Login |
 
 

Overview

Hours per week : 4

Credits : 6

Participants: Min: 2 Max: 18 Expected: 14

Type of course: Seminar  (Bachelor/Master)

Language: english

Content

The availability of automated microscopy and fast imaging methods in combination with fluorescent protein labeling produce a huge amount of n-dimensional image data. A manual analysis of this image data is very time-consuming and hardly reproducible. Therefore, an increasing demand for algorithms to quantify biological image data can be observed. Reliable algorithms for (semi-)automatic quantification of large-scale movie or single image data are required. Computational methods such as standard image processing or machine learning approaches are therefore investigated in this seminar course.

The seminar course is divided into two parts, theory and practice:

Theory:
In a 25min talk the algorithm is presented and advantages/disadvantages in the field of scientific image processing are discussed.

Practice:
The algorithm has to be implemented in Java, with the help of the KNIME – Image Processing framework and ImgLib2.

Registration via StudIS required.


Preliminary discussion of the topics: Tue 16.04.2013, 13:00 - 14:00 in Z 816 (kitchen) 

Links

KNIME Image Processing Framework:
tech.knime.org/community/image-processing

ImgLib2:
imglib2.net


Topics

Cell Lifecycle Classification
Spot-Detection
Tracking
Feature Set Evaluation
Thinning
Matching
Rule-based Cellclump-Splitting
TurboPixels
Region-Merging - Segmentation Pyramids
Variogram-based Segmentation
Rolling Ball
Colocalization
Hough-Transforms
Stiching
Image Stabilization / Registration