Combining residual attention mechanisms and generative adversarial networks for hippocampus segmentation

This research discussed a deep learning method based on an improved generative adversarial network to segment the hippocampus. Different convolutional configurations were proposed capture information obtained by segmentation network. In addition, Pixel2Pixel was proposed. The generator codec structure combining residual and attention mechanism detailed information. discriminator used neural discriminate results of generated model that expert. Through continuously transmitted losses discriminator, reached optimal state hippocampus segmentation. T1-weighted magnetic resonance imaging scans related labels 130 healthy subjects from Alzheimer's disease Neuroimaging Initiative dataset as training test data; similarity coefficient, sensitivity, positive predictive value evaluation indicators. Results showed could achieve efficient automatic thus has practical relevance for correct diagnosis diseases, such disease.