Abstract:PURPOSE:Liver tumor segmentation is a crucial prerequisite for computer-aided diagnosis of liver tumors. In the clinical diagnosis of liver tumors, radiologists usually examine multiphase CT images as these images provide abundant and complementary information of tumors. However, most known automatic segmentation methods extract tumor features from CT images merely of a single phase, in which valuable multiphase information is ignored. Therefore, it is highly demanded to develop a method effectively incorporating multiphase information for automatic and accurate liver tumor segmentation.METHODS:In this paper, we propose a phase attention residual network (PA-ResSeg) to model multiphase features for accurate liver tumor segmentation. A phase attention (PA) is newly proposed to additionally exploit the images of arterial (ART) phase to facilitate the segmentation of portal venous (PV) phase. The PA block consists of an intraphase attention (intra-PA) module and an interphase attention (inter-PA) module to capture channel-wise self-dependencies and cross-phase interdependencies, respectively. Thus, it enables the network to learn more representative multiphase features by refining the PV features according to the channel dependencies and recalibrating the ART features based on the learned interdependencies between phases. We propose a PA-based multiscale fusion (MSF) architecture to embed the PA blocks in the network at multiple levels along the encoding path to fuse multiscale features from multiphase images. Moreover, a 3D boundary-enhanced loss (BE-loss) is proposed for training to make the network more sensitive to boundaries.RESULTS:To evaluate the performance of our proposed PA-ResSeg, we conducted experiments on a multiphase CT dataset of focal liver lesions (MPCT-FLLs). Experimental results show the effectiveness of the proposed method by achieving a dice per case (DPC) of 0.7787, a dice global (DG) of 0.8682, a volumetric overlap error (VOE) of 0.3328, and a relative volume difference (RVD) of 0.0443 on the MPCT-FLLs. Furthermore, to validate the effectiveness and robustness of PA-ResSeg, we conducted extra experiments on another multiphase liver tumor dataset and obtained a DPC of 0.8290, a DG of 0.9132, a VOE of 0.2637, and a RVD of 0.0163. The proposed method shows its robustness and generalization capability in different datasets and different backbones.CONCLUSIONS:The study demonstrates that our method can effectively model information from multiphase CT images to segment liver tumors and outperforms other state-of-the-art methods. The PA-based MSF method can learn more representative multiphase features at multiple scales and thereby improve the segmentation performance. Besides, the proposed 3D BE-loss is conducive to tumor boundary segmentation by enforcing the network focus on boundary regions and marginal slices. Experimental results evaluated by quantitative metrics demonstrate the superiority of our PA-ResSeg over the best-known methods.

Robust Extraction for Low-Contrast Liver Tumors Using Modified Adaptive Likelihood Estimation

DeepRecS: From RECIST Diameters to Precise Liver Tumor Segmentation

Segmentation for CT Image Based on Improved Level-Set Approach

Segmentation of liver and spleen based on computational anatomy models

Automatic Segmentation of Liver from CT Scans with CCP-TSPM Algorithm

PA-ResSeg: A Phase Attention Residual Network for Liver Tumor Segmentation from Multi-phase CT Images

Semi-automatic Liver Tumor Segmentation with Adaptive Region Growing and Graph Cuts

3d Liver Segmentation Using Multiple Region Appearances And Graph Cuts

Efficient Liver Segmentation in CT Images Based on Graph Cuts and Bottleneck Detection

Tumor conspicuity enhancement-based segmentation model for liver tumor segmentation and RECIST diameter measurement in non-contrast CT images

Automatic 3D CT liver segmentation based on fast global minimization of probabilistic active contour

A variational approach to liver segmentation using statistics from multiple sources.

A Unified Level Set Framework Combining Hybrid Algorithms for Liver and Liver Tumor Segmentation in CT Images

Liver Tumor Segmentation Based on Level Set

A New Adaptive Method for 3 D Liver Segmentation with Active Contours

Shape-intensity Prior Level Set Combining Probabilistic Atlas and Probability Map Constrains for Automatic Liver Segmentation from Abdominal CT Images.

Fully Automatic Liver Segmentation in CT Images Using Modified Graph Cuts and Feature Detection.

A New Segmentation Framework Based On Sparse Shape Composition In Liver Surgery Planning System

Artificial Intelligence-Enhanced Couinaud Segmentation for Precision Liver Cancer Therapy

Adaptive Level Set Method for Segmentation of Liver Tumors in Minimally Invasive Surgery Using Ultrasound Images

Segmentation of Liver Tumor Via Nonlocal Active Contours