Automatic volumetry of the hippocampus with adaptive techniques

The small brain structure called "hippocampus" is importantly involved in memory mechanisms.  Changes in the size and/or tissue structure of the hippocampus have been demonstrated in such diverse disorders as Alzheimer's dementia, partial complex epilepsy, depression, Cushing's syndrome and chronic stress. Determination of the contour of the hippocampus from MR images is therefore of great importance in several clinical situations such as, e.g., the preoperative investigation of patients with therapy resistent epilepsy and the prediction of further cognitive decline in patients with MCI (mild cognitive disorder). However, such a determination by visual inspection of the images ("manual segmentation") is very time-consuming and, furthermore, gives different results in the hands of different specialists.

Many research groups are therefore trying to find a way of automatically delimiting the hippocampus from MRI data. This is however a difficult task since the hippocampus borders to several different kinds of tissue, and since the border is not even always visible. Within our project, several approaches have been tested; see Starck et al (2002), Pettersson et al (2006). The latter paper is a pilot study using a new non-rigid registration technique, the so-called morphon, for finding the hippocampal borders. For lack of funding, the development of our automatic method has been put on hold after that study. 

Within the project, we have also developed a new computer assisted manual method, Hipposegm, for segmentation and volumetry of the hippocampus and other brain structures. It has up til now been applied to three different clinical populations, including one with MCI and mild dementia (Eckerström et al 2007, 2008, 2010), and one where the patients had had inadvertent radiation to the basal brain because of radiation treatment of cancer in the head and neck region (Olsson et al, submitted). Several other studies are planned in detail and will start as soon as funding becomes available. The manual segmentation data from the clinical studies will be used to train the automatic method if and when we get the opportunities to develop our ideas in this field further.

The project leaders are Helge Malmgren, Department of Philosophy, GU) and Hans Knutsson (Department of Biomedical Engineeering, Linköping University). Among our project partners are Magnus Borga (same Linköping Department), Sven Ekholm (University of Rochester) and Göran Starck (Department of Diagnostic Radiology, Göteborg). The main funding for the project has come from the Swedish Research Council (VR).


M. Borga, H. Malmgren & H. Knutsson, FSED - feature selective edge detection. In Proceedings of 15th International Conference on Pattern Recognition, volume 1, 229–232, Barcelona, Spain 2000. IAPR.

G. Starck, M. Borga, M. Friberg, E. Olsson, S. Ribbelin, H. Knutsson, S. Ekholm & H. Malmgren, Fully automatic segmentation of the hippocampus in MR images. Poster presentation at ESMRMB 2002: 19th Annual Meeting of the European Society for Magnetic Resonance in Medicine and Biology. (JPEG format).

J. Pettersson, H. Knutsson & M. Borga, Normalised Convolution and Morphons for Non-Rigid Registration. Proceedings of the SSBA Symposium on Image Analysis. Umeå, Sweden: SSBA, 2006.

C. Eckerström, E. Olsson, S. Rolstad, Å. Edman, A. Wallin & H. Malmgren, The Göteborg MCI study - absolute and normalized hippocampal volumes in the prediction of dementia. Poster presented at Svenska Läkaresällskapets Riksstämma, Nov. 28-30, 2007. (Pdf format).

Eckerström, C. ; Olsson, E. ; Borga, M. et al (2008). Small baseline volume of left hippocampus is associated with subsequent conversion of MCI into dementia. The Göteborg MCI study.  Journal of the Neurological Sciences 272 (1-2) s. 48-59.

Eckerström, C. ; Andreasson, U. ; Olsson, E. et al (2010). Combining hippocampal volume and CSF biomarkers improves predictive value in MCI. Dementia and Geriatric Cognitive Disorders 29, s. 294-300.

Olsson, E., Eckerström, C., Berg, G. et al, Hippocampal volumes in patients exposed to low dose radiation to the basal brain. Submitted.

Back to main (Swedish) ANN page

Back to main (English) ANN page

Last update: April 9, 2010 by Helge Malmgren