iBet uBet web content aggregator. Adding the entire web to your favor.
iBet uBet web content aggregator. Adding the entire web to your favor.



Link to original content: https://pubmed.ncbi.nlm.nih.gov/32961402/
NeuroMark: An automated and adaptive ICA based pipeline to identify reproducible fMRI markers of brain disorders - PubMed Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2020:28:102375.
doi: 10.1016/j.nicl.2020.102375. Epub 2020 Aug 11.

NeuroMark: An automated and adaptive ICA based pipeline to identify reproducible fMRI markers of brain disorders

Yuhui Du  1 , Zening Fu  2 , Jing Sui  3 , Shuang Gao  4 , Ying Xing  5 , Dongdong Lin  2 , Mustafa Salman  6 , Anees Abrol  2 , Md Abdur Rahaman  2 , Jiayu Chen  2 , L Elliot Hong  7 , Peter Kochunov  7 , Elizabeth A Osuch  8 , Vince D Calhoun  6 ; Alzheimer's Disease Neuroimaging Initiative
Affiliations

NeuroMark: An automated and adaptive ICA based pipeline to identify reproducible fMRI markers of brain disorders

Yuhui Du et al. Neuroimage Clin. 2020.

Abstract

Many mental illnesses share overlapping or similar clinical symptoms, confounding the diagnosis. It is important to systematically characterize the degree to which unique and similar changing patterns are reflective of brain disorders. Increasing sharing initiatives on neuroimaging data have provided unprecedented opportunities to study brain disorders. However, it is still an open question on replicating and translating findings across studies. Standardized approaches for capturing reproducible and comparable imaging markers are greatly needed. Here, we propose a pipeline based on the priori-driven independent component analysis, NeuroMark, which is capable of estimating brain functional network measures from functional magnetic resonance imaging (fMRI) data that can be used to link brain network abnormalities among different datasets, studies, and disorders. NeuroMark automatically estimates features adaptable to each individual subject and comparable across datasets/studies/disorders by taking advantage of the reliable brain network templates extracted from 1828 healthy controls as guidance. Four studies including 2442 subjects were conducted spanning six brain disorders (schizophrenia, autism spectrum disorder, mild cognitive impairment, Alzheimer's disease, bipolar disorder, and major depressive disorder) to evaluate validity of the proposed pipeline from different perspectives (replication of brain abnormalities, cross-study comparison, identification of subtle brain changes, and multi-disorder classification using identified biomarkers). Our results highlight that NeuroMark effectively identified replicated brain network abnormalities of schizophrenia across different datasets; revealed interesting neural clues on the overlap and specificity between autism and schizophrenia; demonstrated brain functional impairments present to varying degrees in mild cognitive impairments and Alzheimer's disease; and captured biomarkers that achieved good performance in classifying bipolar disorder and major depressive disorder.

Keywords: Brain disorders; Independent component analysis; NeuroMark; Reproducible and comparable biomarkers; fMRI.

PubMed Disclaimer

Conflict of interest statement

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Figures

None
Graphical abstract
Fig. 1
Fig. 1
Schematic flowchart of the NeuroMark pipeline. Step 1: Calculate group-level independent components (ICs) from two independent datasets, human connectome project (HCP) and genomics superstruct project (GSP) datasets, respectively. Step 2: Match ICs using correlations between their spatial maps and then identify highly replicated intrinsic connectivity networks (ICNs) as the network templates. Step 3: Calculate the individual-level ICNs and their related time courses (TCs) by taking the network templates as prior information in adaptive-ICA. Functional connectivity features such as static and dynamic functional network connectivity (FNC) can be obtained and then compared across datasets, studies, and disorders.
Fig. 2
Fig. 2
The pipeline of classifying BD and MDD patients using brain functional networks (i.e. ICNs) as features, in which an unbiased 10-fold cross-validation procedure was applied.
Fig. 3
Fig. 3
Visualization of the identified network templates, which were divided into seven functional domains based on their anatomical and functional properties. In each subfigure, one color in the composite maps corresponds to an ICN.
Fig. 4
Fig. 4
The spatial correlation matrix between the matched two groups of functional networks. It is seen that the diagonal values are high, indicating the selected network templates are common and reproducible between the GSP and HCP datasets.
Fig. 5
Fig. 5
Correspondence between the HCP and GSP datasets. (A) The number of matched components with the correlation > 0.4, >0.5, >0.6, >0.7, >0.8, and > 0.9, respectively. (B) The number of matched meaningful networks with the correlation > 0.4, >0.5, >0.6, >0.7, >0.8, and > 0.9, respectively, in which the correlation was computed using the whole-brain voxels. (C) The number of matched meaningful networks with the correlation > 0.4, >0.5, >0.6, >0.7, >0.8, and > 0.9, respectively, in which the correlation was computed only using the important voxels with positive Z-scores. Note: corr means correlation.
Fig. 6
Fig. 6
The correspondence and uniqueness of the subject-specific ICNs. Similarity between the network templates and subject-specific ICNs is shown in (A) for all subjects in each dataset (e.g. FBIRN) and (B) for the subjects in each group of dataset (e.g. FBIRN-SZ), using boxplots. In (A) and (B), each sample in the boxplots denotes the template-ICN similarity of one subject. Similarity across the subject-specific ICNs is shown in (C) for all subjects in each dataset and (D) for the subjects in each group of dataset, using bars. In (C) and (D), the inter-subject ICN similarity is shown using a bar. Note: the correlations were computed based on the whole-brain voxels.
Fig. 7
Fig. 7
Results of study 1, which shows that there are reproducible sFNC alterations of SZ between the FBIRN and MPRC datasets. (A) and (D): Mean sFNC matrices across all subjects for the FBIRN and MPRC datasets, respectively. (B) and (E): T-values of all sFNCs obtained from two-sample t-tests for FBIRN and MPRC, respectively. (C) and (F): T-values for the sFNCs passing the multiple comparisons correction (p < 0.05 with Bonferroni correction) for FBIRN and MPRC, respectively. “BFN” denotes Bonferroni correction. (G): Mean sFNC strength across subjects for the HC and SZ groups in the common impairments between FBIRN and MPRC data. For each commonly changed sFNC, the averaged values in SZ patients of FBIRN dataset, SZ patients of MPRC dataset, and HCs of the two datasets are shown, respectively.
Fig. 8
Fig. 8
Results of study 2, which supports that SZ and ASD show common alterations in sFNC. (A): Mean sFNC pattern across all subjects for ABIDEI. (B) and (C): T-values of all sFNCs and T-values of the sFNCs passing the multiple comparisons correction (p < 0.05 with Bonferroni correction), obtained from two-sample t-tests of HC vs. ASD for ABIDEI. “BFN” denotes Bonferroni correction. (D): Mean sFNC strength of each group (ASD, SZ and HC) in the common impairments between SZ and ASD. For each commonly impaired sFNC, the averaged connectivity values in ASD patients of ABIDEI, SZ patients of FBIRN and MPRC, and HCs of the three datasets are shown, respectively. (E) and (F): The significant correlations (r and p values) between FNC measures and clinical symptoms, with p < 0.05. The T-value from testing group difference between HC and disorder by two-sample t-test is also included in each subfigure. Taking (F) for an instance, it shows the correlation between FNC measure (corresponding to IC 69 and IC 21) and ADOS score in ASD patients. The T-value of FNC measure from two-sample t-test between HC and ASD is also shown in the title part.
Fig. 9
Fig. 9
Results of study 3. The results revealed gradually changing patterns from healthy controls (HCs) to early mild cognitive impairment (EMCI) to late MCI (LMCI) to Alzheimer’s disease (AD), measured by dynamic functional network connectivity (dFNC) measures. Upper: Group differences in the fraction rate of occurrences of dFNC states among HC, MCI, and AD. Middle: The discriminating dFNC states, along with the count of subjects that have at least one window clustered into the state. Bottom: Group differences in the fraction rate of occurrences of dFNC states among HC, EMCI, LMCI, and AD. Regarding the fraction rate of occurrences in each state, bar and error bar represent the mean and the standard error of mean, respectively. Significant group differences (false discovery rate corrected, q = 0.05) are indicated by asterisks.
Fig. 10
Fig. 10
Results of study 4. The evaluated measures included individual-class accuracy of bipolar disorder (BD) and major depressive disorder (MDD) (BD_acc and MDD_acc), individual-class precision (BD_prec and MDD_prec), overall accuracy (Overall_acc), balanced accuracy (Bala_acc), and balanced precision (Bala_prec). For each measure, we show the values from 100 classification runs using both boxplot and violinplot, respectively.
Fig. 11
Fig. 11
Spatial maps of six most discriminative ICNs, each of which was selected from one functional domain (i.e., SC, AU, SM, VI, CC, and DM).
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Abrol A., Damaraju E., Miller R.L., Stephen J.M., Claus E.D., Mayer A.R., Calhoun V.D. Replicability of time-varying connectivity patterns in large resting state fMRI samples. Neuroimage. 2017;163:160–176. - PMC - PubMed
    1. Allen, E.A., Damaraju, E., Plis, S.M., Erhardt, E.B., Eichele, T., Calhoun, V.D., 2014. Tracking whole-brain connectivity dynamics in the resting state. Cereb. Cortex 24(3), 663–676. - PMC - PubMed
    1. Andreasen N.C., Paradiso S., O'Leary D.S. Cognitive dysmetria“ as an integrative theory of schizophrenia: a dysfunction in cortical-subcortical-cerebellar circuitry? Schizophr. Bull. 1998;24(2):203–218. - PubMed
    1. Anticevic A., Cole M.W., Repovs G., Murray J.D., Brumbaugh M.S., Winkler A.M., Savic A., Krystal J.H., Pearlson G.D., Glahn D.C. Characterizing thalamo-cortical disturbances in schizophrenia and bipolar illness. Cerebral Cortex. 2014;24(12):3116–3130. - PMC - PubMed
    1. Bailey A., Luthert P., Dean A., Harding B., Janota I., Montgomery M., Rutter M., Lantos P. A clinicopathological study of autism. Brain. 1998;121(Pt 5):889–905. - PubMed

Publication types

MeSH terms