Automatic Segmentation of Hippocampal Subfields MRI Based on FPN-DenseVoxNet

摘要

In view of the segmentation of multi-scale spatial structure of the human brain hippocampal subfields magnetic resonance image, this paper proposes a three-dimensional fully convolution DenseNet including a feature pyramid (Feature Pyramid Networks, FPN) structure, which is simply called FPN-DenseVoxNet. The network mainly integrates the FPN structure into a DenseNet with 3D convolution kernel, which better integrates the bottom and high-level features of the neural network to realize the full use of multi-level features. In the case of deep network layers, it could still restore shallow features and restore image details, thus could finally improve the accuracy of the subdivision of the hippocampus. It also contains self-attention blocks to recover more details. This paper compares the proposed method with the existing method on two public datasets. The experimental results show that the method proposed in this paper obtains a higher accuracy on the hippocampal subfields segmentation.