Our vision - MIRA
In 2019, nearly 85 000 medical imaging exams using CT, MRI, PET and ultrasound were performed at St. Olavs hospital. At the same time there is a huge demand for solutions in digital pathology to interpret and improve robustness in diagnostics of whole slide images. In clinical practice, image interpretation still depends on manual handling and personal judgement by experienced radiologists and clinicians, which is a limited and cost-intensive resource.
AI and machine learning constitute the next game changer in medical imaging, with potential benefits along the whole chain of events from image acquisition to medical decision. Robust and transparent AI systems based on medical imaging and clinical information can enable accurate and time-efficient diagnosis, treatment selection, and follow-up of for many patient groups. The outcome will be cost-effective clinical tools to prepare the hospitals for future patient loads, while at the same time improving patient care and quality of life.
Challenges to successfully develop sustainable AI solutions are still present. A major issue is access to relevant data. Machine learning and AI also require Big Data, and to enable useful and generalizable end products, the data should preferably be of multicenter origin. Regulations on data protection and privacy in the EU/EAA (GDPR) is an important regulative for processing of personal data in general, but it also poses hinders for data sharing and analysis, which needs to be tackled. Safe data storage and access to computational power within safe environments are other issues. Moreover, exaggerative claims from poorly designed AI studies have added significantly to creating an AI hype.
It is our responsibility to properly validate new AI tools using independent test sets and clinically relevant questions. The medical imaging environment in Trondheim has several opportunities within this market space. There is a long tradition for medical technology research and innovation, where the integrated university hospital has a central position. MIRA is focusing on AI applications for MR, ultrasound and CT imaging, where Trondheim-based research groups are in the forefront nationally and internationally, but also on applications with a clear synergy potential such as digital pathology. Our overall aim is to provide innovative solutions through image analysis that benefit the patient inside and outside the hospital. By merging domain-specific knowledge from medicine, image technology and AI, we will create an innovative environment with the capacity to validate and implement new AI frameworks for use in the daily clinical arena.