Skip to main content

Brain image processing with Alpha-Stable distributions. Applications to intensity normalization,
segmentation and diagnosis of Parkinsonian syndrome and Alzheimer's disease

Final Report Summary - ALPHA-BRAIN-IMAGING (Brain image processing with Alpha-Stable distributions. Applications to intensity normalization,segmentation and diagnosis of Parkinsonian syndrome and Alzheimer's disease.)

The ALPHA-BRAIN-IMAGING Project has been developed in the Machine Learning and Computational Intelligence Group in the Institute of Biophysics (University of Regensburg) by the Spanish researcher Diego Salas González in collaboration with the scientist in charge Prof. Dr. Elmar W. Lang.

In this project, D. Salas-Gonzalez has developed brain image processing methods using alpha-stable distributions with applications to intensity normalization and segmentation of brain images; and diagnosis of Parkinsonian syndrome and Alzheimer's disease. The results achieved in this interdisciplinary project will definitely have applications and impact in the European society and its health, which is an objective of the '2020 Vision for the European Research Area’. Specifically, Parkinson’s disease and Alzheimer type dementia are a research priority in developed countries, as their population is becoming older and, therefore, there will be more prevalence of these neurodegenerative diseases in the future.

The main goal and overall objective of this project was to attract the attention of neuroimaging experts to the potentialities and wide range of applications of the alpha-stable distribution: a heavy-tailed, non-symmetric distribution with similar desirable properties to the Gaussian density. The Gaussian distribution has been used ubiquitously in neuroimaging. The results obtained in this project, support that the alpha-stable density can be potentially used as an alternative to the Gaussian distribution.

In addition, in order to show the wide range of application in neuroimaging of the alpha-stable distribution, four different research objectives were envisaged in this project. They included basic research, strategic research, applied research and transfer of knowledge:

i) Intensity normalization of FP-CIT Single Photon Emission Computerized Tomography brain images.
ii) Segmentation of Magnetic Resonance Images (MRI).
iii) Feature extraction for Parkinsonian syndrome diagnosis.
iv) Feature extraction for Alzheimer's disease diagnosis.

Successful results have been obtained in all of the objectives previously envisaged. Specifically, 4 journal articles and 7 conference contributions have been obtained in this period. Besides that, 2 additional journal articles are currently under preparation.

Furthermore, the successful results obtained open a novel field of research which will be definitely exploited by the researcher in the following years.