Neuromorphometric signatures of diseases affecting the brain
Title: Neuromorphometric signatures of diseases affecting the brain
DNr: NAISS 2024/5-351
Project Type: NAISS Medium Compute
Principal Investigator: Rolf A Heckemann <>
Affiliation: Göteborgs universitet
Duration: 2024-07-01 – 2025-01-01
Classification: 20603


Various conditions -- brain diseases as well as systemic diseases -- affect the human brain's structural configuration. For each condition, this raises the question whether quantifying such changes through morphometric analysis of brain images can yield biomarkers for predicting diagnosis, disease progression, and therapy response. I have developed image segmentation software (MAPER and Pincram) that automatically determines morphometric quantities on magnetic resonance images of the brain. In several collaborations, clinicians acquire images as part of clinical studies and transmit them to me for analysis, generally with the goal of identifying imaging biomarkers that might be of interest in the context of the respective disease or clinical specialty. With this application, I seek an allocation of computation time to conduct processing of images from such collaboration studies. The allocation I am requesting is larger than usual because of the specific juncture that the project is at: algorithmic improvements to MAPER have been achieved, which creates the necessity (and opportunity) to create a new normative reference morphometry database using studies where healthy volunteers have been scanned (IXI, The Human Connectome Project, etc.). Some subprojects depend on this reference database particularly strongly, because no control group has been scanned as part of the study (e.g. ESOP -- Epilepsy Surgery Outcome Prediction (PI Prof. Bertil Rydenhag), NeumRA (PI Prof. Maria Bokarewa, GU). Images in these projects need reprocessing with the new MAPER version and reanalysis based on the new reference database. Other subprojects will require image segmentation for the first time: in LIM (Less is More -- a multicentre study on extremely preterm-born infants), we had to find a way to correct motion artefact in the images first. Now that this problem is solved, the computationally intensive morphometric analysis of the included images will begin. A completely new effort with anticipated large computational demands is the GroundTruth project, where we will create and validate (e.g. through leave-one-out cross-validation) a novel type of atlas database (source of training data). In addition, I am planning to carry out research projects (master's projects and similar) for methodological assessment and improvement.