Semi-automatic segmentation of thoracic structures in computed tomography scans

• Grant number: 21/05704-2
• Support type: Scholarships in Brazil - Scientific Initiation
• Effective date (Start): July 01, 2021
• Effective date (End): June 30, 2022
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal researcher: Alexandre Xavier Falcão
• Grantee: Ilan Francisco da Silva
• Home Institution: Instituto de Computação (IC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Associated research grant: 14/12236-1 - AnImaLS: Annotation of Images in Large Scale: what can machines and specialists learn from interaction?, AP.TEM

Abstract

The ALTIS method, developed by the research group, provides a fast and accurate automatic segmentation tool of the lungs and trachea in a chest computed tomography scan. Moreover, the method is robust, since its methodology was developed considering anomalies that deform the lungs. However, the method fails in severe cases of pulmonary anatomy impairment or in the presence of structures attached to the pleura, such as tumors or consolidations. The present work aims at exploring deep learning of radiomic features from the image to improve the ALTIS method and provide a tool based on graph search algorithms to perform a semi-automatic correction of the segmentation, exploring the features learned by deep learning.