Hybrid Cascaded Neural Network for Liver Lesion Segmentation

Abstract

Automatic liver lesion segmentation is a challenging task while having a significant impact on assisting medical professionals in the designing of effective treatment and planning proper care. In this paper we propose a cascaded system that combines both 2D and 3D convolutional neural networks to effectively segment hepatic lesions. Our 2D network operates on a slice by slice basis to segment the liver and larger tumors, while we use a 3D network to detect small lesions that are often missed in a 2D segmentation design. We employ this algorithm on the LiTS challenge obtaining a Dice score per case of 68.1%, which performs the best among all non pre-trained models and the second best among published methods. We also perform two-fold cross-validation to reveal the over- and under-segmentation issues in the LiTS annotations.

References in corpus (8)

• The Liver Tumor Segmentation Benchmark (LiTS)
• RA-UNet: A hybrid deep attention-aware network to extract liver and tumor in CT scans
• Automatic Liver and Tumor Segmentation of CT and MRI Volumes using Cascaded Fully Convolutional Neural Networks
• H-DenseUNet: Hybrid Densely Connected UNet for Liver and Tumor Segmentation from CT Volumes
• Hierarchical Convolutional-Deconvolutional Neural Networks for Automatic Liver and Tumor Segmentation
• Bottleneck Supervised U-Net for Pixel-wise Liver and Tumor Segmentation
• Feature Fusion Encoder Decoder Network For Automatic Liver Lesion Segmentation
• 3D Anisotropic Hybrid Network: Transferring Convolutional Features from 2D Images to 3D Anisotropic Volumes

Cited by in corpus (3)

• Medical Image Segmentation Using Deep Learning: A Survey
• ASC-Net : Adversarial-based Selective Network for Unsupervised Anomaly Segmentation
• MDA-Net: Multi-Dimensional Attention-Based Neural Network for 3D Image Segmentation