Teeth segmentation of dental periapical radiographs based on local singularity analysis

Area of Science:

• Medical Imaging
• Computer Vision
• Dental Diagnostics

Background:

• Accurate tooth segmentation in periapical radiographs is essential for diagnosing periapical lesions and periodontitis.
• Dental X-rays often suffer from image quality issues like noise, low contrast, and uneven illumination, complicating analysis.

Purpose of the Study:

• To develop an effective method for segmenting individual teeth in periapical radiographs.
• To improve the accuracy and reliability of automated tooth segmentation for dental diagnostic applications.

Main Methods:

• Image enhancement using adaptive power law transformation.
• Local singularity analysis with Hölder exponent for feature extraction.
• Tooth recognition via Otsu's thresholding and connected component analysis.
• Tooth boundary delineation using snake models and morphological operations.

Main Results:

• The proposed method successfully segmented 105 out of 106 teeth in 28 periapical radiographs.
• High mean segmentation accuracy (TP=0.8959, FP=0.0093) was achieved for the 75 teeth suitable for dental examination.

Conclusions:

• The presented four-stage scheme effectively segments teeth in periapical radiographs, even with image quality limitations.
• This automated approach shows significant potential for enhancing the accuracy of dental diagnostic procedures.