No abstract

There is substantial experimental evidence for dysregulation of several microRNA (miRNA) expression levels in Alzheimer’s disease (AD). MiRNAs modulate critical brain intracellular signaling pathways and are associated with AD core pathophysiological mechanisms. First, we conducted a real-time quantitative PCR-based pilot study to identify a set of brain-enriched miRNAs in a monocentric cohort of cognitively normal individuals with subjective memory complaints, a condition associated with increased risk of AD. Second, we investigated the impact of age, sex, and the Apolipoprotein E ε4 (APOE ε4) allele, on the identified miRNA plasma concentrations. In addition, we explored the cross-sectional and longitudinal association of the miRNAs plasma concentrations with regional brain metabolic uptake using amyloid-β (Aβ)-positron emission tomography (Aβ-PET) and 18F-fluorodeoxyglucose-PET (18F-FDG-PET). We identified a set of six brain-enriched miRNAs—miRNA-125b, miRNA-146a, miRNA-15b, miRNA-148a, miRNA-26b, and miRNA-100. Age, sex, and APOE ε4 allele were not associated with individual miRNA abundance. MiRNA-15b concentrations were significantly lower in the Aβ-PET-positive compared to Aβ-PET-negative individuals. Furthermore, we found a positive effect of the miRNA-15b*time interaction on regional metabolic 18F-FDG-PET uptake in the left hippocampus. Plasma miRNA-125b concentrations, as well as the miRNA-125b*time interaction (over a 2-year follow-up), were negatively associated with regional Aβ-PET standard uptake value ratio in the right anterior cingulate cortex. At baseline, we found a significantly negative association between plasma miRNA-125b concentrations and 18F-FDG-PET uptake in specific brain regions. In an asymptomatic at-risk population for AD, we show significant associations between plasma concentrations of miRNA-125b and miRNA-15b with core neuroimaging biomarkers of AD pathophysiology. Our results, coupled with existing experimental evidence, suggest a potential protective anti-Aβ effect of miRNA-15b and a biological link between miRNA-125b and Aβ-independent neurotoxic pathways.

No abstract

No abstract

Deep learning methods have achieved impressive results for 3D medical image segmentation. However, when the network is only guided by voxel-level information, it may provide anatomically aberrant segmentations. When performing manual segmentations, experts heavily rely on prior anatomical knowledge. Topology is an important prior information due to its stability across patients. Recently, several losses based on persistent homology were proposed to constrain topology. Persistent homology offers a principled way to control topology. However, it is computationally expensive and complex to implement, in particular in 3D. In this paper, we propose a novel loss function to introduce topological priors in deep learning-based segmentation, which is fast to compute and easy to implement. The loss performs a projected pooling within two steps. We first focus on errors from a global perspective by using 3D MaxPooling to obtain projections of 3D data onto three planes: axial, coronal and sagittal. Then, 2D MaxPooling layers with different kernel sizes are used to extract topological features from the multi-view projections. These two steps are combined using only MaxPooling, thus ensuring the efficiency of the loss function. Our approach was evaluated in several medical image datasets (spleen, heart, hippocampus, red nucleus). It allowed reducing topological errors and, in some cases, improving voxel-level accuracy.