You are on page 1of 1

Thesis Topic (B.Sc. / M.Sc.

Convolution Neural Network


Application for Cell Segmentation in
Biological Research
Yeast Cell
One of the primary methods through
which information is acquired from
biological experiments is by optical Department 18
microscope imagery. A common Electrical Engineering and
Information Technology
bottleneck to scientific analysis is the
Institute for
segmentation of cells or regions of Telecommunications
interest. Accurate determination of cell
Bioinspired
contours is an important step in Communication Systems
quantifying morphological features as www.bcs.tu-darmstadt.de
well as cell fluorescence intensity. A
Prof. Dr. Heinz Koeppl
convolutional neural network (U-net)
is to be adapted to segment
individual cells.
Tim Prangemeier

Recently a range of machine tim.prangemeier@bcs.tu-


darmstadt.de
learning techniques have been 06151 16 20886

applied to such data, however,


all of these still have some Christian Wildner
drawbacks. This is in particular (christian.wildner@bcs.tu-

the case when cell trap darmstadt.de,


+49 06151 16 51240)
structures are included on the
images (Fig. 1). The use of
convolutional neural networks Figure 1: Microscope image of
promises accurate segmentation yeast cells (green) immobilised
in microscopic trap structures
with relatively small training (grey) [bottom], with Thesis Topic (B.Sc. / M.Sc.)
data sets. conventionally segmented cell
contour (red) [top]. • Machine Learning
• Image Processing
• Automation
Contact Tim Prangemeier (tim.prangemeier@bcs.tu-darmstadt.de, • Cell Segmentation
+49 6151 16 20886),or Christian Wildner • Supervised Learning
(christian.wildner@bcs.tu-darmstadt.de, • Bio-application

+49 06151 16 51240) for further information on this topic.


Deutsch / English

You might also like