INFORMATICS AND INFORMATION PROCESSES

Automated generation of three-dimensional model of bronchus tree based on CT image

E. V. Aristova, A. A. Smirnov

Institute of Natural Sciences, Ural Federal University named after the first President of Russia Boris Yeltsin, Ekaterinburg

Abstract: This article presents the development of an algorithm for the automatic segmentation of the bronchial tree from CT images and the construction of its 3D model. The proposed method is based on texture analysis using reference segments and supervised machine learning algorithms, including Decision Tree and Random Forest. The effectiveness of the algorithm was tested on a dataset comprising 15 CT scans with over 300 annotated reference segments. Evaluation metrics such as Accuracy, Recall, Precision, F1, and IoU were used to assess performance, with Random Forest demonstrating the best results. The study outlines areas for future improvement, such as reference segment accumulation and image preprocessing. This development has potential applications in clinical practice for bronchoplasty planning and serves as a diagnostic and visualization tool for complex respiratory structures.

Keywords: bronchial tree, CT, segmentation, machine learning, texture analysis.

Received: 22.07.2025
Revised: 24.10.2025
Accepted: 22.07.2025

DOI: 10.60797/IRJ.2025.160s.12