artificial intelligence, that have demonstrated abilities for pattern recognition to help physician to detect, monitor, and analyze diseases in the medical field. In dentistry, CNNs have already been used to detect oral diseases and classify dental conditions such as dental caries (Douglass et al., 1986), bone loss (Douglass et al., 1986), and endodontic lesions (Aminoshariae et al., 2021) with radiographs. CNNs can enable more comprehensive, systematic, and timely evaluation and documentation of dental images. There are currently few studies utilizing CNNs for periodontal disease diagnosis, and even fewer that have applied CNNs to diagnosing periodontitis with 2017 World Workshop on the Classification of Periodontal and Peri-implant Disease and Conditions. This study will develop a CNN that can perform tooth numbering on radiographs, determine the diagnostic quality of the radiographs, and assess the severity of bone loss, rate of progression of periodontitis, and periodontal defect and furcation involvement.Objective: To develop a deep machine learning model that can process dental periapical images to classify tooth type and number, diagnose furcation involvement, and categorize periodontal disease using the 2017 World Workshop on the Classification of Periodontal and Peri-implant Disease and Conditions.Material and methods: Three-hundred patients with full mouth standardized radiographs were included in the study. Three calibrated periodontists analyzed each tooth from the periapical films to perform (1) tooth numbering and tooth type classification as molars, premolars, incisors and canines, (2) radiographic bone loss (RBL) calculation and categorization of each tooth as healthy (no RBL), stage I ( 33% RBL); and (3) identification of buccal/lingual and interproximal radiographic furcation. Radiographic defect morphology was classified as horizontal, vertical. Teeth from the radiographs were excluded if they lacked diagnostic quality, such as severe elongation or foreshortening of the tooth, or root apex or cementoenamel junction of the tooth were not identifiable. The collected data was used to construct and test the CNN model.After being processed, the radiographic images were applied to the CNN model with global average pooling. Test results from deep machine learning were confirmed by five-fold cross validation and performance accuracy was calculated. Cohen's Kappa values were calculated to evaluate the inter-examiner agreements of clinician on diagnoses of RBL and compare with the results from machine learning.Results: For tooth numbering, the CNN model was able to achieve a mean accuracy of 94.8%. The CNN classified molars most accurately (99%), but also classified premolars (98.6%), incisors (98.8%) and canines (95.2%) with high mean accuracy.Using 2017 classification, the CNN classified RBL with sensitivities of 76.2%, 57.7%, 65.5% and 79.3% for no bone loss, stage I, stage II, and stage III, respectively, and specificity over 80% for each classification. The CNN was able to detect and classify no bone loss at 76.2%, stage I at 57.8%, stage II at 65.5% and stage III at 79.3% mean accuracy.For buccal and lingual furcation involvement, the CNN model achieved a sensitivity of 91% and specificity at 82%. The overall mean accuracy of the model was 85%. The detection of proximal furcation involvement reached sensitivity of 98% and specificity of 56%, with mean accuracy at 80%.Conclusions: The accuracy of periodontal bone loss categorization by deep machine learning surpassed 57%. This model can potentially be used as a diagnostic tool to increase efficiency of radiographic periodontal diagnosis, reducing dental practitioners' workload. However, its use for routine clinical diagnosis will require additional data collection for better model construction and training.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024

Mode of access: World Wide Web

ISBN: 9798379765347Subjects--Topical Terms:

674038
Dentistry.
Subjects--Index Terms:

Convolutional neural networksIndex Terms--Genre/Form:

554714
Electronic books.

Applying Convolutional Neural Networks for Radiographic Diagnosis of Periodontally Involved Teeth.
LDR:05399ntm a22004097 4500 001 1152590
005 20241209114603.5
006 m o d
007 cr mn ---uuuuu
008 250605s2023 xx obm 000 0 eng d
020 $a 9798379765347
035 $a (MiAaPQ)AAI30531135
035 $a AAI30531135
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Chang, Kearny. $3 1479628
245 1 0 $a Applying Convolutional Neural Networks for Radiographic Diagnosis of Periodontally Involved Teeth.
264 0 $c 2023
300 $a 1 online resource (81 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Masters Abstracts International, Volume: 85-01.
500 $a Advisor: Chang, Jennifer.
502 $a Thesis (M.Sc.D.)--The University of Texas School of Dentistry at Houston, 2023.
504 $a Includes bibliographical references
520 $a Introduction: Convolutional neural networks (CNNs) are a type of deep machine learning, a branch of artificial intelligence, that have demonstrated abilities for pattern recognition to help physician to detect, monitor, and analyze diseases in the medical field. In dentistry, CNNs have already been used to detect oral diseases and classify dental conditions such as dental caries (Douglass et al., 1986), bone loss (Douglass et al., 1986), and endodontic lesions (Aminoshariae et al., 2021) with radiographs. CNNs can enable more comprehensive, systematic, and timely evaluation and documentation of dental images. There are currently few studies utilizing CNNs for periodontal disease diagnosis, and even fewer that have applied CNNs to diagnosing periodontitis with 2017 World Workshop on the Classification of Periodontal and Peri-implant Disease and Conditions. This study will develop a CNN that can perform tooth numbering on radiographs, determine the diagnostic quality of the radiographs, and assess the severity of bone loss, rate of progression of periodontitis, and periodontal defect and furcation involvement.Objective: To develop a deep machine learning model that can process dental periapical images to classify tooth type and number, diagnose furcation involvement, and categorize periodontal disease using the 2017 World Workshop on the Classification of Periodontal and Peri-implant Disease and Conditions.Material and methods: Three-hundred patients with full mouth standardized radiographs were included in the study. Three calibrated periodontists analyzed each tooth from the periapical films to perform (1) tooth numbering and tooth type classification as molars, premolars, incisors and canines, (2) radiographic bone loss (RBL) calculation and categorization of each tooth as healthy (no RBL), stage I ( 33% RBL); and (3) identification of buccal/lingual and interproximal radiographic furcation. Radiographic defect morphology was classified as horizontal, vertical. Teeth from the radiographs were excluded if they lacked diagnostic quality, such as severe elongation or foreshortening of the tooth, or root apex or cementoenamel junction of the tooth were not identifiable. The collected data was used to construct and test the CNN model.After being processed, the radiographic images were applied to the CNN model with global average pooling. Test results from deep machine learning were confirmed by five-fold cross validation and performance accuracy was calculated. Cohen's Kappa values were calculated to evaluate the inter-examiner agreements of clinician on diagnoses of RBL and compare with the results from machine learning.Results: For tooth numbering, the CNN model was able to achieve a mean accuracy of 94.8%. The CNN classified molars most accurately (99%), but also classified premolars (98.6%), incisors (98.8%) and canines (95.2%) with high mean accuracy.Using 2017 classification, the CNN classified RBL with sensitivities of 76.2%, 57.7%, 65.5% and 79.3% for no bone loss, stage I, stage II, and stage III, respectively, and specificity over 80% for each classification. The CNN was able to detect and classify no bone loss at 76.2%, stage I at 57.8%, stage II at 65.5% and stage III at 79.3% mean accuracy.For buccal and lingual furcation involvement, the CNN model achieved a sensitivity of 91% and specificity at 82%. The overall mean accuracy of the model was 85%. The detection of proximal furcation involvement reached sensitivity of 98% and specificity of 56%, with mean accuracy at 80%.Conclusions: The accuracy of periodontal bone loss categorization by deep machine learning surpassed 57%. This model can potentially be used as a diagnostic tool to increase efficiency of radiographic periodontal diagnosis, reducing dental practitioners' workload. However, its use for routine clinical diagnosis will require additional data collection for better model construction and training.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2024
538 $a Mode of access: World Wide Web
650 4 $a Dentistry. $3 674038
650 4 $a Biomedical engineering. $3 588770
653 $a Convolutional neural networks
653 $a Furcation
653 $a Periapical
653 $a Periodontal bone loss
653 $a Periodontitis
653 $a Radiographic bone loss
653 $a Machine learning
655 7 $a Electronic books. $2 local $3 554714
690 $a 0567
690 $a 0541
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a The University of Texas School of Dentistry at Houston. $b Periodontics. $3 1479629
773 0 $t Masters Abstracts International $g 85-01.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30531135 $z click for full text (PQDT)