Pulmonary arteries segmentation from CT images using PA‐Net with attention module and contour loss

Abstract

Background: Pulmonary embolism is a kind of cardiovascular disease that threatens human life and health. Since pulmonary embolism exists in the pulmonary artery, improving the segmentation accuracy of pulmonary artery is the key to the diagnosis of pulmonary embolism. Traditional medical image segmentation methods have limited effectiveness in pulmonary artery segmentation. In recent years, deep learning methods have been gradually adopted to solve complex problems in the field of medical image segmentation.Purpose: Due to the irregular shape of the pulmonary artery and the adjacent‐complex tissues, the accuracy of the existing pulmonary artery segmentation methods based on deep learning needs to be improved. Therefore, the purpose of this paper is to develop a segmentation network, which can obtain higher segmentation accuracy and further improve the diagnosis effect.Methods: In this study, the pulmonary artery segmentation performance from the network model and loss function is improved, proposing a pulmonary artery segmentation network (PA‐Net) to segment the pulmonary artery region from 2D CT images. Reverse Attention and edge attention are used to enhance the expression ability of the boundary. In addition, to better use feature information, the channel attention module is introduced in the decoder to highlight the important channel features and suppress the unimportant channels. Due to blurred boundaries, pixels near the boundaries of the pulmonary artery may be difficult to segment. Therefore, a new contour loss function based on the active contour model is proposed in this study to segment the target region by assigning dynamic weights to false positive and false negative regions and accurately predict the boundary structure.Results: The experimental results show that the segmentation accuracy of this proposed method is significantly improved in comparison with state‐of‐the‐art segmentation methods, and the Dice coefficient is 0.938 ± 0.035, which is also confirmed from the 3D reconstruction results.Conclusions: Our proposed method can accurately segment pulmonary artery structure. This new development will provide the possibility for further rapid diagnosis of pulmonary artery diseases such as pulmonary embolism. Code is available at https://github.com/Yuanyan19/PA‐Net.

Advancements in cardiac structures segmentation: a comprehensive systematic review of deep learning in CT imaging

“…In addition, it was reported that without using contrast media, contouring the borders between the structures can be difficult ( 11 ). Furthermore, using only CT axial slices to train the model can produce lower segmentation quality ( 5 ). Image resolution can affect the segmentation quality.…”

“…Training the model with different CT acquisitions for instance CT images with and without contrast media can enhance the DSC scores ( 11 ). Moreover, training the model with three CT planes (axial, sagittal, and coronal) might improve the segmentation's quality ( 5 ).…”

“…The features in question can be anatomical structures (such as the cardiac chambers) or pathological lesions (such as an area of myocardial infarction). Cardiac structures segmentation helps in the assessment of cardiac structures ( Figure 1 ) and can assist with earlier detection of cardiac abnormalities ( 1 , 5 , 7 – 9 ). Traditionally, segmentation has been performed manually by radiologists.…”

“…Dedicated cardiac CT is performed for a range of indications, including assessment of coronary artery disease, valvulopathy, and congenital heart disease, as well as preoperative planning and postoperative follow-up. The heart can also be assessed on CT undertaken for non-cardiac indications-such as CT pulmonary angiography for pulmonary embolism or CT angiography for acute aortic syndrome-allowing the identification of incidental but potentially significant findings and cardiac complications of non-cardiac disease (1)(2)(3)(4)(5).…”

Delineation of clinical target volume and organs at risk in cervical cancer radiotherapy by deep learning networks

“…Next we compare the CTV delineation quality of several DL models, including the proposed PPAF-net, PPF-net, the RA-CTVNet, 23 model by Rhee et al, 20 U-Net, 24 U-Net++, 25 EANet, 44 DeepLab v3+, 22 MSRF-Net, 45 TransUNet, 46 SegFormer, 47 SETR, 48 PA-Net 49 and OAU-Net. 50 The PPF-net is the PPAF-net without the attention module.…”