Speaker: Antonia Harms, University Hospital Bonn (grid.15090.3d)
Title: BrainPrint captures stage-specific lateral shape asymmetry in autoimmune limbic encephalitis
Emcee: Lei Zhang
Backend host: Sophia Batchelor
Details: [ Ссылка ]
Lab website: translationalneuroimaging.de/bonn
Presented during Neuromatch Conference 3.0, Oct 26-30, 2020.
Summary: Limbic Encephalitis (LE) is an autoimmune disease often associated with temporal lobe epilepsy. It is categorized by serological subgroups, that differ in clinical progress, therapy and prognosis.
Our aim is to develop accurate imaging biomarkers for clinical and scientific implementation.
We investigate lateral shape asymmetry of an ensemble of 10 cortical and subcortical structures, by applying BrainPrint, a surface-based method that uses Eigenvalues of the Laplace-Beltrami operator on triangular (boundary) and tetrahedral (volumetric) meshes as shape descriptors. The intraindividual lateral shape asymmetry of each structure is calculated as the Euclidean distance between the shape descriptors.
We are comparing lateral shape asymmetry of patients with antibodies against the glutamic acid decarboxylase 65 (GAD, n=34), leucine-rich glioma-inactivated protein 1 (LGI1, n=15), contactin-associated protein-like 2 (CASPR2, n=9) to age-matched healthy controls (HC, n=41). We divide the sero-groups by disease stage (time between first LE-associated symptoms and date of MRI) into an early stage (below or equal 24 months) and a chronic stage (above 24 months).
We find significant difference in the amygdala volume between the LGI1 patients of the acute group and the HC. Regarding the shape-asymmetry we find differences for the amygdala (LGI1 early), the hippocampus (CASPR2 early, LGI1 chronic), the caudate nucleus (CASPR2 early), the thalamus (GAD and LGI1 early and chronic) and the striatum (GAD, LGI1, CASPR2 chronic).
Our findings mirror the higher sensitivity of shape information as compared to volumetry. Since we already identified sero-specific features in volume differences in earlier studies, we now expect to refine those findings by adding information about shape. Our aim is to develop a decision tree classifier, so that in future MR-imaging analysis will contribute accurately to diagnosis, prognosis and description of progress of LE.
![](https://s2.save4k.ru/pic/WzJRr8DQBkc/maxresdefault.jpg)