IADR Abstract Archives

Developing a System for Automatic Detection & Characterisation of Enamel Caries

Objective: The purpose of this study was to develop an automated method for precise caries detection and characterization of x-ray micro computed tomography (microCT) images based on machine-based recognition and image processing methods and to quantify different parameters of enamel lesions

Method: Images of natural and artificial enamel lesions were obtained using a microCT machine (XRADIA). Imaging was undertaken using continuous mode exposures at 0.5 s intervals and binning value of 2, resulting in a resolution of 14.8 µm. The stacked file of reconstructed images was produced using reconstruction software (XRADIA) and exported images were saved in TIFF format. An algorithm was developed based on research needs and the algorithm was coded into machine language using MATLAB software. Multilevel thresholding was used for segmentation of the images based on gray level values which were normalized using hydroxyapatite phantoms. The developed program was used for batch processing of TIFF images.

Result: Processing of the images created characteristic mineral maps of various types of lesions which visualized specific pattern and morphology of different zones of the lesions. The program quantified the lesion parameters for both natural and artificial lesions based on the imaging resolution and the number of color pixels. The quantified parameters included the number of each color pixel, surface area and volume of each zone of the lesion, amount of mineral loss or gain for each zone of the lesion by weight and volume percent and finally the speed of remineralisation/demineralisation for each zone and the whole lesion.

Conclusion: The developed algorithm and program proved to be a reliable method for visualization, segmentation and quantification of enamel caries lesions. This method can be used for a variety of research applications including cariology and remineralisation studies and also can be used for visualization and quantification of clinical CBCT images.

Division: Australian/New Zealand Division Meeting
Meeting: 2014 Australian/New Zealand Division Meeting (Brisbane, Australia)
Location: Brisbane, Australia
Year: 2014
Final Presentation ID:
Abstract Category|Abstract Category(s): Scientific Groups
Authors
  • Shahmoradi, Mahdi  ( Faculty of Dentistry, University of Sydney, Sydney, , Australia )
  • Swain, Michael  ( Faculty of Dentistry, The University of Sydney, Sydney, , Australia )
  • SESSION INFORMATION
    Cariology Research