Cortical graph neural network for AD and MCI diagnosis and transfer learning across populations.
ABSTRACT: Combining machine learning with neuroimaging data has a great potential for early diagnosis of mild cognitive impairment (MCI) and Alzheimer's disease (AD). However, it remains unclear how well the classifiers built on one population can predict MCI/AD diagnosis of other populations. This study aimed to employ a spectral graph convolutional neural network (graph-CNN), that incorporated cortical thickness and geometry, to identify MCI and AD based on 3089 T1-weighted MRI data of the ADNI-2 cohort, and to evaluate its feasibility to predict AD in the ADNI-1 cohort (n = 3602) and an Asian cohort (n = 347). For the ADNI-2 cohort, the graph-CNN showed classification accuracy of controls (CN) vs. AD at 85.8% and early MCI (EMCI) vs. AD at 79.2%, followed by CN vs. late MCI (LMCI) (69.3%), LMCI vs. AD (65.2%), EMCI vs. LMCI (60.9%), and CN vs. EMCI (51.8%). We demonstrated the robustness of the graph-CNN among the existing deep learning approaches, such as Euclidean-domain-based multilayer network and 1D CNN on cortical thickness, and 2D and 3D CNNs on T1-weighted MR images of the ADNI-2 cohort. The graph-CNN also achieved the prediction on the conversion of EMCI to AD at 75% and that of LMCI to AD at 92%. The find-tuned graph-CNN further provided a promising CN vs. AD classification accuracy of 89.4% on the ADNI-1 cohort and >90% on the Asian cohort. Our study demonstrated the feasibility to transfer AD/MCI classifiers learned from one population to the other. Notably, incorporating cortical geometry in CNN has the potential to improve classification performance.
Project description:Mild cognitive impairment (MCI) and Alzheimer's Disease (AD) are complex diseases with their molecular architecture not elucidated. APOE, Amyloid Beta Precursor Protein (APP), and Presenilin-1 (PSEN1) are well-known genes associated with both MCI and AD. Recently, epigenetic alterations and dysregulated regulatory elements, such as microRNAs (miRNAs), have been reported associated with neurodegeneration. In this study, differential expression analysis (DEA) was performed for genes and miRNAs based on microarray and RNA-Seq data. Global gene profile of healthy individuals, early and late mild cognitive impairment (EMCI and LMCI, respectively), and AD was obtained from ADNI Cohort. miRNA global profile of healthy individuals and AD patients was extracted from public RNA-Seq data. DEA performed with limma package on ADNI Cohort data highlighted eight differential expressed (DE) genes (AGER, LINC00483, MMP19, CATSPER1, ARFGAP1, GPER1, PHLPP2, TRPM2) (false discovery rate (FDR) p-value < 0.05) between EMCI and LMCI patients. Previous molecular studies showed associations between these genes with dementia and neurological-related pathways. Five dysregulated miRNAs were identified by DEA performed with RNA-Seq data and edgeR (FDR p-value < 0.002). All reported miRNAs in AD interact with the aforementioned genes. Our integrative transcriptomic analysis was able to identify a set of miRNA-gene interactions that may be involved in cognitive and neurodegeneration processes.
Project description:BACKGROUND:Alzheimer's disease (AD) is one of the leading causes of death in the US and there is no validated drugs to stop, slow or prevent AD. Despite tremendous effort on biomarker discovery, existing findings are mostly individual biomarkers and provide limited insights into the transcriptomic decoupling underlying AD. We propose to explore the gene co-expression patterns in multiple AD stages, including cognitively normal (CN), early mild cognitive impairment (EMCI), late MCI and AD. METHODS:We modified traiditonal joint graphical lasso to model our asusmption that the co-expression networks in consecutive disease stages are largely similar with critical differences. In addition, we performed subsequent network comparison analysis for identification of stage specific transcriptomic decoupling. We focused our analysis on top AD-enriched pathways. RESULTS:We observed that 419 edges in CN, 420 edges in EMCI, 381 edges in LMCI and 250 edges in AD were frequently estimated with non zero weights. With modified JGL, the weight of all estimated edges in CN, EMCI and LMCI are zero. In AD group, 299 edges were occasionally estimated to be nonzero and the average correlation between genes was 0.0023. For co-expression change during AD progression, there are 66 pairs of genes that demonstrated a continuously decreasing or increasing co-expression from CN to EMCI, LMCI and AD.The network level clustering coefficient remains stable from CN to LMCI and then decreases significantly when progressing to AD. When evaluating edge level differences, we identified eight gene modules with continuously decreasing or increasing co-expression patterns during AD progression. Five of them shows significant changes from CN to EMCI and thus have the potential to serve system biomarkers for early screening of AD. CONCLUSION:We employed a modified joint graphical lasso for estimation of co-expression networks for multiple stages of AD. Comparing with graphical lasso, our modified joint graphical lasso model accounts for the similarity in consecutive disease stages. Our results on real data set revealed five gene clusters with obvious co-expression pattern change from CN to EMCI, which could be used as potential system-level biomarkers for early screening of AD.
Project description:Using data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) population, we examined (1) cross-sectional relationships between amyloid deposition, hypometabolism, and cognition, and (2) associations between amyloid and hypometabolism measurements and longitudinal cognitive measurements.We examined associations between mean cortical florbetapir uptake, mean (18) F-fluorodeoxyglucose-positron emission tomography (FDG-PET) within a set of predefined regions, and Alzhiemer's Disease Assessment Scale (ADAS-cog) performance in 426 ADNI participants (126 normal, 162 early mild cognitive impairment [EMCI], 85 late MCI [LMCI], 53 Alzheimer disease [AD] patients). For a subset of these (76 normal, 81 LMCI) we determined whether florbetapir and FDG-PET were associated with retrospective decline in longitudinal ADAS-cog measurements.Twenty-nine percent of normal subjects, 43% of EMCI patients, 62% of LMCI patients, and 77% of AD patients were categorized as florbetapir positive. Florbetapir was negatively associated with concurrent FDG and ADAS-cog in both MCI groups. In longitudinal analyses, florbetapir-positive subjects in both normal and LMCI groups had greater ongoing ADAS-cog decline than those who were florbetapir negative. However, in normal subjects, florbetapir positivity was associated with greater ADAS-cog decline than FDG, whereas in LMCI, FDG positivity was associated with greater decline than florbetapir.Although both hypometabolism and ?-amyloid (A?) deposition are detectable in normal subjects and all diagnostic groups, A? showed greater associations with cognitive decline in normal participants. In view of the minimal cognitive deterioration overall in this group, this suggests that amyloid deposition has an early and subclinical impact on cognition that precedes metabolic changes. At moderate and later stages of disease (LMCI/AD), hypometabolism becomes more pronounced and more closely linked to ongoing cognitive decline.
Project description:Olfactory impairment is associated with prodromal Alzheimer's disease (AD) and is a risk factor for the development of dementia. AD pathology is known to disrupt brain regions instrumental in olfactory information processing, such as the primary olfactory cortex (POC), the hippocampus, and other temporal lobe structures. This selective vulnerability suggests that the functional connectivity (FC) between the olfactory network (ON), consisting of the POC, insula and orbital frontal cortex (OFC) (Tobia et al., 2016), and the hippocampus may be impaired in early stage AD. Yet, the development trajectory of this potential FC impairment remains unclear. Here, we used resting-state functional magnetic resonance imaging (rs-fMRI) data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) to investigate FC changes between the ON and hippocampus in four groups: aged-matched cognitively normal (CN), early mild cognitive impairment (EMCI), late mild cognitive impairment (LMCI), and AD. FC was calculated using low frequency fMRI signal fluctuations in the ON and hippocampus (Tobia et al., 2016). We found that the FC between the ON and the right hippocampus became progressively disrupted across disease states, with significant differences between EMCI and LMCI groups. Additionally, there were no significant differences in gray matter hippocampal volumes between EMCI and LMCI groups. Lastly, the FC between the ON and hippocampus was significantly correlated with neuropsychological test scores, suggesting that it is related to cognition in a meaningful way. These findings provide the first in vivo evidence for the involvement of FC between the ON and hippocampus in AD pathology. Results suggest that functional connectivity (FC) between the olfactory network (ON) and hippocampus may be a sensitive marker for Alzheimer's disease (AD) progression, preceding gray matter volume loss.
Project description:Amyloid beta (A?) plaques aggregation is considered as the "start" of the degenerative process that manifests years before the clinical symptoms appear in Alzheimer's Disease (AD). The aim of this study is to use back propagation neural networks (BPNNs) in 18F-florbetapir PET data for automated classification of four patient groups including AD, late mild cognitive impairment (LMCI), early mild cognitive impairment (EMCI), and significant memory concern (SMC), versus normal control (NC) for early AD detection. Five hundred images for AD, LMCI, EMCI, SMC, and NC, i.e., 100 images for each group, were used from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. The results showed that the automated classification of NC/AD produced a high accuracy of 87.9%, while the results for the prodromal stages of the disease were 66.4%, 60.0%, and 52.9% for NC/LCMI, NC/EMCI and NC/SMC, respectively. The proposed method together with the image preparation steps can be used for early AD detection and classification with high accuracy using A? PET dataset.
Project description:We examined differences in cerebral blood flow (CBF) measured by Arterial Spin Labeled perfusion MRI (ASL MRI) across the continuum from cognitively normal (CN) older adults to mild Alzheimer's Disease (AD) using data from the multi-site Alzheimer's Disease Neuroimaging Initiative (ADNI). Measures of CBF, in a predetermined set of regions (meta-ROI), and hippocampal volume were compared between CN (n = 47), patients with early and late Mild Cognitive Impairment [EMCI (n = 32), LMCI (n = 35)], and AD (n = 15). Associations between these measures and disease severity, assessed by Clinical Dementia Rating scale sum of boxes (CDR SB), were also assessed. Mean meta-ROI CBF was associated with group status and significant hypoperfusion was observed in LMCI and AD relative to CN. Hippocampal volume was associated with group status, but only AD patients had significantly smaller volumes than the CN. When examining the relationship between these measures and disease severity, both were significantly associated with CDR SB and appeared to provide independent prediction of status. In light of the tight link between CBF and metabolism, ASL MRI represents a promising functional biomarker for early diagnosis and disease tracking in AD and this study is the first to demonstrate the feasibility in a multi-site context in this population. Combining functional and structural measures, which can be acquired in the same scanning session, appears to provide additional information about disease severity relative to either measure alone.
Project description:The Clinical Core of the Alzheimer's Disease Neuroimaging Initiative (ADNI) has provided clinical, operational, and data management support to ADNI since its inception. This article reviews the activities and accomplishments of the core in support of ADNI aims. These include the enrollment and follow-up of more than 800 subjects in the three original cohorts: healthy controls, amnestic mild cognitive impairment (now referred to as late MCI, or LMCI), and mild Alzheimer's disease (AD) in the first phase of ADNI (ADNI 1), with baseline longitudinal, clinical, and cognitive assessments. These data, when combined with genetic, neuroimaging, and cerebrospinal fluid measures, have provided important insights into the neurobiology of the AD spectrum. Furthermore, these data have facilitated the development of novel clinical trial designs. ADNI has recently been extended with funding from an NIH Grand Opportunities (GO) award, and the new ADNI GO phase has been launched; this includes the enrollment of a new cohort, called early MCI, with milder episodic memory impairment than the LMCI group. An application for a further 5 years of ADNI funding (ADNI 2) was recently submitted. This funding would support ongoing follow-up of the original ADNI 1 and ADNI GO cohorts, as well as additional recruitment into all categories. The resulting data would provide valuable data on the earliest stages of AD, and support the development of interventions in these critically important populations.
Project description:Background This study aimed to investigate feasible gray matter microstructural biomarkers with high sensitivity for early Alzheimer’s disease (AD) detection. We propose a diffusion tensor imaging (DTI) measure, “radiality”, as an early AD biomarker. It is the dot product of the normal vector of the cortical surface and primary diffusion direction, which reflects the fiber orientation within the cortical column. Methods We analyzed neuroimages from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database, including images from 78 cognitively normal (CN), 50 early mild cognitive impairment (EMCI), 34 late mild cognitive impairment (LMCI), and 39?AD patients. We then evaluated the cortical thickness (CTh), mean diffusivity (MD), which are conventional AD magnetic resonance imaging (MRI) biomarkers, and the amount of accumulated amyloid and tau using positron emission tomography (PET). Radiality was projected on the gray matter surface to compare and validate the changes with different stages alongside other neuroimage biomarkers. Results The results revealed decreased radiality primarily in the entorhinal, insula, frontal, and temporal cortex with further progression of disease. In particular, radiality could delineate the difference between the CN and EMCI groups, while the other biomarkers could not. We examined the relationship between radiality and other biomarkers to validate its pathological evidence in AD. Overall, radiality showed a high association with conventional biomarkers. Additional ROI analysis revealed the dynamics of AD-related changes as stages onward. Conclusion Radiality in cortical gray matter showed AD-specific changes and relevance with other conventional AD biomarkers with high sensitivity. Moreover, radiality could identify the group differences seen in EMCI, representative of changes in early AD, which supports its superiority in early diagnosis compared to that possible with conventional biomarkers. We provide evidence of structural changes with cognitive impairment and suggest radiality as a sensitive biomarker for identifying early AD.
Project description:BackgroundAlzheimer’s disease (AD) is a chronic progressive neurodegenerative disease impacting an estimated 44 million adults worldwide. The causal pathology of AD (accumulation of amyloid-beta and tau), precedes hallmark symptoms of dementia by more than a decade, necessitating development of early diagnostic markers of disease onset, particularly for new drugs that aim to modify disease processes. To evaluate differentially methylated positions (DMPs) as novel blood-based biomarkers of AD, we used a subset of 653 individuals with peripheral blood (PB) samples in the Alzheimer’s disease Neuroimaging Initiative (ADNI) consortium. The selected cohort of AD, mild cognitive impairment (MCI), and age-matched healthy controls (CN) all had imaging, genetics, transcriptomics, cerebrospinal protein markers, and comprehensive clinical records, providing a rich resource of concurrent multi-omics and phenotypic information on a well-phenotyped subset of ADNI participants.ResultsIn this manuscript, we report cross-diagnosis differential peripheral DNA methylation in a cohort of AD, MCI, and age-matched CN individuals with longitudinal DNA methylation measurements. Epigenome-wide association studies (EWAS) were performed using a mixed model with repeated measures over time with a P value cutoff of 1 × 10?5 to test contrasts of pairwise differential peripheral methylation in AD vs CN, AD vs MCI, and MCI vs CN. The most highly significant differentially methylated loci also tracked with Mini Mental State Examination (MMSE) scores. Differentially methylated loci were enriched near brain and neurodegeneration-related genes (e.g., BDNF, BIN1, APOC1) validated using the genotype tissue expression project portal (GTex).ConclusionsOur work shows that peripheral differential methylation between age-matched subjects with AD relative to healthy controls will provide opportunities to further investigate and validate differential methylation as a surrogate of disease. Given the inaccessibility of brain tissue, the PB-associated methylation marks may help identify the stage of disease and progression phenotype, information that would be central to bringing forward successful drugs for AD.
Project description:Based on previous studies, a preclinical classification for Alzheimer's disease (AD) has been proposed. However, 1) specificity of the different neuronal injury (NI) biomarkers has not been studied, 2) subjects with subtle cognitive impairment but normal NI biomarkers (SCINIB) have not been included in the analyses and 3) progression to mild cognitive impairment (MCI) or dementia of the AD type (DAT), referred to here as MCI/DAT, varies between studies. Therefore, we analyzed data from 486 cognitively normal (CN) and 327 DAT subjects in the AD Neuroimaging Initiative (ADNI)-1/GO/2 cohorts.In the ADNI-1 cohort (median follow-up of 6 years), 6.3% and 17.0% of the CN subjects developed MCI/DAT after 3 and 5 years follow-up, respectively. NI biomarker cutoffs [structural magnetic resonance imaging (MRI), fluorodeoxyglucose positron emission tomography (FDG-PET) and cerebrospinal fluid (CSF) tau] were established in DAT patients and memory composite scores were calculated in CN subjects in a cross-sectional sample (n?=?160). In the complete longitudinally followed CN ADNI cohort (n?=?326, median follow-up of 2 years), CSF and MRI values predicted an increased conversion to MCI/DAT. Different NI biomarkers showed important disagreements for classifying subjects as abnormal NI [kappa?=?(-0.05)-(0.33)] and into AD preclinical groups. SCINIB subjects (5.0%) were more prevalent than AD preclinical stage 3 subjects (3.4%) and showed a trend for increased progression to MCI/DAT.Different NI biomarkers lead to different classifications of ADNI subjects, while structural MRI and CSF tau measures showed the strongest predictive value for progression to MCI/DAT. The newly defined SCINIB category of ADNI subjects is more prevalent than AD preclinical stage individuals.