Automatic lung segmentation for the inclusion of juxtapleural nodules and pulmonary vessels using curvature based border correction

Lung field segmentation is a significant step in the development of a Computer-Aided Diagnosis system and efficiently used for the quantitative analysis of lung CT images. However, this task is complicated especially in the presence of juxtapleural nodules, pulmonary vessels. In this paper, a more robust and accurate approach for lung segmentation is proposed. Our algorithm determines the dominant points indicating the concave and convex region along the lung boundary and these dominant points are connected by applying dominant point marching algorithm. The lung segmentation algorithm was tested on 36 subjects from the Lung Imaging Database Consortium. Results indicate that our proposed method can successfully re-include juxtapleural nodules and pulmonary vessels and achieves the average volumetric overlap fraction of 96.97%, average under segmentation rate of 2.324% and average over segmentation rate of 0.79%. The average processing time required for each subject is 182.448 s and for each slice is 0.953 s, which is comparatively faster than the manual segmentation done by radiologists. Experimental results suggest that our proposed method is faster and more robust than other state of the art methods. Hence proposed method can be used as an efficient tool for the lung segmentation in clinical practice. (c) 2018 The Authors. Production and hosting by Elsevier B.V. on behalf of King Saud University. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Automated summary

This paper proposes a more robust and accurate lung segmentation method which can successfully handle juxtapleural nodules and pulmonary vessels, achieving high levels of segmentation accuracy. The method is faster than manual segmentation done by radiologists.