[Establishment and evaluation of a computer-based software system for detection of initial approximal caries]

Area of Science:

• Dentistry
• Radiology
• Computer Science

Background:

• Early detection of approximal caries is crucial for effective dental treatment.
• Radiographic interpretation for initial caries can be subjective and prone to error.

Purpose of the Study:

• To develop and assess a computer-based software system, the Caries Diagnosing System (CDS).
• To evaluate the performance of CDS in detecting initial approximal caries using radiography.

Main Methods:

• 190 approximal tooth surfaces from 95 extracted posterior teeth were analyzed using bite-wing radiography.
• A CDS was developed using Matlab and VC language to analyze radiographic characteristics of approximal lesions.
• The CDS automatically detected lesions, and its performance was compared against 8 dentists' visual estimations and histological analysis.

Main Results:

• The CDS achieved a sensitivity of 0.728 and a Pearson's correlation of 0.722.
• The system demonstrated superior performance compared to optic estimation.
• The reproducibility of the CDS was found to be good.

Conclusions:

• The Caries Diagnosing System (CDS) can accurately report the site and extent of approximal caries lesions.
• CDS offers a sensitive and accurate method for detecting initial approximal caries in clinical practice.
• The system provides objective data, enhancing diagnostic capabilities beyond visual assessment.