After six months of hard work, we are now delivering the new features developed thanks to ELISE Open Call. The focus area we have been concentrating on is at the crossroad between Machine Learling for Computer Vision and for NLP – Natural Language Processing (namely NLG – Natural Language Generation).
Main deliverables:
- The new architecture based on Multimodal Deep Neural Network and tiles splitting for image analisys and classification
- The enrichment with a novel AI for Automatic Generation of a Medical Reports from Images, thanks to state-of-the-art NLG (Natural Language Generation) technology.
- Creation of training images data-set.
The collaboration with our Scientific Advisor allowed us to get relevant feedbacks on the soundness of our approach for the generation of the image report, that was built according to the state-of-the-art in NLG, a Transformer based technology based on an encoder-decoder approach
“ELISE Open Call was the driver for a strong acceleration of DeepMammo and a unique opportunity to scale to an international vision of the project”.