TDA-Clustering Strategies for the Characterization of Brain Organoids
Résumé
We propose to use Topological Data Analysis (TDA) to in-
tent deciphering the morphological development of these cultures seg-
mented with U-Net. To classify various shape at three developmental
stages, we propose to combine TDA with a Kmean Clustering or with a
Support Vector Machine. We calculated some characteristics as regres-
sions on the rendered presentations to compare mean representations
from each stage. Results show a specific morphological pattern (whatever
the kind of TDA clustering) appears between 9 and 14 days correspond-
ing to the neuroepithelial formations which has to be further studied and
validated on an other dataset. Hence, our method provides an indicator
of neuroepithelial growth formation stage prediction in brain organoids,
it has to be compared with others classification methodologies.