Biomedical Engineering Reference
In-Depth Information
Fig. 3 Image of seven randomly arranged microbead populations. The seven microbead
populations were used in a DNA hybridization assay at a packing density of about 700
microbeads per mm
2
. Each microbead population can be identified by its PopID, and ligand
fluorescence (red outer circle = positive) is quantified and reported as refMFI. Microbeads of one
population are represented redundantly
with a diameter of 6-30 lm. A microbead of 6 lm creates an image with a diameter
of nine pixels on the camera sensor (pixel size 6.5 9 6.5 lm), which is a sufficient
resolution for image processing. The Nyquist-Shannon [
15
] sampling theorem is
not violated at any wavelength used, so visual artifacts are avoided. Because depth
of focus is only about 2.7 lm[
16
], auto-focusing is mandatory. This is automatically
done by software algorithms. Each image captured with the above configuration
covers an area of 670 9 900 lm. Such an image theoretically contains up to 500
single microbeads.
The intensity of fluorescence excitation is not constant, neither in time nor in the
camera field of view. To solve this serious problem for measurement, VideoScan
uses only ratios between different fluorescence channels, never absolute values. The
ratio of these dyes is specific for every microbead population. Additionally the
diameter can be used as a classification parameter. The surface of each microbead
population is modified with specific probes to detect different analytes in a sample.
The analyte is stained with a third fluorescence dye. The binding of an analyte leads
to a fluorescent halo around the microbead (Figs.
3
,
4
and
5
). The intensity is
determined by the image processing algorithm. This absolute value is referenced to
one or both encoding dyes. The resulting referenced mean fluorescence intensity
(refMFI) is independent of the device used and time point of measurement.
The measurement process is controlled by software without any user interaction
necessary. After automatic focusing, an image is captured with a filter set matched
to the first encoding dye (first encoding channel). This image is analyzed to
identify single microbeads. Because fluorescence intensities between microbeads
vary significantly due to the differences in the encoding dye concentration, an
edge-based detection algorithm is applied [
17
]. Microbeads can be identified
unambiguously by their circular shape, because microbeads are nearly perfect
spheres. Non-circular shaped or overexposed objects are excluded from further
processing, because they are considered to be artifacts, e.g., agglomerations, dust,
