The Institute of Corpuscular Physics (IFIC), The Institute of Corpuscular Physics (IFIC), a joint centre of the University of Valencia and the CSIC, is developing a prototype for the deployment and validation of radiodiagnostic aid systems based on Artificial Intelligence models, which will facilitate the interpretation of clinical findings and evidence in radiological images both in public hospitals and in health centres and small private clinics.
Named DIRAC, the project is led by the IFIC, which collaborates with the FISABIO Foundation and the Institute of Telecommunications and Multimedia Applications (iTEAM), of the Polytechnic University of Valencia (UPV). Besides, it has funding of the Valencian Agency of the Innovation (AVI), in the frame of the announcement of helps in competitive concurrence corresponding to 2020.
DIRAC proposes a methodology based on mini-controllers that facilitate the implementation of Artificial Intelligence (AI) models, capable of interoperating with medical imaging formats in an autonomous, secure and efficient way. The incorporation of this technology into the healthcare sector will not only improve productivity and the accuracy and interpretation of the medical image, but will also reduce operational costs.
Currently, AI models for the automatic detection of pathologies from chest radiographs are already being developed. Specifically, and within the framework of the pandemic, algorithms have been designed and integrated that detect lung disease by COVID-19 early, in the framework of a project of the Carlos III Institute of Health that IFIC has also been a beneficiary of.
To carry out this initiative, the IFIC has made use of one of the largest collections of radiological images labelled in the world, which belongs to the Medical Imaging Bank of the Valencian Community, managed by FISABIO. Also, for the efficient training and validation of deep learning models and Artificial Intelligence it has Artemisa, the computer platform of the IFIC, of high capacity of calculation and dedicated to the development of these models.