The work package on machine learning will develop data analysis methods in support of the other work packages. The aim is to use expertise in artificial intelligence to help solving the medical research questions in the 180° North project.
Our work focuses in particular on how to fuse the information from data acquisitions with PET, MR and CT scanners. The individual imaging modalities provide insight in different aspects of the patient’s anatomy and physiology, but together they offer a more complete picture and can help reveal the underlying medical truth. In this context, we are particularly interested in developing methods for accurate translation between the medical image modalities (PET, MR and CT). By converting data from one image domain to another, we may complement the patient information with image modalities that have not been recorded, which may help to detect changes, anomalies and relevant artifacts in the images with a sensitivity that a human interpreter – meaning a qualified medical expert – cannot provide.
We will also address well-known physical, biological and technical limitations in preclinical PET imaging that lower the quantification accuracy during kinetic analysis in imaging of small animals, by developing novel PET quantification methods using machine learning. We will further use established machine learning methods, as well as own developed techniques, to detect, visualize, and quantify changes in the tumor tissue automatically over time from PET image data. This will be necessary in order to quantify subtle changes in the treated tissue at early time points after therapy. Lastly, we will build machine learning models that, based on collected data, will be able to predict outcome and efficiency of a given treatment already at an early stage of the treatment.
All these objectives will be essential for both brain, lung and breast cancer work packages. The research is performed in close collaboration with Siemens Healthineers, the producer of the medical sensors at Tromsø PET Center, and other research institutions within the 180° North project and its network.
WP1: Immuno-PET and lung cancer
Despite the demonstrated effectiveness of immunoregulatory agents such as immune checkpoint blockers (ICB) on refractory cancers, these therapies work satisfactorily only in a reduced subset of patients. Further, ICB treatments are not exempt of risks and are associated to very high costs. Reliable response biomarkers are needed to identify responders and non-responders, and conventional imaging…
WP2: Radionuclide targeted therapy and imaging in Glioblastoma
The plasticity of GB tumor cells and their ability to infiltrate adjoining brain tissue limits the effectiveness of current cancer therapies. Microglia plays an important role in GB progression. Inhibition of EGFR and CSF-1R decreases microglia-stimulated invasion of GB cells. Specific radiopharmaceuticals targeting EGFR or CSF-1R will be developed (WP3/Bergen) and applied as (i) diagnostic…
WP3: Phagocyte targeting in breast cancer
In spite of an enormous global research effort, astonishing preclinical cancer cures, and the approval of multiple formulations, nanomedicine’s impact on cancer patient care remains limited. Recently, it is becoming evident that this unsatisfactory exploitation may be tackled by considering nanodrugs’ extensive interaction with the immune system. Moreover, our collaborators recently demonstrated these interactions can…
WP4: Advancing oncological PET imaging using machine learning
The work package on machine learning will develop data analysis methods in support of the other work packages. The aim is to use expertise in artificial intelligence to help solving the medical research questions in the 180° North project. Our work focuses in particular on how to fuse the information from data acquisitions with PET,…