New artificial intelligence algorithms facilitate diagnosis of difficult cancers

New artificial intelligence algorithms are developed to facilitate the diagnosis and treatment of cancerous tumors that are difficult to interpret clinically.

A team of European researchers and companies, led by the Universitat Politècnica de València (UPV), through the CVBLab-HUMAN-tech group, has developed innovative artificial intelligence algorithms to facilitate the diagnosis and treatment of cancerous tumors that are difficult to interpret clinically. This is one of the main results of the CLARIFY project, which is currently in its final months of execution and focuses specifically on three types of cancer: triple-negative breast cancer (TNBC), high-risk non-muscle invasive bladder cancer (HR-NMIBC) and spitzoid melanocytic lesions (SML).

In all three cases, the diagnosis is complex and challenging, which we are already addressing with this project. With these algorithms, we are taking another step forward in facilitating the interpretation of histological images and, ultimately, the diagnosis of these types of cancer for medical professionals." 

Valery Naranjo, project leader and coordinator of the Computer Vision and Behaviour Analysis Lab (CVBLab) at the UPV's Human-Centered Technology Research Institute (HUMAN-tech)

In addition to these algorithms, the project has already achieved other results of great interest, such as databases of the cancers under study, which can become valuable resources for the medical and scientific community, and advances in the secure management of data in the cloud, with requirements inherent to the health field and with several application scenarios within the project.

Excellence training and European research network

CLARIFY has also carried out a multidisciplinary training programme, which has enabled twelve young researchers of various nationalities to complete their PhD studies at one of the institutions involved in the project, learn about the latest scientific advances in artificial intelligence, cloud computing, and digital pathology and enhance new skills, which will be the key to meeting to the challenges of the engineering-medicine binomial in the coming years.

"And it has driven the launch of a European research network that brings together experts from both fields and helps to improve the exchange of knowledge, always with the same goal: to contribute to facilitating decision-making in the diagnosis and treatment of breast, bladder and skin cancers studied," adds Sandra Morales, researcher at CVBLab-I3B of the Universitat Politècnica de València.

Along with the UPV, the project also involves the INCLIVA Health Research Institute, the University of Granada, the University of Amsterdam, the Erasmus MC University Hospital Rotterdam, the University of Stavanger, the University Hospital of Stavanger, and the companies bitYoga and Tyris Software.

Final project meeting in Valencia

The project, funded by the European Union's Horizon 2020 programme under the Innovative Training Networks (ITN), will hold its final conference in just two months at the Universitat Politècnica de València, where all the results of four years of research will be presented. It will be held on 25 and 26 January at the Polytechnic City of Innovation, UPV Science Park, with the aim of facilitating the debate between the research and clinical community and laying the groundwork for future developments and collaborations.

This meeting is open to all scientific and medical communicators interested in the topics addressed by CLARIFY. Registration is free and can be made via the website https://www.clarify-project.eu/final-conference/.

The call for papers is now open

In addition, 30 November is the deadline for submission of papers related to the topics to be addressed during the conference: artificial intelligence methods applied to histological images, data infrastructure, and cloud-oriented algorithms applied to histological images, computer-aided diagnosis tools from histological images, and content-based image retrieval tools from these images. Abstracts of previously published high-quality papers are accepted and encouraged, and selected abstracts will be presented in a poster session at the conference.