IADR Abstract Archives

Automatically Identifying Teeth Numbers on Bitewing X-rays

Objectives: Dental x-rays are an important diagnostic tool for detecting damage and disease not visible during a regular dental exam. As dental x-rays move into the digital era, a software tool which can automatically retrieve findings from these radiographs will improve the consistency of the data and reduce the time required for clinicians to review radiographs. As the first step to develop an automated finding retrieval system, our objective was to develop an algorithm to automatically identify teeth numbers on bitewing x-rays.
Methods: The algorithm was developed based on the Universal Numbering System and only for identifying numbers for permanent teeth. A bitewing x-ray was first divided into two sections (upper and lower), then the algorithm processed the two sections separately. There were three main functions. The first one was used to roughly draw the boundary around each tooth and calculate the size of area within the boundary. The second function was to identify the type of the tooth (molar, premolar, canine and incisor) using the size data obtained from the first function. The third one was to decide the teeth numbers based on the number of each type of teeth and their relative locations. 30 bitewing x-rays from adult patients were used for evaluating the accuracy of the algorithm.
Results: The algorithm identified 93% of molars, 85% of premolars, 92% of canines, and 100% of incisors correctly. About 91.7% of the misidentified teeth had large restorations, and the one tooth which lost its crown was also misidentified.
Conclusions: The algorithm demonstrated the potential of identifying the teeth numbers for majority of teeth on bitewing x-rays. More work needs to be done for teeth with large restorations.
IADR/AADR/CADR General Session
2020 IADR/AADR/CADR General Session (Washington, D.C., USA)
Washington, D.C., USA
2020
0797
e-Oral Health Network
  • Li, Shuning  ( Indiana University Purdue University Indianapolis , Indianapolis , Indiana , United States )
  • Wang, Hua  ( Indiana University Purdue University Indianapolis , Indianapolis , Indiana , United States )
  • Thyvalikakath, Thankam  ( Indiana University School of Dentistry , Indianapolis , Indiana , United States )
  • NONE
    Poster Session
    eOral Health Network I
    Results
     MolarPremolarCanineIncisor
    # of Teeth for Testing11097132
    # of Teeth Correctly Identified10282122