Our abstract “Multimodal Imaging Disease Progression Scores as Quantitative Traits in GWAS of the ADNI Cohort” was selected for an oral presentation at OHBM 2017, Vancouver, Canada!
On top of that lead author Marzia Scelsi received a Merit Abstract Award for the 2017 OHBM Annual Meeting. Congrats Marzia!
M. A. Scelsi, M. Lorenzi, J. M. Schott, S. Ourselin, A. Altmann, “Multimodal Imaging Disease Progression Scores as Quantitative Traits in GWAS of the ADNI Cohort”
Quantitative trait genome-wide association studies (GWAS) in late-onset Alzheimer’s disease (AD) using imaging biomarkers focused either on cross-sectional or on longitudinal phenotypes derived from a single imaging modality. However, since clinical and imaging biomarkers in AD are highly interrelated, association studies based on single biomarkers may miss genetic factors influencing their joint variation. In order to account for their joint variability, in this work we propose to use two well-established AD biomarkers to define a disease progression score (DPS), and to subsequently perform a GWAS using this DPS as a novel quantitative phenotype.
Our work on distributed partial least squares (PLS) has been accepted for an oral presentation at SIPAIM 2016. An earlier version of this paper was presented by Marco Lorenzi at MASAMB (Matematical and Statistical aspects of Molecular Biology) in Cambridge in early October.
Title: Secure multivariate large-scale multi-centric analysis through on-line learning: an imaging genetics case study
Authors: Marco Lorenzi, Boris Gutman, Paul M. Thompson, Daniel C. Alexander, Sebastien Ourselin, Andre Altmann
Abstract: State-of-the-art data analysis methods in genetics and related fields have advanced beyond massively univariate analyses. However, these methods suffer from the limited amount of data available at a single research site. Re- cent large-scale multi-centric imaging-genetic studies, such as ENIGMA, have to rely on meta-analysis of mass univariate models to achieve critical sample sizes for uncovering statistically significant associations. Indeed, model parameters, but not data, can be securely and anonymously shared between partners. We propose here partial least squares (PLS) as a multivariate imaging-genetics model in meta-studies. In particular, we propose an online estimation approach to partial least squares for the sequential estimation of the model parameters in data batches, based on an approximation of the singular value decomposition (SVD) of partitioned covariance matrices. We applied the proposed approach to the challenging problem of modeling the association between 1,167,117 genetic markers (SNPs, single nucleotide polymorphisms) and the brain cortical and sub-cortical atrophy (354,804 anatomical surface features) in a cohort of 639 individuals from the Alzheimer’s Disease Neuroimaging Initiative. We compared two different modeling strategies (sequential- and meta-PLS) to the classic non-distributed PLS. Both strategies exhibited only minimal approximation errors of model parameters. The proposed approaches pave the way to the application of multivariate models in large scale imaging-genetics meta-studies, and may lead to novel understandings of the complex brain phenotype-genotype interactions.
Two of our contributions to the recent AAIC in Toronto have been highlighted in an Alzforum news item. Click here to read the article.
We just added the Re-Annotator annotation for the custom microarray chip from Agilent used by the Allen Institute for Brain Sciences. Get the annotation here. This annotation was also used for our recent paper in Science. We are also taking requests for frequently used chips to be added to the page, just email us! If you use Re-Annotator or the resulting annotations, please cite our paper.
Our paper with the title “Partial Least Squares Modeling for Imaging-Genetics in Alzheimer’s Disease: Plausibility and Generalization” was accepted for presentation at ISBI 2016.
Two of our recent works were highlighted!
Our paper entitled “Sex modifies APOE-related risk of developing Alzheimer’s Disease” was highlighted in “Recognizing the problem of delayed entry in time-to-event studies: Better late than never for clinical neuroscientists” for its proper use of survival analysis.
Further, our paper entitled “Regional Brain Hypometabolism Is Unrelated to Regional Amyloid Plaque Burden” received a scientific comment by Sorg and Grothe within the same issue with the title “The complex link between amyloid and neuronal dysfunction in Alzheimer’s disease“.
Our paper with the title “Validation of non-REM sleep stage decoding from resting state fMRI using linear support vector machines” has been published in NeuroImage. The paper can freely accessed via this link until December 31st, 2015.
A growing body of literature suggests that changes in consciousness are reflected in specific connectivity patterns of the brain as obtained from resting state fMRI (rs-fMRI). As simultaneous electroencephalography (EEG) is often unavailable, decoding of potentially confounding sleep patterns from rs-fMRI itself might be useful and improve data interpretation. Linear support vector machine classifiers were trained on combined rs-fMRI/EEG recordings from 25 subjects to separate wakefulness (S0) from non-rapid eye movement (NREM) sleep stages 1 (S1), 2 (S2), slow wave sleep (SW) and all three sleep stages combined (SX). Classifier performance was quantified by a leave-one-subject-out cross-validation (LOSO-CV) and on an independent validation dataset comprising 19 subjects. Results demonstrated excellent performance with areas under the receiver operating characteristics curve (AUCs) close to 1.0 for the discrimination of sleep from wakefulness (S0|SX), S0|S1, S0|S2 and S0|SW, and good to excellent performance for the classification between sleep stages (S1|S2:~0.9; S1|SW:~1.0; S2|SW:~0.8). Application windows of fMRI data from about 70 s were found as minimum to provide reliable classifications. Discrimination patterns pointed to subcortical–cortical connectivity and within-occipital lobe reorganization of connectivity as strongest carriers of discriminative information. In conclusion, we report that functional connectivity analysis allows valid classification of NREM sleep stages.
Our paper ‘Node-based Gaussian Graphical Model for Identifying Discriminative Brain Regions from Connectivity Graphs’ has won the best paper award at MLMI 2015 (held in conjunction with MICCAI in Munich, Germany).
Our paper entitled “Regional brain hypometabolism is unrelated to regional amyloid plaque burden” is now available online at Brain.
In its original form, the amyloid cascade hypothesis of Alzheimer’s disease holds that fibrillar deposits of amyloid are an early, driving force in pathological events leading ultimately to neuronal death. Early clinicopathological investigations highlighted a number of inconsistencies leading to an updated hypothesis in which amyloid plaques give way to amyloid oligomers as the driving force in pathogenesis. Rather than focusing on the inconsistencies, amyloid imaging studies have tended to highlight the overlap between regions that show early amyloid plaque signal on positron emission tomography and that also happen to be affected early in Alzheimer’s disease. Recent imaging studies investigating the regional dependency between metabolism and amyloid plaque deposition have arrived at conflicting results, with some showing regional associations and other not. We extracted multimodal neuroimaging data from the Alzheimer’s disease neuroimaging database for 227 healthy controls and 434 subjects with mild cognitive impairment. We analysed regional patterns of amyloid deposition, regional glucose metabolism and regional atrophy using florbetapir (18F) positron emission tomography, 18F-fluordeoxyglucose positron emission tomography and T1-weighted magnetic resonance imaging, respectively. Specifically, we derived grey matter density and standardized uptake value ratios for both positron emission tomography tracers in 404 functionally defined regions of interest. We examined the relation between regional glucose metabolism and amyloid plaques using linear models. For each region of interest, correcting for regional grey matter density, age, education and disease status, we tested the association of regional glucose metabolism with (i) cortex-wide florbetapir uptake; (ii) regional (i.e. in the same region of interest) florbetapir uptake; and (iii) regional florbetapir uptake while correcting in addition for cortex-wide florbetapir uptake. P-values for each setting were Bonferroni corrected for 404 tests. Regions showing significant hypometabolism with increasing cortex-wide amyloid burden were classic Alzheimer’s disease-related regions: the medial and lateral parietal cortices. The associations between regional amyloid burden and regional metabolism were more heterogeneous: there were significant hypometabolic effects in posterior cingulate, precuneus, and parietal regions but also significant positive associations in bilateral hippocampus and entorhinal cortex. However, after correcting for global amyloid burden, few of the negative associations remained and the number of positive associations increased. Given the wide-spread distribution of amyloid plaques, if the canonical cascade hypothesis were true, we would expect wide-spread, cortical hypometabolism. Instead, cortical hypometabolism appears to be linked to global amyloid burden. Thus we conclude that regional fibrillar amyloid deposition has little to no association with regional hypometabolism.