The National Information Processing Institute (OPI PIB) has received funding to develop an innovative platform that facilitates prostate cancer diagnosis. The cutting-edge tool will rely on artificial intelligence (AI) to assist medical experts in implementing the most optimal therapies. Every year, 1.2 million people are diagnosed with prostate cancer; in Poland, it is the most common solid tumour among men. OPI PIB’s innovative platform will help doctors reduce the time needed to identify prostate lesions and reduce the number of painful biopsies. The project has been funded by the INFOSTRATEG programme of the Polish National Centre for Research and Development (NCBR), and is scheduled to launch on 1 October, 2021. OPI PIB experts will consult the medical aspects of the project with a team of doctors at the Lower Silesian Oncology Centre in Wrocław.
The vast potential of artificial intelligence
Over 30 experts at the Laboratory of Applied Artificial Intelligence (LSSI) at OPI PIB work on the application of AI in medicine. They focus on research standards, image analysis, and the development of AI models to support prostate cancer diagnoses, in addition to projects that connect new technologies and medicine. The specialists’ efforts have led to the creation of eRADS, an innovative research platform that standardises descriptions in medical reports. The tool helps radiologists evaluate the clinical relevance of lesions subjectively, and is based on the five-point PI-RADS (Prostate Imaging-Reporting and Data System) scale. The platform also enables the collection of data from examinations. Although the state-of-the-art tool is not yet commonly accessible to doctors, it is expected that this will change in light of the NCBR funding. The tool will first be implemented as an e-learning platform – then, following clinical trials, as a fully-fledged structural reporting system supported by AI algorithms.
Dr. Jarosław Protasiewicz, Head of OPI PIB said: ‘The Laboratory of Applied Artificial Intelligence has been working on applying machine learning and deep learning in prostate cancer diagnosis for many years. The research results are highly promising. I am glad that we have managed to receive funding from the INFOSTRATEG programme; we will now be able to provide medical facilities with a modern tool that will quickly and effectively identify potential cancer lesions. Although the tool is not going to replace doctors, it will certainly make their work easier. Artificial intelligence has great potential, and Poland has phenomenal IT experts. It is high time we implemented their innovations in many areas of our lives’.
Artificial intelligence vs. cancer
Researchers at OPI PIB will soon implement eRADS: an innovative tool that describes medical examination results, improves communication between radiologists and clinicians, and, ultimately, leads to higher-quality healthcare services. eRADS also supports the creation of reliable, properly described reference datasets that can be used to advance research in cancer diagnosis and treatment methods. Reliable data is key to the development of AI-based methods that could assist doctors in their diagnoses. The project’s reference database of multi-parametric magnetic resonance imaging (mpMRI) prostate imaging data—comprising a substantial body of cases supplemented with full medical documentation, including images—is unique in Poland. The database will be developed and project tasks related to medical aspects conducted alongside experts at the Lower Silesian Oncology Center in Wrocław, which was the first in Poland to implement the pilot programme of the Polish National Oncology Network. The hospital earned an accolade for its initiative in the Value Best Healthcare Dragon’s Grant & Endorsement awards, held in the Netherlands in 2021.
Piotr Sobecki, Head of the Laboratory of Applied Artificial Intelligence at OPI PIB said: ‘Prostate cancer has become a growing problem, both from the medical and social perspectives. The diagnostic methods that are becoming popular include medical imaging and, in particular, mpMRI. The evaluation of an mpMRI scan is a complex and multi-aspect task, while the quality of reporting of lesions and communication between radiologists and clinicians play key roles in implementing suitable treatment plans. With staff shortages among radiologists and the long training necessary to become an experienced specialist capable of properly evaluating mpMRI results, the use of artificial intelligence is of utmost importance. Our eRADS platform, which relies on machine learning, can deliver invaluable support to doctors’.
The first phase of the project includes the development of an initial version of the system with basic functionalities (integration with the imaging database and delivery of structural reports), which does not integrate with imaging analysis algorithms. The second phase will focus on integration with models, which will allow users to analyse mpMRI images. At the end of the second phase, a generic version of the system will be developed, which will be capable of accepting and analysing new cases – the data for which will be provided externally. In the third phase, the system will be optimised and enhanced. Additionally, a modern e-learning platform that provides information on the structural reporting of mpMRI prostate scans according to the PI-RADS scale will be developed. Educational resources will be available as a result of integration with the reference imaging database.
Dr. Rafał Jóźwiak of the Laboratory of Applied Artificial Intelligence at OPI PIB, the project’s manager said: ‘We are going to conduct the project according to an innovative concept of integration of standardised structural reports with trained machine learning methods that analyse mpMRI images. The results of the image analyses will be used to support radiologists in different aspects of their work. These will include automatic measurement of the size and calculation of the volume of prostate glands, creation of visual cues regarding the locations of suspicious lesions, evaluation of the clinical relevance of identified lesions on the PI-RADS scale, and searching for similar cases’.