Quantitative Analysis of Muscle Fibre Sarcomers
Internship, 3 months, 50%
Students: Lei Chen, Wenxin Wang, Li Dong
Coach: Erik Franken, Somebody of Hilbers’ Group?, Bart ter Haar Romeny
Start: September 2007
Application:
Sarcomers are the basic structures of myofibrils, which make up the muscle fibres.
The overlapping area of the two molecules, active in the contraction process (actin
and myosin) can be seen as a darker band in 2-photon microscopic images of muscle
fibres. The spacing regularity is a marker for local inhomogeneities in contraction,
which can be caused by metabolic deviations (diabetes), blood supply of biochemical
responses.
Sarcomer bands.
Coherence enhancing
diffusion.
Nonlinear diffusion in the
Orientation score
In this project we need to establish useful criteria to classify ‘normal’ appearance of
sarcomer bands in healthy muscle fibres. The elongated fibrils have a dashed-line
appearance, making a quantitative analysis non-trivial.
First, an enhancement step is necessary to remove noise and to close the gaps between
the edgels. Classical methods, as coherence-enhancing diffusion, fail at crossings of
elongated structures, where orientation is locally not defined.
Recently a completely new enhancement method has been developed in the BMIA
group, based on recent findings in how orientations are measured and analysed in the
human visual system.
The Methods:
Coherence enhancing diffusion and edge enhancing diffusion are well-known
techniques in image analysis to enhance noisy images with elongated structures, such
as the images in the application explained above (but also many other biomedical
applications e.g.\ medical images with blood vessels or catheters). The standard
methods work quite well on images where the elongated structures do not cross and
exhibit relatively small curvature. However, crossing lines do occur in the application
at hand.
To overcome this problem the Biomedical image analysis group is working on nonlinear diffusion in the Euclidean motion group. An image is transformed to an
orientation score using an invertible transformation. In an orientation score,
orientation is made an explicit dimension, making it easier to perform orientationsensitive operations (see figure).
The methods are very promising but there is still a lot of room for improvements. To
learn more about how the methods work in practice, this internship project focusses
on the application described above.
Project:
Specific Requests:
 Small literature study about:
o Coherence enhancing diffusion techniques.
o The application
 Experiment with existing implementations (and implementing additional parts as
needed) of diffusion on 3D images or 2D orientation scores on muscle fibre
sacromer images.
 Build a prototype sarcomer analysis program in Mathematica and/or C++ that is
capable of statistical analysis of orientations, band-gaps and density of sarcomers.
 Tested on a series of normal muscle fibre images, and a series of pathological
fibres, such as of patients with a severe degree of diabetes.
 Write a report.
Student profile:
 In master phase of BMT (or Electrical Engineering, Mathematics, or Computer
Science.)
 Affinity with experimenting with image processing algorithms and programming
in Mathematica and/or C/C++.
Reference:
[1] http://www.deeptissue.de/english/dehn1.htm
[2] E. Franken,
Euclidean motion
(SSVM07).
[3] J. Weickert.
Verlag, Stuttgart,
R. Duits, and B. ter Haar Romeny. Nonlinear diffusion on the
group. Conference on Scale Space and Variational methods
Anisotropic Diffusion in Image Processing. ECMI Series. TeubnerGermany, 1998.