Pulmonary Fissure Detection in CT Images Based on Expectation Maximization Algorithm and Surface Fitting Model

Pulmonary fissures are important landmarks for recognition of lung anatomy.It is difficult to automatically detect pulmonary fissures due to factors such as deformation,incompleteness and disruption,a reliable and valuable approach based on expectation maximization algorithm and surface fitting model is proposed to detect pulmonary fissures.Firstly,we fuse the prior knowledge of trachea and pulmonary arteries to reduce the impact of other pulmonary structures for fissure region of interest localization.Then fissure orientation field is constructed to effectively discriminate between fissures and noise,and an expectation maximization algorithm is utilized to identify pulmonary fissures.Finally,a surface fitting model is used to make up for the undetected fissures for pulmonary fissure detection.The proposed method is validated on lung CT images of 20 patients with chronic obstructive pulmonary disease.Compared with manual fissure references,our method obtain a high segmentation accuracy with median F1-score of 0.979 and 0.971 for the left and right lung images,respectively.Experimental results show that the proposed method has a good performance in pulmonary fissure detection.