Adaptive Region Growing based on detecting the seed point in the central trachea to the pre-segment respiratory system

This work proposes an efficient and valuable respiratory system pre-segmentation method from computed tomography (CT) images for medical software. The aim is to improve the efficiency of lung cancer treatment, early pulmonary nodule detection, and help for COVID 19 diagnostics. The proposed approach starts with looking for (in a specific range of Hounsfield pixel values) the center of the best area of the central trachea, this will serve as a seed point for a 3D region growing to pre-segment lung and trachea. Our approach has approved its effectiveness for respiratory system pre-segmentation and does not require any image preprocessing algorithms. For a dataset of 70 series (different patients) of a thoracic image, the method pre-segments the respiratory system in very remarkable durations with an average of 71,86 milliseconds for detecting the seed point and an average of about 40,69 seconds for the pre-segmentation of the respiratory system.Our experience shows that our algorithm can automatically pre-segment the respiratory system in a short time, and this work is easily scalable to become a fully automatic approach for lung and trachea segmentation.