Morphometry of Brain Structures

Anatomical descriptions of the macroscopic human brain are obtained by Magnetic Resonance Imaging (MRI) within a few minutes. The use of computer vision algorithms to analyze acquired data can significantly increase the information yield from these examinations and provide new insight for the underlying scientific, diagnostic or therapeutic question. The resulting quantitative characterization of structures, tissues and their changes with time can be used in further statistical studies: to classify brains, to draw conclusions about structural differences in subject groups (i.e., by gender, structural abnormalities) and to track individual changes with time (aging, CNS diseases, therapeutic interventions). Learn more about our results in the segmentation of brain compartments, texture analysis of brain structures, the description of brain shape, and the parcellation of the neocortical surface.


Quantification of Brain Disease Processes

High resolution MR images of the brain are used in clinical practice to reveal disease processes in the brain (e.g., as a consequence of head trauma, intra-cerebral haemorrhages, cerebral infarcts, tumors or inflammatory processes). Lesion properties (i.e., position, extent, density) are known to be related to cognitive handicaps of a patient. While a semi-quantitative analysis of MR tomograms based on visual inspection (e.g., rating scales) is common today in certain clinical protocols, tools for a quantitative analysis are still rare. Automatic segmentation of pathological findings is still considered a non-trivial task. Because there is no single feature that discriminates lesions from healthy tissue, approaches for lesion segmentation must be optimized towards a specific detection problem. We focused on the quantification of focal brain lesions (e.g., due to cerebral infarcts), diffuse lesions (e.g., due to inflammatory processes), small lesions (e.g., due to degenerative processes), time-dependent changes using biomechanical models (e.g., disease processes, therapeutic interventions), and the discrimination of healthy and pathological aging (e.g., Alzheimer's disease).


New Methods in Functional Neuroimaging

Cognitive processes in humans are usually studied by stimulation experiments. Responses are recorded by various neuroimaging methods, such as functional MRI, EEG, MEG, SPECT and PET, that measure different correlates of the underlying cognitive process. Because these different methods yield complementary information about the anatomical, metabolic and neurophysiological state of the brain, an integrated data evaluation is highly desirable and lead to results not achievable with a single modality. A major research topic of the SIP group is best described as "multi-modal imaging". Here, we summarize approaches for an advanced analysis of functional neuroimaging data (in fMRI and EEG, the introduction of new imaging techniques (fNIRS), as well as the combination of different imaging modalities (EEG/fMRI).


The BRIAN System

Various neuroimaging techniques map specific aspects of the brain's structure and function. Besides structural information (i.e. anatomical MRI), data from functional techniques (fMRI, PET, SPECT, EEG, MEG) are recorded for analyzing cognitive processes. A combined data evaluation is highly desirable to obtain complementary information about the anatomical, metabolical and neurophysiological state of the brain. A comprehensive processing environment called BRIAN (Brain Image Analysis) was developed to facilitate a combined analysis of neuroimaging data. The BRIAN systems consists of an interactive kernel that provides easy-to-use viewers for visualization and editors for modifying or marking data, and a large set of separate modules that implement specific processing tasks.