Optimizing parameters of an open-source airway segmentation algorithm using different CT images

摘要: Computed tomography (CT) helps physicians locate and diagnose pathological conditions. In some conditions, having an airway segmentation method which facilitates reconstruction of the from chest CT images can help hugely in assessment lung diseases. Many efforts have been made to develop algorithms, but methods are usually not optimized be reliable across different scan parameters. this paper, we present a simple semi-automatic algorithm segment tracheal bronchial anatomy using open-source 3D Slicer platform. The is based on region growing approach where trachea, right left bronchi cropped segmented independently three thresholds. its parameters efficient acquisition performance proposed has evaluated EXACT’09 cases local clinical as well breathing pig phantom multiple scans changing particular, investigate kernel, radiation dose slice thickness considered. Volume, branch count, length leakage presence evaluated. A new for evaluation developed correlation between metrics All considered successfully with good results terms presence. Results data comparable other teams’ methods, obtained by against EXACT09 challenge, whereas prove reliability platforms As expected, parameter affecting most, kernel seem particularly affect segmentation. system represents first quantitative presented repeatable tool like-for-like comparison platforms. suggest that stable starting point development complete algorithm.