MRI is an exceptionally powerful and versatile measurement technique that can provide high-resolution anatomical details (T1- and T2-weighted imaging; T1W and T2W), but also non-invasive assessment of multiple physiological parameters such as vascularization (dynamic contrast enhanced imaging - DCE), cellularity (diffusion weighted imaging – DWI) and metabolism (magnetic resonance spectroscopic imaging).

In clinical practice, the image interpretation still depends on manual handling and personal judgement by experienced radiologists, which is a limited and cost-intensive resource. The quantitative and multi-dimensional nature of the data is also underused. Our hypothesis is that robust and transparent AI (artificial intelligence) systems based on MRI and clinical information can meet these clinical objectives, thereby enabling accurate and time-efficient diagnosis, treatment selection, and follow-up of prostate cancer patients. This will result in cost-effective clinical tools to prepare the hospitals for future patient loads, while at the same time improving patient care and quality of life.

Medical imaging has become a cornerstone in all diagnostics as well as in therapy planning, and the requirement of precise imaging protocols for personalised care are increasing. At the same time, treatment options such as radionuclide therapy and proton therapy are becoming available and viable options, which also require robust imaging and multimodal reading. Providing the right treatment to the right patient at the right time is a major challenge because current imaging tools for risk assessment, diagnosis and therapy monitoring still are inadequate. PET(positron emission tomography) combined with MRI, and its potential to guide therapy, is still not common clinical routine in Norway. 

PET-MR provides complementary information for prostate cancer diagnostics

In patients with metastatic disease, imaging which measures response to systemic therapy would be a benefit. PET/MRI have this potential benefit at several timepoints through the prostate cancer disease course.

Machine learning approaches can aid in utilizing the complementary quantitative information. CIMORe aims to develop new machine learning methods to better exploit the image data from PET and MR examinations. To bring this field beyond state of the art for the betterment of patients across Norway, the CIMORe group has also taken part to facilitate multi-center clinical nuclear medicine research by implementation and development of novel methodology and analyses in the diagnosis, follow up and treatment settings.