Privacy-preserving quality control of neuroimaging datasets in federated environment

Saha, Debbrata K.; Calhoun, Vince D.; Du, Yuhui; Fu, Zhibin; Panta, Sandeep R.; Kwon, Soo Min; Sarwate, Anand D.; Plis, Sergey M.

doi:10.1101/826974

Cited by 6 publications

(5 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…tSNE using EMD score t-distributed stochastic neighbor embedding (tSNE) leads to a powerful and flexible visualization of high-dimensional data by giving each datapoint a location in a two or threedimensional map [59,60]. Decentralized stochastic neighbor embedding (dSNE) is used to separate the subgroups in the data using their distance metrics [61,62]. tSNE considers each shape a data point in the subspace and uses EMD distances to select the neighbors for the embeddings.…”

Section: Kernel Density Estimator (Kde)mentioning

confidence: 99%

“…Thus, studies employ subgrouping/clustering of the subjects to minimize the dissimilarity and making a more rational comparison (Luchins, Weinberger et al 1979; Scarr, Cowie et al 2009; Rahaman, Damaraju et al 2020). A typical sliding window plus clustering (SWC) analysis approach resolves the problems by continuously modeling the system through a fixed set of connectivity patterns or states (Allen, Damaraju et al 2014; Damaraju, Allen et al 2014; Miller, Yaesoubi et al 2016; Vergara, Mayer et al 2018; Saha, Abrol et al 2019). Moreover, a recent study of tri-clustering dFNC data suggests three-dimensional connectivity states and reveals more significant group differences than previous studies (Rahaman, Damaraju et al 2020).…”

Section: Definitions and Backgroundmentioning

confidence: 99%

See 1 more Smart Citation

Statelets: High dimensional predominant shapes in dynamic functional network connectivity

Rahaman

Damaraju

Saha

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Dynamic functional network connectivity (dFNC) analysis has been attracting interest over the past years by elucidating crucial details of brain activation patterns in severe neurological or psychiatric disorders. Consequently, a considerable amount of work has been conducted to analyze dFNC and leverage it for an improved understanding of neuro-dynamics and cognition. However, state-of-art methods mostly focus on fixed patterns of connectivity that reoccur. Such approaches do not capture the more transient feature space of the brain's neural system and its dynamic properties. Likewise, the dynamical system approaches to model the system continuously and do not well capture unmodelled heterogeneity across subjects and time. Here, we seek an improved understanding of these connectivity states and the mechanism through which their dynamics vary across individuals. Relying on the concept of shapelets to model the temporal dynamics of functional associations between intrinsic brain networks, we develop the 'statelet' - a high dimensional state-shape representation of dynamics. We handle scale differences using the earth mover distance as our similarity metric and utilizing kernel density estimation to build a probability density profile for each shapelet. Unlike clustering the time series, we approximate short, variable-length connectivity patterns using the probability density profile of the shapelets from lower dimensions. We apply the proposed approach to study patients with schizophrenia (SZ) exhibit and identify reduced modularity increased statelet recurrence in their brain network organization relative to healthy controls (HC). An analysis of the connections' consistency across time reveals significant differences within visual, sensorimotor, and default mode regions where HC subjects show higher consistency than SZ. The introduced statelet-approach enables the handling of dynamic information in cross-modal applications to study healthy and disordered brains and multi-modal fusion within a single dataset.

show abstract

Section: Kernel Density Estimator (Kde)mentioning

confidence: 99%

Section: Definitions and Backgroundmentioning

confidence: 99%

Statelets: High dimensional predominant shapes in dynamic functional network connectivity

Rahaman

Damaraju

Saha

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Multiple neuroimaging algorithms have been federated and can be run in COINSTAC. Examples of implemented algorithms include decentralized voxel-based morphometry (Gazula et al, 2018), decentralized t-distributed stochastic neighbor embedding (Saha et al, 2017(Saha et al, , 2021, decentralized dynamic functional network connectivity (Baker et al, 2020), and decentralized support vector machine with differential privacy (Sarwate et al, 2014).…”

Section: Introductionmentioning

confidence: 99%

Federated analysis in COINSTAC reveals functional network connectivity and spectral links to smoking and alcohol consumption in nearly 2,000 adolescent brains

Gazula

Rootes-Murdy

Holla

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

With the growth of decentralized/federated analysis approaches in neuroimaging, the opportunities to study brain disorders using data from multiple sites has grown multi-fold. One such initiative is the Neuromark, a fully automated spatially constrained independent component analysis (ICA) that is used to link brain network abnormalities among different datasets, studies, and disorders while leveraging subject-specific networks. In this study, we implement the neuromark pipeline in COINSTAC, an open-source neuroimaging framework for collaborative/decentralized analysis. Decentralized analysis of nearly 2000 resting-state functional magnetic resonance imaging datasets collected at different sites across two cohorts and co-located in different countries was performed to study the resting brain functional network connectivity changes in adolescents who smoke and consume alcohol. Results showed hypoconnectivity across the majority of networks including sensory, default mode, and subcortical domains, more for alcohol than smoking, and decreased low frequency power. These findings suggest that global reduced synchronization is associated with both tobacco and alcohol use. This work demonstrates the utility and incentives associated with large-scale decentralized collaborations spanning multiple sites.

show abstract

“…Recently, there is growing interest in developing techniques to classify subjects into diagnostic groups using functional connectivity in various research domains , Kim et al 2016, Plitt, Barnes, and Martin 2015, Rahaman et al 2019, Saha et al 2017, Saha et al 2019. Some recent studies have evaluated classification among bipolar and schizophrenia patients using the features generated from functional connectivity (Arbabshirani et al 2013, Shen et al 2010, Su et al 2013.…”

Section: Introductionmentioning

confidence: 99%

A classification-based approach to estimate the number of resting fMRI dynamic functional connectivity states

Saha

Damaraju

Rashid

et al. 2020

Preprint

View full text Add to dashboard Cite

Recent work has focused on the study of dynamic (vs static) brain connectivity in resting fMRI data. In this work, we focus on temporal correlation between time courses extracted from coherent networks or components called functional network connectivity (FNC). Dynamic functional network connectivity (dFNC) is most commonly estimated using a sliding window-based approach to capture short periods of FNC change. These data are then clustered to estimate transient connectivity patterns or states. Determining the number of states is a challenging problem. The elbow criterion is a widely used approach to determine the optimal number of states. In our work, we present an alternative approach that evaluates classification (e.g. healthy controls versus patients) as a measure to select the optimal number of states (clusters). We apply different classification strategies to perform classification between healthy controls (HC) and patients with schizophrenia (SZ) for different numbers of states (i.e. varying the model order in the clustering algorithm). We compute cross-validated accuracy for different model orders to evaluate the classification performance. Our results are consistent with our earlier work which shows that overall accuracy improves when dynamic connectivity measures are used separately or in combination with static connectivity measures. Results also show that the optimal model order for classification is different from that using the standard k-means model selection method and that such optimization improves resulting in cross-validated accuracy. The optimal model order obtained from the proposed approach also gives significantly improved classification performance over the traditional model selection method. In sum, the observed results suggest that if one's goal is to perform classification, using the proposed approach as a criterion for selecting the optimal number of states in dynamic connectivity analysis leads to improved accuracy in hold-out data.

show abstract

Privacy-preserving quality control of neuroimaging datasets in federated environment

Cited by 6 publications

References 55 publications

Statelets: High dimensional predominant shapes in dynamic functional network connectivity

Statelets: High dimensional predominant shapes in dynamic functional network connectivity

Federated analysis in COINSTAC reveals functional network connectivity and spectral links to smoking and alcohol consumption in nearly 2,000 adolescent brains

A classification-based approach to estimate the number of resting fMRI dynamic functional connectivity states

Contact Info

Product

Resources

About