High Content Cellular
Screening
CSIRO is undertaking advanced image analysis research and
development for rapid screening of multi-band fluorescence microscope
images of cells for high content, high throughput screening for drug
discovery.
By pursuing disease targets and drug candidates rich in biological,
chemical and metabolic information, scientists and companies can make more
accurate choices about the most promising leads to move into clinical
development. Cell-based fluorescence assays can probe a wide range of cellular
functions to enable improved efficacy and toxicity assessment in a high
throughput format, including:
- Changes in morphology and the cytoskeleton
- Cellular differentiation
- Cell-cell interactions and adhesion
- Apoptosis
- Chemotaxis and motility
- Spatial distribution changes such as receptor trafficking,
translocation of signalling molecules and complex formation
Our Value
Our technology is based upon automated image analysis. Images of single
or multi-cell preparations will be processed, in a few seconds, to provide
accurate, reliable and meaningful quantitation of compound or gene
knockout effects on cell
structure and morphology. It draws on enormous depth of intellectual
property in advanced mathematical and colour morphology, including
segmentation of moving images. Our core expertise is in advanced methods
for image segmentation, suitable for image processing applications
involving complex microstructure.
Our proprietary algorithms are realized in a rigorously engineered
library, proven in multi-platform applications. This library delivers
significantly more capability than shrink wrapped image analysis products.
Our algorithms have been tested in successful high content, high
throughput screening projects.
Screening and Pathway Analysis Applications
Some examples of applications include:
Most examples of automated high content screening are commercially
sensitive. The images we present here illustrate our technological
capabilities in this area. Our capacity to handle subtle and demanding
changes in image morphology is illustrated in the area of automated melanoma
detection.
Cell Counting
|
 |
 |
|
A typical neuron cell image captured using three fluorescence channels. The nuclei are
shown in the blue and the cytoplasm in the green channels. |
Our
cell counting algorithm is better able to separate touching cells
than standard methods. |
Neurite Detection for
single channel images
|
|
|
|
|
A
typical neuron cell image captured using a single fluorescence
channel.
|
Standard
solutions just measure neurite number & length.
|
Our
solution measures the more biologically relevant complexity of
neurite branching.
|
Neurite Detection
for two channel images
|
 |
 |
|
A
typical neuron cell image captured using two fluorescence channels
– blue for nuclei, green for cytoplasm and neurites. |
Our
solution measures the complexity of neurite branching even in images
where the cells are densely populated and irregularly sized. |
Nucleus-Cytoplasm translocation for commonly used cell
lines
|
 |
|
|
|
A typical cytoplasm channel image with densely
packed cells from a common cell line. When well separated, these
cells are roughly elliptical in shape and the cytoplasm is
symmetrically located around the nucleus. |
In widefield microscopy, it is difficult to
segment the cytoplasm directly. So cytoplasm measurements are
usually taken using a surrogate for the cytoplasm - an isotropic
"donut" region around the nucleus. |
When cells are densely packed, isotropic donuts
will include some background. Our solution allows the use of
anisotropic "donuts". This reduces the likelihood that low
signal background will bias the cytoplasm measurement. |
Nucleus-Cytoplasm
translocation for more difficult primary cells
|
 |
 |
|
A typical cytoplasm channel image with densely
packed primary cells. Even when well separated, these cells are not
elliptical in shape and nor is the cytoplasm symmetrically located
around the nucleus. |
We have developed a new cytoplasm surrogate using
adaptive donuts. These expand within the cytoplasm differentially to
avoid the background. This significantly reduces the false alarm
rate from incorrect cytoplasm measurements. |
Cytoplasm-vesicle
translocation
 |
 |
 |
| A
typical cytoplasm-to-vesicle translocation assay – a negative
result is shown above, a positive result is shown below. |
Standard
solutions just count the vesicles/dots in the full image. So
heterogeneous responses cannot be detected. |
Our
solution allocates each vesicle/dot to a single cell. So atypical
cells with high or low response can be detected. |
 |
 |
 |
Cell Phenotyping for
orphan protein or gene knockout studies
|
Compartments below (left to right): ER - Golgi - Actin -
Mitochondria - Nucleolus |
|
 |
|
Compartments above (left to right): Lysosome - Tubulin - Golgi
body - Transferrin receptor - Nucleus |
Having fluorescently tagged the major organelles in a cell, we extract
image analysis features which quantify the spatial appearance of each
organelle. These features are insensitive to position and rotation of the
cell and are designed to have discriminating power and to be meaningful to
biologists. We then use statistical classification methods to classify:
- a known organelle's appearance into normal or abnormal classes -
gene knockout experiments, or
- an unknown protein into one or more normal organelles - in orphan
protein experiments.
Commercial Partners
-
Molecular Devices (formerly Axon Instruments) (http://www.axon.com)
are a major biotechnology instrumentation company that CSIRO has worked
with in the development of their
ImageXpressTM cellular screening system.
-
Atto Bioscience (now BD Biosciences)
specializes in technologies for live cell-based assays and affiliated
technologies including confocal high-throughput imaging.
The company has licensed CSIRO's neurite
outgrowth detection software to run in
the AttoVision software for its Pathway HT cell screening system.
-
Evotec Technologies GmbH (http://www.evotec-technologies.com)
is a supplier of innovative tools and technologies for life sciences and
pharmaceutical drug discovery. The company has licensed CSIRO's neurite
outgrowth detection software to run in
the Acapella software for its Opera
ultra high-throughput cell screening system.
For further information, please contact:
Pascal Vallotton
Leader, Biotech Imaging
CSIRO Mathematics, Informatics and Statistics
Locked Bag 17, North Ryde NSW 1670 AUSTRALIA
Phone: +61 (0)2 9325 3208
Fax: +61 (0)2 9325 3200
Email: pascal.vallotton@csiro.au
|
Leanne Bischof
Biotech Imaging
CSIRO Mathematics, Informatics and Statistics
Locked Bag 17, North Ryde NSW 1670 AUSTRALIA
Phone: +61 (0)2 9325 3206
Fax: +61 (0)2 9325 3200
Email: leanne.bischof@csiro.au |
|
To top