Projects / Programmes
Metabolic brain characteristics in neurodegenerative dementia syndromes and its correlation with histopathological brain changes
| Code |
Science |
Field |
Subfield |
| 3.03.00 |
Medical sciences |
Neurobiology |
|
| 1.02.00 |
Natural sciences and mathematics |
Physics |
|
| Code |
Science |
Field |
| 3.01 |
Medical and Health Sciences |
Basic medicine |
| 1.03 |
Natural Sciences |
Physical sciences |
Neurodegenerative diseases, biomarkers, brain imaging, network analysis, deep brain learning, histopathology, 2-[18F]FDG-PET, metabolic patterns, Alzheimer's disease, dementia with Levy bodies, dementia in Parkinson's disease, frontotemporal dementia, Creutzfeldt Jacobs disease
Data for the last 5 years (citations for the last 10 years) on
June 9, 2023;
A3 for period
2017-2021
Data for ARRS tenders (
04.04.2019 – Programme tender,
archive
)
| Database |
Linked records |
Citations |
Pure citations |
Average pure citations |
| WoS |
819 |
18,954 |
18,054 |
22.04 |
| Scopus |
698 |
21,966 |
20,966 |
30.04 |
Researchers (27)
Organisations (3)
Abstract
Dementia poses a major health, social and economic burden worldwide. In the aging Europe society, 10.5 million people are suffering from dementia. The major unmet need in dementia care is a lack of causative treatment. One of the reasons for that are insufficient biomarkers for various dementia syndromes that would enable early and correct diagnosis and consequently enrolment of patients with specific dementia syndromes into clinical trials. In this project we will identify and validate novel metabolic brain biomarkers of various dementia syndromes. We will for the first time validate them with the results of histopathological brain analysis. Finally, we will develop two novel automated classification algorithms based on »machine learning« methods for the analysis of metabolic brain images to improve diagnostic accuracy of dementia syndromes. The most common neurodegenerative dementias are dementia due to Alzheimer's disease (AD), dementia with Lewy bodies (DLB), Parkinson’s disease dementia (PDD) and frontotemporal dementia (FTD). AD, DLB, PDD and FTD are slowly progressing diseases as opposed to the rapidly progressive Creutzfeldt-Jacob disease (CJD).There is a significant overlap in early clinical presentation in these syndromes, with misdiagnosis rate at around 30%. While few biomarkers are already available for AD, there is a lack of biomarkers for non-AD dementias. There is an urgent need for further improvement of AD biomarkers and for the development of biomarkers for non-AD dementias. We aim to first, identify and validate specific disease-related patternsfor AD, DLB, PDD, FTD and CJD in Slovenian cohort of patients with these neurodegenerative dementias with the network analysis of their 2-[18F]FDG-PET brainimages. Second, we plan for the first time to perform a histopathological validation of AD and CJD metabolic patterns by correlating the topography and expression of AD and CJD metabolic patterns with the results of histopathological brain examination performed at autopsy. Third, we will develop two different automated types of 2-[18F]FDG-PET images analyses for the differential diagnosis of neurodegenerative dementias and compare their performance metrics. 2-[18F]FDG-PET scans of 20 patients with each of AD, DLB, PDD, FTD and CJD as well as 20 normal controls (NC) will be analysed with an automatic voxel-based scaled subprofile model analysis based on principal component analysis with the ScAnVP software to identify the characteristic patterns. These newly identified and dementia specific patterns will be then validated on an independent cohort of various dementia patients and NC, as well as by the correlation of metabolic pattern characteristics of AD and CJD with the specific histopathological brain findings. Finally, we will develop, for the first time, a multinomial logistic regression classification algorithmthat will classify patients’ 2-[18F]FDG-PET images to six classes, i.e. AD, DLB, PDD, FTD, CJD or indeterminate dementia. We will also develop a deep neural network modelconsisting of multi-layer convolutional neural network and a classifier that will classify brain images to diagnostic classes. The statistical metrics of the two algorithms will be assessed and compared, using the patients’ clinical diagnosis (or pathological diagnosis in AD and CJD) on the follow-up at least two years after scanning as gold standard. We believe, based on our preliminary results, that results of this research will significantly improve the accuracy of the diagnosis of dementia syndromesand enhance the utility of metabolic brain imaging in dementia research and in patients' diagnostic workup. To reach our goals, we have assembled an interdisciplinary research teamof neurologists, nuclear medicine specialists, pathologists and medical physicists from University Medical Centre Ljubljana and University of Ljubljana. This experienced team assures unique multidisciplinary platform for the proposed project.
