Deep Radiomics Decision Support System for Prostate Cancer Management

Non-invasive medical imaging, particularly MRI, has become essential in prostate cancer management, providing quantitative, multi-dimensional, and multi-parametric image data. Clinical evaluation and interpretation of radiological images thus form a crucial aspect of disease management pathway. Presently, this task is primarily performed manually and qualitatively by experienced radiologists following standardized guidelines. While this has improved healthcare outcomes, manual evaluation and interpretation of images for clinical decision-making pose bottlenecks in clinical practice. It is cumbersome, subjective, time-consuming, and underutilizes the available quantitative data, sometimes leading to under- or over-diagnosis. Moreover, the disproportionate scarcity of experienced radiologists compared to the increasing demand for MRI examinations, driven by a growing patient population, poses scalability challenges for future healthcare needs. There is, therefore, an unmet clinical need for optimized and automated radiological workflows that are reproducible, objective, scalable, and capable of exploiting the full spectrum of clinical information embedded in images for improved patient care. This has triggered clinical and commercial interest in AI decision support systems for radiological applications, especially prostate cancer management. The CIMORe group at ISB, NTNU, has recently developed a cutting-edge AI decision support tool that analyzes MRI scans and clinical data to provide quantitative metrics for prostate cancer detection and stratification. Unlike black-box AI, this approach offers transparent and comprehensible insights to clinicians. Through the qualification grant funding from RCN, we explored the commercialization potential of the technology, including market analysis and business models, as well as technical aspects such as mode of delivery, interoperability, and user-friendliness, along with regulatory considerations for technology deployment. As part of the commercialization investigation, the project participated in the REACH incubator program in Silicon Valley, USA, organized by the Nordic Innovation House, Palo Alto, specifically tailored for Nordic research project or startups. On the technical front, a proof-of-technology clinical testing has been initiated at the St. Olavs Hospital, Trondheim.

In this qualification grant project, we explored the commercialization potential and technical aspects of delivering an AI decision support system for more accurate and efficient detection of prostate cancer. This system was developed by the Cancer Imaging and Multi-Omics Research (CIMORe) Group at ISB, NTNU. As part of this endeavor, proof-of-technology clinical trials have been initiated at St. Olav Hospital in Trondheim. At the research level, this project has equipped us with the expertise to transfer and implement research projects or ideas into tangible products or services. On a societal and patient level, the clinical integration of the AI decision support system holds promise for enabling early detection and accurate diagnosis of prostate cancer, leading to improved treatment outcomes and enhanced quality of life for patients. Moreover, leveraging AI in radiology could streamline the prostate cancer treatment pathway, ultimately reducing costs. Addressing global health challenges is a key priority of the UN Sustainable Development Goals. Lack of adequately skilled healthcare personnel and appropriate technology pose significant obstacles to achieving this goal. Our project has the potential to contribute to addressing these challenges. From a global health perspective, the utilization of AI for prostate cancer diagnosis could help narrow the gap in disease care, particularly between developed and underdeveloped countries. In regions with limited resources and personnel, such as Sub-Saharan Africa, where there are roughly 2 radiologists per million people compared to 116 per million in Europe and North America, AI-based decision support systems could play a crucial role in facilitating and augmenting the work of radiologists. Furthermore, the integration of AI decision support systems into radiological decision-making processes will create opportunities for implementing novel approaches to data analysis and radiological image interpretation, potentially leading to the development of new skills and job opportunities. This could significantly impact the training of future radiologists