Ruprecht-Karls-Universitšt Heidelberg

HCI->MIP->Research->Segmentation of NMR Spectral Images of the Brain


Segmentation of NMR Spectral Images of the Brain


B.M. Kelm, B.H. Menze, F.A. Hamprecht

Keywords: spatio-spectral image analysis, classification

NMR spectroscopy  allows the in-vivo study of physiological and pathophsiological cell processes through the visualization of specific cell metabolites. This allows to detect tumorous tissue with very high specificity and has been applied for clinical examinations in the breast, prostate and brain.

Recent technological advances allow the recording of three-dimensional volumes of spectral NMR signals at very high resolution. Since the spectral images have to be analyzed in order to access their diagnostic information an automated approach is indispensable.

In this project an approach for the segmentation of spectral images is to be implemented and examined.  Candidates should have some programming skills, preferably in Matlab or R, and should be familiar with or interested in  statistical pattern recognition.

NMR Spectral Image of the Brain
Morphological MR image (greenish) and high-resolution NMR spectrosopic image (blue).

Project page: Classification of MR spectra

Please contact Michael Kelm or Fred Hamprecht for details.

Last update: 06.10.2010, 12:27
zum Seitenanfang