International dental journal · 2026
Student Performance in Bitewing Caries Detection: Artificial Intelligence Versus Alternative E-learning.
СтатьяСравнительное исследованиеНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Nagyová Valéria, Blaňár Dominik, Kybic Jan, Schwendicke Falk, Tichý Antonín
Аннотация
INTRODUCTION AND AIMS: This study compared 3 methods for teaching caries detection in bitewings: a prerecorded lecture, a preannotated dataset, and an artificial intelligence (AI)-based web application.
METHODS: Fifty-two dental students annotated carious lesions in 50 bitewings using minimum bounding boxes. After initial annotations, students were divided into 3 groups according to the training method: the Lecture Group (n = 16) received a prerecorded lecture on caries detection in bitewings, the Dataset Group (n = 17) had access to 50 bitewings annotated by a dentist, and the AI Group (n = 19) used an AI-based web application. After training, all students annotated caries in 50 previously unseen bitewings. Student annotations before and after training were compared to a reference standard of 3 experienced dentists. The evaluation was stratified according to the training method and stage of studies: preclinical (n = 16), junior clinical (n = 15), and senior clinical (n = 21).
RESULTS: All training methods significantly improved the mean number of errors, intersection over union of matching annotations, and accuracy. Sensitivity increased significantly in the Dataset Group (from 0.62 ± 0.14 to 0.78 ± 0.08) and the AI Group (from 0.68 ± 0.15 to 0.73 ± 0.12), as opposed to the Lecture Group, where a significant increase in specificity was observed (from 0.94 ± 0.09 to 0.96 ± 0.05). The stage of studies impacted the results; the extent of improvement decreased with increasing clinical experience.
CONCLUSION: While the 3 training methods varied in their impact on the confusion matrix components, they yielded comparable overall improvements. The AI-based web application could serve as an educational tool for caries detection in bitewings, especially for dental students with limited clinical experience.
CLINICAL RELEVANCE: This study shows that learning bitewing caries detection with an AI tool yields improvements comparable to other tested e-learning methods. Evaluating and comparing established e-learning and AI teaching methods is key to optimising AI-assisted education for better learning outcomes in dental training.
METHODS: Fifty-two dental students annotated carious lesions in 50 bitewings using minimum bounding boxes. After initial annotations, students were divided into 3 groups according to the training method: the Lecture Group (n = 16) received a prerecorded lecture on caries detection in bitewings, the Dataset Group (n = 17) had access to 50 bitewings annotated by a dentist, and the AI Group (n = 19) used an AI-based web application. After training, all students annotated caries in 50 previously unseen bitewings. Student annotations before and after training were compared to a reference standard of 3 experienced dentists. The evaluation was stratified according to the training method and stage of studies: preclinical (n = 16), junior clinical (n = 15), and senior clinical (n = 21).
RESULTS: All training methods significantly improved the mean number of errors, intersection over union of matching annotations, and accuracy. Sensitivity increased significantly in the Dataset Group (from 0.62 ± 0.14 to 0.78 ± 0.08) and the AI Group (from 0.68 ± 0.15 to 0.73 ± 0.12), as opposed to the Lecture Group, where a significant increase in specificity was observed (from 0.94 ± 0.09 to 0.96 ± 0.05). The stage of studies impacted the results; the extent of improvement decreased with increasing clinical experience.
CONCLUSION: While the 3 training methods varied in their impact on the confusion matrix components, they yielded comparable overall improvements. The AI-based web application could serve as an educational tool for caries detection in bitewings, especially for dental students with limited clinical experience.
CLINICAL RELEVANCE: This study shows that learning bitewing caries detection with an AI tool yields improvements comparable to other tested e-learning methods. Evaluating and comparing established e-learning and AI teaching methods is key to optimising AI-assisted education for better learning outcomes in dental training.