Can deep learning and explainable AI approaches support the diagnosis of dementia? Results from a proof-of-concept study

Veranstaltung der UMG

On May 20, 2026 PD Dr. Martin Dyrba, who focuses on deep learning explainability in his research at DZNE, will give a lecture as part of the Late Lunch Seminars on the question "Can deep learning and explainable AI approaches support the diagnosis of dementia? Results from a proof-of-concept study".

Alzheimer’s disease and other types of dementia have a specific atrophy profile that can be determined from MRI scans and automatically evaluated by current deep learning models. However, interpreting MRI findings clinically and associating them with disease processes remains challenging. In the DFG-funded EXPLAINATION project, PD Dr. Dyrba and his group developed strategies to improve model explainability, and evaluated the clinical utility of various types of explanations.

The Late Lunch Seminars, the Medical Faculty’s scientific lecture series, aim to promote networking across a wide range of research topics. The talks include presentations from local scientists and external project partners as well as inaugural lectures by newly appointed professors. The lectures are primarily aimed at the scientific staff of University Medicine, the University and non-university research institutions in Greifswald.

The talk is open to the university public and will be held in English. Registration is not required.

Contact
Universitätsmedizin Greifswald KöR
Dekanat/Wissenschaftlicher Vorstand
Fleischmannstr. 8, 17489 Greifswald
Telefon +49 3834 86 5012
dekanatmed.uni-greifswaldde

 

Organization

  • Dekanat Universitätsmedizin

Location

  • Universitätsmedizin, Hörsaal Nord, Ferdinand-Sauerbruch-Straße
    Universitätsmedizin, Ferdinand-Sauerbruch-Straße, 1
    17489 Greifswald