Many of the phenotypic variations for which we'd like to screen
can be very difficult to score by eye. Some of our screens rely
on subtle changes in cell morphology and staining intensity. Others
involve repetitive and time consuming tasks like object counting
and classification.
Computers are very good at quantitation of subtle variations in
images and even better at doing repetitive tasks. We have developed
a number of microscope-based assays that rely on computational image
analysis.
By converting visual information to numbers, we can rapidly and
with high precision assess the effects of small molecules on large
numbers of cells. Quantitative analysis eases the comparison of
replicate assays and allows us to much more precisely assign degrees
of strength to an observed perturbation, which is necessary for
meaningful dose-response and structure activity analyses.
In general, we believe that new techniques for the automated analysis
of digital images have the potential to reveal fundamental new cell
biology and gives us new insights into the interactions of small
molecules with cells. The ICCB/ICG is working with the
Open Microscopy Environment (OME)
project based at MIT to help develop tools for automated
analysis of images produced in high throughput screens.
|
