Random Walk and Graph Cut for Co-Segmentation of Lung Tumor on PET-CT Images

Abstract

Abstract-Accurate lung tumor delineation plays an important role in radiotherapy treatment planning. Since the lung tumor has poor boundary in PET images and low contrast in CT images, segmentation of tumor in PET and CT images is a challenging task. In this study, we effectively integrate the two modalities by making fully use of the superior contrast of PET images and superior spatial resolution of CT images. Random walk and graph cut method are integrated to solve the segmentation problem, in which random walk is utilized as an initialization tool to provide object seeds for graph cut segmentation on PET and CT images. The co-segmentation problem is formulated as an energy minimization problem which is solved by max-flow/min-cut method. A graph includes two sub-graphs and a special link is constructed, in which one sub-graph is for PET and another is for CT, and the special link encodes a context term which penalizes the difference of the tumor segmentation on the two modalities. To fully utilize the characteristics of PET and CT images, a novel energy representation is devised. For PET, a downhill cost and a 3D derivative cost is proposed. For CT, a shape penalty cost is integrated into the region and boundary function which helps constrain the tumor location during the segmentation. We validate our algorithm on a dataset which consists of 18 PET-CT images. The experimental results indicate the proposed method is superior to the graph cut method solely using the PET or CT is more accurate compared with the random walk method, random walk co-segmentation method, and non-improved graph-cut method.Index Terms-Image segmentation, interactive segmentation, graph cut, random walks, prior information, lung tumor, positron emission tomography (PET), computed tomography (CT).

Scene Composition in Augmented Virtual Presenter System

“…In this case, we consider to apply interactive segmentation methods in accurate occlusion processing, e.g. GrabCut [11], Random Walk [12][13] [14]. The result of accurate occlusion using GrabCut is shown in Fig.…”

Hierarchical forest attributes for multimodal tumor segmentation on FDG-PET/contrast-enhanced CT

“…However, two inclusive notions are gaining share to enrich image information concept: multimodality and region-based features. Multimodality [14] refers to the use of semantically different images as data sources. On the other hand, region-based features can be computed [15] for image characterization.…”

Efficient Segmentation of Lung Using FIDRM Filter and Successive Peak Quantization Transform

“…Too many FPs will render to follow the SVM classification, if the method is too unfastened. In the method [15], the random walk and graph cut methods are integrated in order to solve the problems in the lung segmentation. The random walk is considered as the initialization tool for the provision of object seeds to the graph cut segmentation on PET and CT images.…”