Vicomtech and Inycom have developed an Image Processing and Artificial Intelligence support system to detect several pathologies
• The system has been developed with the Galician Health Service, SERGAS • This solution based on Artificial Intelligence allows the specialist to carry out image analysis on radiology and digital pathology studies. • Currently, work is being done on testing the application in a reference hospital.
Vicomtech and Inycom have combined their capabilities to develop a support system for the detection of various pathologies based on Image analysis and Artificial Intelligence. The results of the CADIA project, which has been carried out for 15 months, have been validated in the case of breast cancer screening and diagnosis. Currently, work is being done towards the adoption of the system and its integration into the workflow of radiologists and pathologists as an additional tool to support decision-making during screening and diagnosis.
The system will be integrated with the applications of the Galician Health Service (SERGAS) systems, allowing the specialist to perform the image analysis on the selected radiology and digital pathology studies and view the results obtained by the algorithms. It integrates Artificial Intelligence models for detection and characterization of masses in mammograms and digital tomosynthesis (DBT), tissue segmentation for evaluation of breast density in mammography and DBT, and diagnosis from digital images of pathological anatomy.
The models developed have been trained from databases generated with suitably selected and annotated imaging studies. The radiology models contemplate SERGAS studies from four different image matrices. For its part, the pathology model has been trained with digitized images within the framework of the project from samples from different hospitals.
This project has brought together the key participation of two groups of specialists. Specifically, 15 radiologists and 15 pathologists have participated and collaborated both in the annotation of the images, as well as in the definition of requirements and in the validation of the demonstrator.
The next steps include the resolution of challenges that Vicomtech and Inycom are currently addressing, such as the improvement of the specificity and sensitivity of the obtained models or the refinement of the application interfaces, as well as the integration into SERGAS workflows for testing the system in a referral hospital.
The CADIA project has been developed by the UTE formed by Vicomtech and Inycom for these purposes, in response to a SERGAS Innovative Public Procurement procedure within the Codigo100 Healthcare Innovation Plan, financed 80% by the Pluriregional Operational Program of Spain FEDER ( POPE) through the collaboration agreement with the Ministry of Science, Innovation and Universities.