Zu den Inhalten springen

FOD processing

PDE-based enhancing of Fiber Orientations

The accurate and reliable processing of diffusion-sensitized magnetic resonance images is a major prerequisite for tractography algorithms or any other derived statistical analysis of diffusion weighted measurements. In this project we formulate the general principle of fiber continuity (FC) which can be applied to a wide range of problems involving the processing of high angular resolution diffusion images. The idea is based on the simple observation that the imaging of fibrous tissue implies certain expectations to the measured images.

FC-based FOD reconstructions can improve the resolution of crossing configurations compared to ordinary CSD and isotropically regularized CSD. Results are obtained on the physical phantom by Poupon et al.

Depending on regulrization strength also difficult fiber bundles can resolved. Obtained tractograms usually get more coherent the stronger the FC influence is.

  • Marco Reisert, Elias Kellner, Valerij Kiselev
    “About the Geometry of Asymmetric Fiber Orientation Distributions”
    IEEE Trans Med Imaging. 2012 Jun;31(6):1240-9. doi: 10.1109/TMI.2012.2187916. Epub 2012 Feb 15.
  • Marco Reisert, Valerij Kiselev
    “Fiber Continuity: An Anisotropic Prior for ODF Estimation.”
    IEEE Transactions on Medical Imaging, 2011, June 2011, Volume: 30, Issue:6, On page(s): 1274 – 1283
  • Marco Reisert and Henrik Skibbe
    “Fiber Continuity Based Spherical Deconvolution in Spherical Harmonic Domain”
    Medical Image Computing and Computer-Assisted Intervention – MICCAI 2013, Lecture Notes in Computer Science Volume 8151, 2013, pp 493-500

STA toolbox

A toolbox for processing orientation distributions in spherical harmonic domain. The STA tool box can be downloaded from here.

With the advent of novel 3D image aquisition techniques their efficient and reliable analysis becomes more and more important. In particular in 3D the amount of data is enourmous and requires for an automated processing. The tasks are manifold, starting from simple image enhancement, image reconstruction and description to object/feature detection to high-level contextual feature extraction. One important property which most of the tasks have in common is their covariance to rotations. Spherical Tensor Algebra (STA) offers a a general framework to fulfill these demands. STA transfers theories from mathematical physics and harmonic analysis into the domain of image analysis and pattern recognition. The main objects of interest are orientation fields. The content of the fields are arbitrary, they may contain local image descriptors, features, orientation scores or filter reponses. STA deals with the processing of such fields in the domain of the irreducible representations of the rotation group.

Features extracted via the STA toolbox can be used for landmark detection. Here an example how a simple random forest classifier can classify/detect certain landmarks/ROIs based on a T1-weighted image.

Steerable deconvolution can enhance drastically the 'lineness' of tubular structures. Here maximum intensity projections of neurites in a fly brain are shown. In the original acquisition dendritic processes/spines dominate the actual dendrites, in the processed image the dendrites are clearly visible.

  • Henrik Skibbe and Marco Reisert
    “Rotation Covariant Image Processing for Biomedical Applications”
    Computational and Mathematical Methods in Medicine, Volume 2013, ID 931507, dx.doi.org/10.1155/2013/931507, Special Issue on Mathematical Methods in Biomedical Imaging
  • Henrik Skibbe and Marco Reisert and Thorsten Schmidt and Thomas Brox and Olaf Ronneberger and Hans Burkhardt
    “Fast Rotation Invariant 3D Feature Computation utilizing Efficient Local Neighborhood Operators”
    IEEE Trans. on PAMI 2012, 10.1109/TPAMI.2011.263
  • Henrik Skibbe and Marco Reisert
    Detection of Unique Point Landmarks in HARDI Images of the Human Brain”
    accepted for presentation at the CDMRI, MICCAI 2012 workshop on computational diffusion MRI
  • Henrik Skibbe, Marco Reisert and Hans Burkhardt
    “SHOG – Spherical HOG Descriptors for Rotation Invariant 3D Object Detection.”
    In Proceedings of the 33rd international conference on Pattern recognition (DAGM'11), Rudolf Mester and Michael Felsberg (Eds.). Springer-Verlag, Berlin, Heidelberg, 326-335. 
  • Henrik Skibbe and Marco Reisert 
    “Dense Rotation Invariant Brain Pyramids for Automated Human Brain Parcellation.”

    In Proceedings of the Informatik 2011, Workshop on Emerging technologies for medical diagnosis and therapy, Berlin

Funding Sources

  • Deutsche Forschungs Gemeinschaft DFG RE 3286/2-1

Dr. Marco Reisert
Group Leader

Tel.: +49 761 270-93860
E-Mail: marco.reisert@uniklinik-freiburg.de

University Medical Center Freiburg
Dept. of Radiology · Medical Physics
Killianstr. 5a
79106 Freiburg