DeepPathFinder™
DeepPathFinder™ can lead you to deep-dive
into tumor microenviromnent
By using not only pathologist annotation but also masked data with multiple IHC as training data.
It is feasible to identify cells on H&E stained slides that are not distinguishable by human perception.
Extract information from morphological features, such as genetic mutations that is difficult to obtain without genomic testing. [Publication]
-
BRAF, KRAS and MSI-H and so on
A spatial combination of the obtained features is used to predict patient prognosis and drug efficacy. [Publication]
-
Predicting prognosis and drug efficacy from pathological features
Multidimensional features can be obtained from cell/tissue features
-
TILs count and density in stroma
-
Fibroblast count and density in stroma
-
Distance from tumor area and each lymphocyte
Quantifying IHC imaging and combination analysis with H&E image
-
HER2, ER, PgR, Ki-67
Workflow
