Search

Alzheimer’s disease (AD) is the most common cause of dementia worldwide and is characterised pathologically by the accumulation of amyloid beta and tau protein aggregates. Currently, there are no approved disease modifying therapies for clearance of either of these proteins from the brain of people with AD. As well as abnormalities in protein aggregation, other pathological changes are seen in this condition. The function of mitochondria in both the nervous system and rest of the body is altered early in this disease, and both amyloid and tau have detrimental effects on mitochondrial function. In this review article, we describe how the function and structure of mitochondria change in AD. This review summarises current imaging techniques that use surrogate markers of mitochondrial function in both research and clinical practice, but also how mitochondrial functions such as ATP production, calcium homeostasis, mitophagy and reactive oxygen species production are affected in AD mitochondria. The evidence reviewed suggests that the measurement of mitochondrial function may be developed into a future biomarker for early AD. Further work with larger cohorts of patients is needed before mitochondrial functional biomarkers are ready for clinical use.

Graph neural network (GNN) models are increasingly being used for the classification of electroencephalography (EEG) data. However, GNN-based diagnosis of neurological disorders, such as Alzheimer’s disease (AD), remains a relatively unexplored area of research. Previous studies have relied on functional connectivity methods to infer brain graph structures and used simple GNN architectures for the diagnosis of AD. In this work, we propose a novel adaptive gated graph convolutional network (AGGCN) that can provide explainable predictions. AGGCN adaptively learns graph structures by combining convolution-based node feature enhancement with a correlation-based measure of power spectral density similarity. Furthermore, the gated graph convolution can dynamically weigh the contribution of various spatial scales. The proposed model achieves high accuracy in both eyes-closed and eyes-open conditions, indicating the stability of learned representations. Finally, we demonstrate that the proposed AGGCN model generates consistent explanations of its predictions that might be relevant for further study of AD-related alterations of brain networks.

Alzheimer’s disease (AD) is the leading form of dementia worldwide. AD disrupts neuronal pathways and thus is commonly viewed as a network disorder. Many studies demonstrate the power of functional connectivity (FC) graph-based biomarkers for automated diagnosis of AD using electroencephalography (EEG). However, various FC measures are commonly utilised, as each aims to quantify a unique aspect of brain coupling. Graph neural networks (GNN) provide a powerful framework for learning on graphs. While a growing number of studies use GNN to classify EEG brain graphs, it is unclear which method should be utilised to estimate the brain graph. We use eight FC measures to estimate FC brain graphs from sensor-level EEG signals. GNN models are trained in order to compare the performance of the selected FC measures. Additionally, three baseline models based on literature are trained for comparison. We show that GNN models perform significantly better than the other baseline models. Moreover, using FC measures to estimate brain graphs improves the performance of GNN compared to models trained using a fixed graph based on the spatial distance between the EEG sensors. However, no FC measure performs consistently better than the other measures. The best GNN reaches 0.984 area under sensitivity-specificity curve (AUC) and 92% accuracy, whereas the best baseline model, a convolutional neural network, has 0.924 AUC and 84.7% accuracy.

Background: Category Fluency Test (CFT) is a common measure of semantic memory (SM). Test performance, however, is also influenced by other cognitive functions. We here propose a scoring procedure that quantifies the correlation between the serial recall order (SRO) of words retrieved during the CFT and a number of linguistic features, to obtain purer SM measures. To put this methodology to the test, we addressed a proof-of-concept hypothesis whereby, in alignment with the literature, older adults would show better SM.Methods: Ninety participants (45 aged 18–21 years; 45 aged 70–81 years) with normal neurological and cognitive functioning completed a 1-min CFT. SRO was scored as an ordinal variable incrementing by one unit for each valid entry. Each word was also scored for 16 additional linguistic features. Participant-specific normalised correlation coefficients were calculated between SRO and each feature and were analysed with group comparisons and graph theory.Results: Younger adults showed more negative correlations between SRO and “valence” (a feature of words pleasantness). This was driven by the first five words generated. When analysed with graph theory, SRO had significantly higher degree and lower betweenness centrality among older adults.Conclusion: In older adults, SM relies significantly less on pleasantness of entries typically retrieved without semantic control. Moreover, graph-theory metrics indicated better optimised links between SRO and linguistic features in this group. These findings are aligned with the principle whereby SM processes tend to solidify with ageing. Although additional work is needed in support of an SRO-based item-level scoring procedure of CFT performance, these initial findings suggest that this methodology could be of help in characterising SM in a purer form.

Background: The incidence of Alzheimer disease (AD) is increasing with the ageing population. The development of low cost non-invasive diagnostic aids for AD is a research priority. This pilot study investigated whether an approach based on a novel dynamic quantitative parametric EEG method could detect abnormalities in people with AD. Methods: 20 patients with probable AD, 20 matched healthy controls (HC) and 4 patients with probable fronto temporal dementia (FTD) were included. All had detailed neuropsychology along with structural, resting state fMRI and EEG. EEG data were analyzed using the Error Reduction Ratio-causality (ERR-causality) test that can capture both linear and nonlinear interactions between different EEG recording areas. The 95% confidence intervals of EEG levels of bi-centroparietal synchronization were estimated for eyes open (EO) and eyes closed (EC) states. Results: In the EC state, AD patients and HC had very similar levels of bi-centro parietal synchronization; but in the EO resting state, patients with AD had significantly higher levels of synchronization (AD = 0.44; interquartile range (IQR) 0.41 vs. HC = 0.15; IQR 0.17, p < 0.0001). The EO/EC synchronization ratio, a measure of the dynamic changes between the two states, also showed significant differences between these two groups (AD ratio 0.78 versus HC ratio 0.37 p < 0.0001). EO synchronization was also significantly different between AD and FTD (FTD = 0.075; IQR 0.03, p < 0.0001). However, the EO/EC ratio was not informative in the FTD group due to very low levels of synchronization in both states (EO and EC). Conclusion: In this pilot work, resting state quantitative EEG shows significant differences between healthy controls and patients with AD. This approach has the potential to develop into a useful non-invasive and economical diagnostic aid in AD.