From “High Content Screening” to “Robust Machine Vision Solutions ...

geckokittenAI and Robotics

Oct 17, 2013 (3 years and 8 months ago)


Presented at: ‘6
Annual Discovery Summit’ April 10-12, 2006, Monte Carlo.

From “High Content Screening” to “Robust Machine Vision Solutions” for
Genomics, Translational Science, Cloning, Bio-production, & (Stem-) Cell Therapy

Johan Geysen

MAIA SCIENTIFIC, Cipalstraat 3, B-2440 Belgium

Integrative trends in the biomedical industry have created novel R&D value chain models collectively called
“translational science”. By capitalizing on genomics and disease-specific assays, the translational science
approach strives for more efficient development of highly specific therapies for targeted, well-described
patient groups. Translational science has created an increasing demand for the integration and flexible
adaptation of disease-specific animal and cell-based compound screening models. Also in pharmaceutical-
and bio-production, there is a similar need for more flexible processes to produce a wider range of products
for smaller patient populations. In (stem-) cell therapy, where donor- and patient-derived cells & tissues are
the therapeutic effector, production ultimately aims at the individual patient.
Non-invasive, label-free bio-imaging applications provide a unique way to meet these needs. Robust
machine vision solutions for label-free automated cell counting, confluence, clonality and colony size
assessment have been developed for the frequently used adherent cell lines CHO, HEK-293, 3T3, E11,
HeLa, for the suspension cell lines, CHO-K1SV, U937, HL60, THP-1 and for hybridoma. Full well image
capture and image analysis can be executed in less than 15 minutes and with single image/well capture in
less than 5 minutes (96-well plate). These applications were shown to be effective in supporting the
execution of 24hour / 7day unattended cloning and hybridoma selection as well as automated clone transfer
and scale-up production processes inside the cell culture and cloning robot “Cello”.
Similar object identification principles have now been developed for primary cell cultures derived from
donors and patients: epidermal keratinocytes, dermal fibroblasts and endothelial cells, peripheral- and cord-
derived blood; dendritic cell cultures amongst others. The applications were shown to be cost-effective, fast
alternatives for surrogate biochemical endpoint assays in high content screening. Because of being non-
invasive, they were successfully applied as ancillary endpoints for cell proliferation and cell viability in
biochemical, expression and differentiation screening. The image documentation tools were shown to
provide valuable information for segregating hits from false positives, whereas the image analysis result
allowed for the correction of screening result for the effect of individual compounds on the actual cell
number in each well.
Starting from the aforementioned know-how, current developments at MAIA SCIENTIFIC include novel
applications for donor- and patient-derived dendritic-, T- and NK cell cultures and their co-cultures in novel
organotypic environments that can improve the outcome of cancer vaccination processes. They will
collectively support unattended cell therapy processes under GMP conditions and will reduce the number of
interferences needed with the patient-derived sample on its way from donor/patient, over co-culture to
cloning and ultimately the cancer vaccine for the patient.
Examples from the above and from other projects will be shown to illustrate how increased productivity and
higher process quality in automated work environments can be achieved in the bio-therapeutics, tissue-
banking, and (stem-) cell therapy industry.