Automatic pancreas segmentation from Computed Tomography (CT) images is a prerequisite of clinical practices such as cancer detection, yet challenging due to the variability in shape. To address this challenge, we propose a Hierarchical Convolutional Neural Network (H-CNN) to fuse multi-scale features, which could remedy the lost image details in progressive convolutional and pooling layers. In our proposed H-CNN, a hierarchical fusion block is designed to fuse low-level and high-level features across different layers. The H-CNN is evaluated on NIH pancreas dataset and outperforms the current state-of-art methods by achieving 86.59% ± 4.33% in terms of DSC. The experimental results confirm the effectiveness of the proposed H-CNN.
CITATION STYLE
Chen, Z., & Zheng, J. (2020). Deep neural network for pancreas segmentation from CT images. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11691 LNAI, pp. 406–413). Springer. https://doi.org/10.1007/978-3-030-39431-8_39
Mendeley helps you to discover research relevant for your work.