Hierarchical Vertex Regression-Based Segmentation of Head and Neck CT Images for Radiotherapy Planning

Abstract

Segmenting organs at risk from head and neck CT images is a prerequisite for the treatment of head and neck cancer using intensity modulated radiotherapy. However, accurate and automatic segmentation of organs at risk is a challenging task due to the low contrast of soft tissue and image artifact in CT images. Shape priors have been proved effective in addressing this challenging task. However, conventional methods incorporating shape priors often suffer from sensitivity to shape initialization and also shape variations across individuals. In this paper, we propose a novel approach to incorporate shape priors into a hierarchical learning-based model. The contributions of our proposed approach are as follows: 1) a novel mechanism for critical vertices identification is proposed to identify vertices with distinctive appearances and strong consistency across different subjects; 2) a new strategy of hierarchical vertex regression is also used to gradually locate more vertices with the guidance of previously located vertices; and 3) an innovative framework of joint shape and appearance learning is further developed to capture salient shape and appearance features simultaneously. Using these innovative strategies, our proposed approach can essentially overcome drawbacks of the conventional shape-based segmentation methods. Experimental results show that our approach can achieve much better results than state-of-the-art methods.

Shape constrained fully convolutional DenseNet with adversarial training for multiorgan segmentation on head and neck CT and low‐field MR images

“…The comparison results with the SRM constrained FC‐ResNet indicate that adversarial training and dense connectivity further benefit the segmentation networks performance the in H&N multiorgan segmentation task. The improvement in DSC using our method in comparison to the model‐based method, and the hierarchical vertex regression method is not significant, but our segmentation is two orders of magnitude faster.…”

“…In Tables and , we compare SC‐GAN‐DenseNet with five state‐of‐the‐art H&N segmentation methods based on a hierarchical atlas, an active appearance model, a patch‐based CNN, a hierarchical vertex regression method, and our previous study using SRM and FC‐ResNet . It is worth noting that the segmentation performance reported in Ref . was evaluated on the same PDDCA dataset with our proposed SC‐GAN‐DenseNet, which enables a direct comparison.…”

“…combined statistical appearance models, geodesic active contours, and multiple atlases to segment OARs in H&N CT images. Nevertheless, without supervision, most of the model‐based methods are sensitive to initialization and insufficiently robust to intersubject shape variations of OARs …”

“…Nevertheless, without supervision, most of the model-based methods are sensitive to initialization and insufficiently robust to intersubject shape variations of OARs. 11 Recently, methods based on deep learning, particularly the convolutional neural networks (CNNs), have demonstrated the potential for medical image segmentation, target detection, registration, and other tasks. [12][13][14][15][16][17][18][19][20] Specifically for H&N OARs CT segmentation, Ibragimov et al 21 modeled the task as multisegmentation subtasks and trained 13 CNNs.…”

AnatomyNet: Deep learning for fast and fully automated whole‐volume segmentation of head and neck anatomy

“…19 Instead of aligning images to a fixed set of exemplars, learning-based methods trained to directly segment OARs without resorting to reference exemplars have also been tried. [20][21][22][23][24] However, most of the learning-based methods require laborious preprocessing steps, and/or handcrafted image features. As a result, their performances tend to be less robust than registration-based methods.…”