Identifying image derived biomarkers in abdominal MRI scans

My PhD research lies at the intersection of industrial collaboration and academic exploration. The central objective lies in the application of explainable AI in medical diagnosis with a focus on the liver and specifically Metabolic Dysfunction-Associated Steatohepatitis (MAS). We aim to extract biomarkers from abdominal MRI scans. Those biomarkers will help us understand the characteristics of MAS and we will create accurate models for disease progression. This endeavor not only will enhance diagnostic precision but also is important to provide non-invasive biomarkers and speed up medical practitioners’ diagnosis, which poses a challenge for them, due to the increasing workload burden, who must analyze a growing number of scans each year.

In collaboration with Novo Nordisk, our work aims to integrate explainable AI with medical expertise and to utilize 3D image data in trials to assess the effect of new treatments/drugs in a quantitative manner. This research will aim at improving diagnosis and treatment, given the ongoing rise of the population of high-risk obese patients, which are at greater risk of developing MAS. The overall project’s goal is to develop a successful biomarker identification methodology to assist MAS patients getting diagnosed with non-invasive techniques. This will help to track the progression of fatty liver disease as early as possible, when symptoms are not apparent while there is still time to reverse the stage of the disease. Moreover, this research will enable quantifying precision medicine advancements for tailored treatments.