Parotid salivary ductal system segmentation and modeling in Sialo-CBCT scans

O. Shauly, L. Joskowicz*, E. G. Istoyler, C. Nadler

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

2 Scopus citations


Sialography Cone-Beam Computerized Tomography (Sialo-CBCT) imaging is commonly used for the assesment and diagnosis of salivary gland pathologies.  Since the examination of the Sialo-CBCT scans is challenging due to the salivary ducts complexity, oral radiologists perform their evaluation on 2D Maximum Intensity Projection images. This results in a mostly qualitative, incomplete, and observer-dependent analysis. We present the first fully automatic method for the segmentation and comprehensive quantitative structural analysis of the parotid salivary ducts in Sialo-CBCT scans. It consists of: 1) segmentation of the primary and secondary ducts; 2) computation of a 3D tree model of the salivary gland; 3) quantitative analysis of the salivary glands features, and; 4) visualization of the tree model and analysis results. We describe a new evaluation methodology for the validation of the salivary ducts model without ground-truth manual delineation. Experimental studies on 62 Sialo-CT scans show that our method successfully identifies 93% and 86% of the primary salivary ducts and the 1st and 2nd order duct branches and bifurcations. The RSME of the primary duct diameter is 0.33 mm (std = 0.04). Our method may be useful for the characterization of salivary gland architecture and for the diagnosis of ductal pathologies.

Original languageAmerican English
Pages (from-to)488-499
Number of pages12
JournalComputer Methods in Biomechanics and Biomedical Engineering: Imaging and Visualization
Issue number5
StatePublished - 2021

Bibliographical note

Publisher Copyright:
© 2021 Informa UK Limited, trading as Taylor & Francis Group.


  • Segmentation
  • Sialo-CBCT scans
  • modelling
  • salivary gland imaging
  • sialography


Dive into the research topics of 'Parotid salivary ductal system segmentation and modeling in Sialo-CBCT scans'. Together they form a unique fingerprint.

Cite this