Deep-Learning and HPC to Boost Biomedical Applications for Health (DeepHealth) project is funded by the EC under the topic ICT-11-2018-2019 “HPC and Big Data enabled Large-scale Test-beds and Applications”. DeepHealth kicked-off in mid January 2019 and is expected to conclude its work in December 2021.
The aim of DeepHealth is to offer a unified framework completely adapted to exploit underlaying heterogeneous HPC and Big Data architectures; and assembled with state-of-the-art techniques in Deep Learning and Computer Vision. In particular, the DeepHealth framework is envisioned to tackle real needs of the health sector and facilitate the daily work of medical personnel and the expert users in terms of image processing and the use and training of predictive models without the need of combining numerous tools. To this end, the project will combine High-Performance Computing (HPC) infrastructures with Deep Learning (DL) and Artificial Intelligence (AI) techniques to support biomedical applications that require the analysis of large and complex biomedical datasets and thus, new and more efficient ways of diagnosis, monitoring and treatment of diseases.
In a nutshell, DeepHealth will develop a flexible and scalable framework for the HPC and Big Data environments, based on the DeepHealth toolkit, composed of two new libraries, free-software and open-source, ready to be integrated into end-user software platforms or applications, and ready to run algorithms on Hybrid HPC and Big Data architectures with heterogeneous hardware: the European Distributed Deep Learning Library (EDDLL) and the European Computer Vision Library (ECVL), and a front-end. The libraries will be integrated in 7 existing biomedical software platforms, which include: a) commercial platforms (e.g. everisLumen, PHILIPS Open Innovation Platform, THALES PIAF, etc.); and b) research-oriented platforms (e.g. CEA’s ExpressIFTM, CRS4’s Digital Pathology, WINGS MigraineNet, etc.). The framework will be validated in 14 pilot test-beds following real world use cases and datasets in three main areas: neurological diseases, tumor detection and early cancer prediction; and digital pathology and automated image annotation. Use cases pilots will allow to train models and evaluate the performance of the proposed solutions in terms of a) time-of-pre-processing-images (toppi), b) the time-to-model-in-production (ttmip) and the time-to-train-models (totm), measured in hours, the speedup and the efficiency of parallelism. Concerning the accuracy of predictive models, standard KPIs such as error rate, sensitivity vs specificity, or just the accuracy will be used.