Seba Al Mokdad, Anas Al Houria, M. A. Talib, M. Moufti, A. Bouridane, Q. Nasir
{"title":"Analytical Overview on Transfer Learning in Processing Dental X-rays","authors":"Seba Al Mokdad, Anas Al Houria, M. A. Talib, M. Moufti, A. Bouridane, Q. Nasir","doi":"10.1145/3561613.3561635","DOIUrl":null,"url":null,"abstract":"Dental x-rays have been a standard piece of dental equipment for many years and are an indispensable diagnostic tool for dentists to detect tooth damage or disease. Recent research has focused on employing computer vision algorithms to automate analysis of dental x-rays. Our study aims to review the work done using transfer learning in dental image processing. AI solutions for dental images have been developed for many purposes, including examining tooth cavities (caries) and restorations and abnormalities in the maxillary sinuses. They have also been used to classify dental implants and determine gender in forensic studies. Transfer Learning is a new approach that is being used to solve a problem that classic deep learning and machine learning techniques could not solve: that of data limitation. Our search has investigated 80 research papers, of which 30 were relevant and analyzed in this paper. The identified studies have discussed a variety of transfer learning models to process different types of x-rays and have reported their efficacy using a variety of metrics. Transfer learning was used to solve various problems depending on the research question. Some papers compared the performance of transfer learning with that of dental experts in analyzing x-ray images, the accuracy of which were surprisingly close to equal. Although the results of the majority of dental applications performed using transfer learning models are encouraging, future research will need to solve the shortcomings highlighted in the present review.","PeriodicalId":348024,"journal":{"name":"Proceedings of the 5th International Conference on Control and Computer Vision","volume":"92 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2022-08-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings of the 5th International Conference on Control and Computer Vision","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/3561613.3561635","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
Dental x-rays have been a standard piece of dental equipment for many years and are an indispensable diagnostic tool for dentists to detect tooth damage or disease. Recent research has focused on employing computer vision algorithms to automate analysis of dental x-rays. Our study aims to review the work done using transfer learning in dental image processing. AI solutions for dental images have been developed for many purposes, including examining tooth cavities (caries) and restorations and abnormalities in the maxillary sinuses. They have also been used to classify dental implants and determine gender in forensic studies. Transfer Learning is a new approach that is being used to solve a problem that classic deep learning and machine learning techniques could not solve: that of data limitation. Our search has investigated 80 research papers, of which 30 were relevant and analyzed in this paper. The identified studies have discussed a variety of transfer learning models to process different types of x-rays and have reported their efficacy using a variety of metrics. Transfer learning was used to solve various problems depending on the research question. Some papers compared the performance of transfer learning with that of dental experts in analyzing x-ray images, the accuracy of which were surprisingly close to equal. Although the results of the majority of dental applications performed using transfer learning models are encouraging, future research will need to solve the shortcomings highlighted in the present review.