Combining Sequence Learning and U-Like-Net for Hippocampus Segmentation

Due to the difficulty in the precise segmentation of hippocampus MRI by ordinary 2D semantic segmentation network,a hippocampus segmentation method combining sequence learning and U-like-net is proposed.The U-like-net consists of two parts:the encoder and the decoder.The encoder is used for extracting and encoding image features.The decoder combines the features and outputs the segmentation mask.To further improve segmentation precision,sequence learning introduces dependency information between adjacent slices with bi-directional convolutional long short-term memory.The experiment on ADNI dataset shows that the proposed method outperforms the usual U-like-net.The visualization results of the proposed network are interpretable and accord with expert knowledge.