CCG/ZGDV participates in a health project where AI technologies lead the way in detecting neoplasms and lung cancer

The PRR HfPT - Health from Portugal Agenda aims to create and develop advanced solutions for health markets, based on innovation and technology, to make our country a world reference hub in this sector.

One of the focuses of this project is developing and disseminating a telematics platform with decision support and distributed review for clinical pathology imaging studies for detecting neoplasms and lung cancer. Currently, figures indicate that 15 percent of deaths from this disease are caused by lung cancer. In addition to this context, the reduced number of pathologists, the lengthy, demanding, and error-prone process of detecting neoplasms, and the increase in requests from these clinical services, require speed in finding solutions to these problems.

This is where "deep learning" comes in, a branch of Artificial Intelligence in which computers, using large amounts of data, recognize, classify, describe, and perform countless tasks. In this specific case, deep learning integrates an intuitive and interactive interface platform that allows healthcare professionals to examine histopathological images in detail, through the ability to zoom in on specific areas, adjust contrasts, and efficiently navigate large data sets, thus guaranteeing a thorough and accurate analysis.

 

PathoBox: the new paradigm in collaborative telepathology

The CCG/ZGDV Institute, together with BMD Software and FMUC (School of Medicine of the University of Coimbra) have already developed this project, since it is the continuation of another - iPATH, where the aim was to detect mitoses in the diagnosis of breast cancer. The platform used in both projects is the same - PathoBox: a centralized archive for digital pathology images, 100% web-based, where the doctor can draw up a complete clinical report, from the image acquisition stage to remote clinical review, using the intelligent tools available in the system.

The CCG/ZGDV is responsible for gathering requirements for intelligent decision support tools and implementing computer vision and deep learning technologies. In this case, the main tasks involve: segmenting the lung tumor; identifying tumor cells versus intratumoral stroma; and detecting and classifying different abnormal cells.

The HfPT project includes 88 co-designers, who aim to launch 99 new products, with a total investment of 90 million euros. The CCG/ZGDV participates in four WorkPackages: WP3 - Sensorisation and Robotics; WP4 - Data Integration, Interpretation and Sharing; WP5 - Digital Consultation/Telemedicine; and WP16 - Health Business Portugal.