AERMNet: Attention-enhanced relational memory network for medical image report generation

Xianhua Zeng, Tianxing Liao, Liming Xu, Zhiqiang Wang

Abstract: Highlights•We propose an attention-enhanced relational memory network model for generating more accurate reports of medical images.•We propose the double LSTM with interaction module to enhance the context information and improve the model ability to generate medical reports.•We propose an attention-enhanced relational memory module to enhance the correlation between generated words.