OPEN Research Support

Thomas Krøigård
Department of neurology

Projekt styring
Projekt status    Open
Data indsamlingsdatoer
Start 01.05.2021  
Slut 31.12.2022  

Agreement between artificial intelligence quantitative EEG analysis and molecular imaging in dementia with Lewy bodies

Short summary

Formålet med forsøget er at finde ud af, om man kan bruge en optagelse af hjernens elektriske aktivitet, et såkaldt EEG, til at påvise forandringer i hjernen der tyder på at en person har udviklet sygdommen demens. Specielt er der et behov for at kunne skelne mellem forskellige undertyper af demens. Den hyppigste undertype er Alzheimers demens, men der findes andre undertyper, som demens med såkaldte Lewy bodies, der har andre symptomer og et andet forløb. Resultatet af EEG-optagelsen vil


Approximately 20% of dementia patients have dementia with Lewy bodies (DLB). Visual hallucinations, visuospatial deficits, and fluctuations in the level of consciousness often dominate this type of dementia. In addition, patients develop Parkinsonism. However, during early disease phases, differentiation from other types of dementia, such as Alzheimer's disease may be difficult. The condition is relatively frequent among elderly due to the high prevalence of dementia in general. Two clinical types with a common pathogenesis of accumulation of alpha-synuclein and loss of dopaminergic neurons have been described: In dementia with Lewy bodies, dementia is the initial presentation, later followed by Parkinsonism. In Parkinson's disease with dementia, dementia is presided by at least one year of idiopathic Parkinson's disease without dementia. As mentioned, differential diagnosis between DLB and Alzheimer's disease may be difficult during early disease phases, as episodic memory deficiency may be an initial symptom in both conditions. Although the underlying mechanism responsible for the memory deficiency is probably different, the clinical presentation may be similar. Differentiation between DLB and Alzheimer's disease has significant clinical implications regarding management and planning of follow up as the patients with the former diagnosis will require specialized evaluation for development of Parkinsonism whereas patients with the latter are best managed in the primary health care setting. The diagnosis of DLB is based on the clinical picture consisting of fluctuations in cognitive function, visual hallucinations and Parkinsonism, but in case of diagnostic uncertainty, the concentration of dopamine transporter in the basal ganglia may be assessed through molecular imaging techniques (DAT scan). The concentration is reduced in DLB, and normal in Alzheimer's disease. The test has a high sensitivity (83-92%) and specificity (67-93%) for differentiation between the two conditions (Davison and O'Brien, 2014). However, DAT scans are relatively time consuming and dependent on injection of a radioactive ligand. Capacity is limited and in general, there are waiting lists for the test. Other molecular imaging techniques, such as FDG-PET are used to support the diagnosis of neurodegenerative diseases in general, including Alzheimer's disease. For some time it has been known that quantitative changes are present in the electroencephalogram (EEG) of patients with dementia. However, EEG analysis based on routine, visual assessment does not contribute to dementia diagnosis, as the changes are subtle. Advanced data analysis based on machine learning and artificial intelligence (AI) allows for identification of these subtle changes. Differentiation between DLB and Alzheimer's disease is further plausible because occipital cortex is primarily affected in DLB as shown in cerebral perfusion studies. Therefore, analysis could provide differentiation between these dementia subtypes in addition to dementia diagnosis. Previous quantitative EEG studies not based on AI have shown significant differences at group levels (Babiloni et al., 2017, 2018; Bonanni et al., 2016). No direct comparison of the diagnostic values of AI based EEG analysis and DAT scan has been performed.

Description of the cohort

A consecutive series of patients referred for molecular imaging due to clinical suspicion of dementia by the dementia outpatient clinic at Odense University Hospital will be invited to participate in the study.

Data and biological material

EEG, molecular imaging, clinical scores.

Collaborating researchers and departments

Department of nuclear medicine

  • Peter Grupe