Comparison of U-net-based Convolutional Neural Networks for Liver Segmentation in CT
This work addresses liver segmentation for medical imaging and surgery planning, but it is incremental as it compares existing U-net variants without introducing new methods.
The study compared 2D and 3D U-net architectures for liver segmentation in CT scans, finding that slice-wise (2D) approaches achieved high accuracy with mean and median Dice coefficients above 0.97, making them preferable over 3D methods due to hardware and software constraints.
Various approaches for liver segmentation in CT have been proposed: Besides statistical shape models, which played a major role in this research area, novel approaches on the basis of convolutional neural networks have been introduced recently. Using a set of 219 liver CT datasets with reference segmentations from liver surgery planning, we evaluate the performance of several neural network classifiers based on 2D and 3D U-net architectures. An interesting observation is that slice-wise approaches perform surprisingly well, with mean and median Dice coefficients above 0.97, and may be preferable over 3D approaches given current hardware and software limitations.