Improvement of fully automated airway segmentation on volumetric computed tomographic images using a 2.5 dimensional convolutional neural net.

•A novel airway segmentation method using 2.5D CNN is proposed.•The 2.5D patches capture the 3D appearance of the airway efficiently.•Iterative optimization by probability-based patching improves the performance.•Validation on multiple dataset shows the reliability and practicality of our method.•Our end-to-end segmentation method is feasible in radiologic practice.