Fig. 6: The detailed architecture of the decoder.
From: A deep learning based automatic report generator for retinal optical coherence tomography images
A report in Chinese was automatically generated by the decoder based on the image features extracted by the encoder. The red box in OCT images highlights the area of “Local Cystoid Edema”. LSTM fused features from the image with the feature from the previous LSTM sequence of “macular”, the decoder then predicted the term “Cystoid Edema” at the subsequent step.
