Machine Interpretation of Medical Images Using Deep Learning

Abstract

We come across different biomedical images. It is difficult to interpret those images as they do not have any description. Image captioning is the process of generating textual description from an image which depends on the object and action in the image. With the advancement in deep learning techniques, we will build models to generate captions for biomedical images. This model will be very useful to accelerate the diagnosis process by telling the abnormalities present in the image. The model will be based on an encoder-decoder framework along with an attention model. The encoder will be using deep CNN to extract image features and the decoder will be using transformers to generate captions. Caption generating involves different complex scenarios starting from collecting the data set, training the model, validating the model, creating trained model to test the image, detecting the image and generating the captions