tina [about] [news] [software] [demos] [projects] [docs] [images] [faq] [people] [teaching] [tina4] [tina5] [WIKI] [KNOPPIX

project : DODECANTS

jump to



The aim of the DODECANTS (Diagnosis Of DEmentias using Cerebro-spinal fluid Analysis iN Twelve Sectors) project is to produce a diagnostic system for dementing diseases, such as Alzheimer's Disease, Fronto- Temporal Dementia, Vascular Dementia etc. A common feature of all of the diseases being studied in this project is that they produce cerebral atrohpy (a reduction in the volume of the brain). Since the volume of the interior of the skull is fixed in adults, this volume change can be observed as an increase in the volume of cerebro-spinal fluid (CSF). Measuring the CSF volume, rather than the volume of grey or white matter, has several advantages: CSF has a markedly different density to other brain tissues, and so can be segmented more accurately in the MR images used in this work, and the CSF volume is only a small fraction of the overall intra-cranial volume, making any changes more apparent. Working on the hypothesis that different dementing diseases result in different patterns of cerebral atrophy, we measure the CSF volume in twelve coarse regions within the skull, producing a low-parameter representation of its distribution. These measurements are then corrected for head size and normal, age-related atrophy, and a reduced set of five relative variables is computed. Locations of data points in this five-dimensional space can then be investigated for correlation with disease, and diagnoses performed through comparison with a sample of patients with known diagnoses drawn from the same demographic. We have achieved remarkably high diagnostic accuracies with this relatively simple technique.
Although the diagnostic accuracy is impressive, it is known to be sub-optimal, since the twelve CSF measurement regions used are not related to the underlying cortical structure. It is reasonable to assume that aligning the regions to apecific biological features would improve the technique, since some of the diseases being studied are known to affect particualr regions of the brain. Therefore we are now extending the orignial proof-of-concept work by investigating this alignment process using a genetic algotrithm optimisation routine.


Developers Paul A. Bromiley email: paul.bromiley@man.ac.uk
Neil Thacker email: neil.thacker@man.ac.uk


The DODECANTS project is currently funded by the MIAS (from Medical Images and Signals to clinical information) IRC (Interdisciplinary Research Collaboration), under EPSRC grant GR/N14248/01 and the UK Medical Research Council Grant No. D2024/31.


This project uses several of the tools (coregistration, segmentation etc.) available in the TINA machine vision software.


Several Tina Memos describe aspects of this work:

2000-007 Quantification of the severity and distribution of cerebral atrophy provides diagnostic information in dementing diseases.
2002-008 Diagnosis of dementing diseases through the distribution of cerebral atrophy: development of a Multi-Objective Evolutionary Algorithm Optimiser.

Valid HTML 4.01!