HIGHLIGHTS
- who: . and colleagues from the University of Tu00fcbingen, Germany have published the research work: Fluorescently labeled nuclear morphology is highly informative of neurotoxicity, in the Journal: (JOURNAL)
- how: Rather than generating large labeled datasets of images though human curation and/or annotation as is usually required for training CNNs (Hughes et_al 2018 Sullivan et_al 2018) the authors used quantification of the GEDI signal directly as a classification label a technique the authors named biomarker-optimized convolutional neural_networks (BO-CNNs). Previously the authors showed that GEDI-CNN models trained against EGFP morphology signal output live/dead . . .
If you want to have access to all the content you need to log in!
Thanks :)
If you don't have an account, you can create one here.