FOKUS-FUNKTION:
Biomarkers in Network Neuroscience
Minimum spanning tree analysis of brain networks:
A systematic review of network size effects,
sensitivity for neuropsychiatric pathology,
and disorder specificity
N. Blomsma1*, B. de Rooy1*, F. Gerritse1, R. van der Spek1, P. Tewarie2, A. Hillebrand2,
W. M. Otte3, C. J. Stam2, and E. van Dellen1,4
1University Medical Center Utrecht, Department of Psychiatry, Brain Center, Utrecht, die Niederlande
2Amsterdam UMC, Vrije Universiteit Amsterdam, Department of Neurology and Department of Clinical
Neurophysiology and MEG center, Amsterdam Neuroscience, Amsterdam The Netherlands
3University Medical Center Utrecht, Department of Child Neurology, Brain Center, Utrecht, die Niederlande
4University Medical Center Utrecht, Department of Intensive Care Medicine, Brain Center, Utrecht, die Niederlande
*Equal contribution.
Schlüsselwörter: Minimum spanning tree, network neuroscience, transdiagnostic, multimodal, network size
ABSTRAKT
Brain network characteristics’ potential to serve as a neurological and psychiatric pathology
biomarker has been hampered by the so-called thresholding problem. The minimum spanning
tree (MST) is increasingly applied to overcome this problem. It is yet unknown whether this
approach leads to more consistent findings across studies and converging outcomes of either
disease-specific biomarkers or transdiagnostic effects. We performed a systematic review on
MST analysis in neurophysiological and neuroimaging studies (N = 43) to study consistency
of MST metrics between different network sizes and assessed disease specificity and
transdiagnostic sensitivity of MST metrics for neurological and psychiatric conditions. Analyse
of data from control groups (12 Studien) showed that MST leaf fraction but not diameter
decreased with increasing network size. Studies showed a broad range in metric values,
suggesting that specific processing pipelines affect MST topology. Contradicting findings
remain in the inconclusive literature of MST brain network studies, but some trends were seen:
(1) a more linelike organization characterizes neurodegenerative disorders across pathologies,
and is associated with symptom severity and disease progression; (2) neurophysiological
studies in epilepsy show frequency band specific MST alterations that normalize after
successful treatment; Und (3) less efficient MST topology in alpha band is found across
disorders associated with attention impairments.
ZUSAMMENFASSUNG DES AUTORS
The potential of brain network characteristics to serve as biomarker of neurological and
psychiatric pathology has been hampered by the so-called thresholding problem. Der
minimum spanning tree (MST) is increasingly applied to overcome this problem. Wir
performed a systematic review on MST analysis in neurophysiological and neuroimaging
studies and assessed disease specificity and transdiagnostic sensitivity of MST metrics for
neurological and psychiatric conditions. MST leaf fraction but not diameter decreased
Keine offenen Zugänge
Tagebuch
Zitat: Blomsma, N., de Rooy, B.,
Gerritse, F., van der Spek, R., Tewarie,
P., Hillebrand, A., Otte, W. M., Stam,
C. J., & van Dellen, E. (2022). Minimum
spanning tree analysis of brain
Netzwerke: A systematic review of
network size effects, sensitivity for
neuropsychiatric pathology, Und
disorder specificity. Netzwerk
Neurowissenschaften, 6(2), 301–319. https://doi
.org/10.1162/netn_a_00245
DOI:
https://doi.org/10.1162/netn_a_00245
zusätzliche Informationen:
https://doi.org/10.1162/netn_a_00245
Erhalten: 29 September 2021
Akzeptiert: 10 Marsch 2022
Konkurrierende Interessen: Die Autoren haben
erklärte, dass keine konkurrierenden Interessen bestehen
existieren.
Korrespondierender Autor:
Edwin van Dellen
e.vandellen@umcutrecht.nl
Handling-Editor:
Olaf Sporns
Urheberrechte ©: © 2022
Massachusetts Institute of Technology
Veröffentlicht unter Creative Commons
Namensnennung 4.0 International
(CC BY 4.0) Lizenz
Die MIT-Presse
l
D
Ö
w
N
Ö
A
D
e
D
F
R
Ö
M
H
T
T
P
:
/
/
D
ich
R
e
C
T
.
M
ich
T
.
/
T
/
e
D
u
N
e
N
A
R
T
ich
C
e
–
P
D
l
F
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
N
e
N
_
A
_
0
0
2
4
5
P
D
T
.
F
B
j
G
u
e
S
T
T
Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
Minimum spanning tree in neuropsychiatric pathology: Systematic review
with increasing network size. Contradicting findings remain in the literature on MST brain
network studies, but some trends were seen: (1) a more linelike organization characterizes
neurodegenerative disorders; (2) in epilepsy there are frequency band specific MST alterations
that normalize after successful treatment; Und (3) less efficient MST topology is found across
disorders associated with attention impairments.
EINFÜHRUNG
A biomarker can be defined as a characteristic that is objectively measured and evaluated to
indicate normal biologic processes, pathogenic processes, or pharmacologic responses to a
therapeutic intervention (De Gruttola et al., 2001). In psychiatry, and to a lesser extent in neu-
rology, clinical practice and therapeutic innovation lack biomarkers (First et al., 2018).
Disturbances in the organization of macroscale brain networks are increasingly recognized
as a pathophysiological characteristic of brain disease (Bullmore & Spurns, 2009). A recurrent
finding across neurological and psychiatric disorders is the loss of network efficiency or inte-
gration, and damage to hub regions (Bassett & Bullmore, 2009; Crossley et al., 2014; Stam &
van Straaten, 2012). Brain network metrics may thus have the potential to serve as biomarkers
that will aid the diagnostic process and guide treatment, provided that one or more reliable
and reproducible indicators can be established (Douw et al., 2019; First et al., 2018). Thus far,
Jedoch, different studies describing changes in brain networks for the same disorder have
yielded contradictory results. These results can at least in part be explained by methodological
issues (Fornito et al., 2013; Tijms et al., 2013; van den Heuvel et al., 2017; Van Diessen et al.,
2013). In brain network research, one key issue is the definition of a ‘true’ connection or
edge based on empirical data. Weighted connection estimates from inherently noisy data will
introduce false positive connections in the network. Außerdem, comparing networks with
differences in mean connection weights introduces possible bias since this influences graph
measurements such as clustering coefficient and path length (van Wijk et al., 2010).
A frequently used solution is to threshold connection weights, but this introduces the so-called
“thresholding problem”: the choice of a threshold is often arbitrary (van Wijk et al., 2010; Fornito
et al., 2012; Stam et al., 2014; Zalesky et al., 2016). Fixed thresholds may lead to different
connection densities across subjects that bias higher order graph characteristics, while fixed
densities include noisy edges or discard true edges (Ercsey-Ravasz et al., 2013). Attempts to
normalize data may thus inherently introduce bias to graph theoretical measures. The thresh-
olding problem contributes to poor reproducibility and limited interpretability of results.
Stam and others put forward a theoretical solution, at least at the macroscale brain network
analysis level, by reconstructing the minimum spanning tree (MST) (Stam et al., 2014). A tree is
defined as a connected graph with a path between each pair of nodes, without forming any
Schleifen. A spanning tree is defined as a subgraph that includes all N nodes of the original graph
and N − 1 links (M). When the sum of the weights of the links is minimized, this is called a MST
of the connected weighted graph (Kruskal, 1956; Prim, 1957). Wichtig, the MST will serve
as the backbone of information flow in the network under conditions where link weights in the
original graph show strong fluctuations (Van Mieghem & Van Langen, 2005). Advantages of
this approach are that the MST of the weighted connectivity matrix is unique, provided that its
weights are unique. MST connections may be binarized to avoid density effects, and the num-
ber of links in the MST is fixed (Stam et al., 2014). And, importantly, MST characteristics can be
interpreted along the lines of conventional metrics that characterize network topology
Minimum spanning tree:
A spanning tree is defined as a
subgraph that includes all N nodes of
the original graph and N − 1 links
(M). When the sum of the weights of
the links is minimized, this is called
a minimum spanning tree of the
connected weighted graph.
Netzwerkneurowissenschaften
302
l
D
Ö
w
N
Ö
A
D
e
D
F
R
Ö
M
H
T
T
P
:
/
/
D
ich
R
e
C
T
.
M
ich
T
.
/
T
/
e
D
u
N
e
N
A
R
T
ich
C
e
–
P
D
l
F
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
N
e
N
_
A
_
0
0
2
4
5
P
D
T
.
F
B
j
G
u
e
S
T
T
Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
Minimum spanning tree in neuropsychiatric pathology: Systematic review
(Tewarie et al., 2015A). Tewarie and others further demonstrated in silico that MST analysis
indeed is reliable and reproducible, in the sense that it is relatively insensitive to bias and noise
in simulated connectivity data (Tewarie et al., 2015A). MST metrics were unaffected by
changes in density and average connectivity of a network, and global MST metrics are even
robust against substantial levels of noise in the input data.
Other, more data-driven thresholding approaches are also available, including efficiency
cost optimization, proportional thresholding, and probabilistic thresholding (De Vico Fallani,
Latora, & Chavez, 2017; van den Heuvel et al., 2017; Váša, Bullmore, & Patel, 2018). Der
MST has been proven to be theoretically and methodologically reliable for specific imaging
modalities (van Dellen et al., 2018). Over the last few years, MST analysis has been applied to
neuroimaging and neurophysiological data obtained with various acquisition techniques, solch
as magnetic resonance imaging (MRT), functional magnetic resonance imaging (fMRT), Diffusion
tensor imaging (DTI), electroencephalography (EEG), and magnetoencephalography (MEG).
Hier, we performed a systematic literature review and critically assessed whether the use of
MST analysis to characterize brain networks holds promise for establishing the reproducibility
and reliability necessary for the development of brain network–based biomarkers for neuro-
logical and psychiatric disorders. Erste, we analyzed how MST metrics are affected by different
imaging modalities, node, and link definitions by comparing MST characteristics in empirical
studies of healthy controls.
Zweitens, we assessed how results from different MST studies compare within and across
categories of pathology, and attempted to interpret the changes in brain networks that occur in
brain disease from a transdiagnostic perspective. After a systematic search for clinical MST
Studien, we categorized findings on (1) neurodevelopmental disorders, (2) adult psychiatric
disorders, (3) neurodegenerative disorders, (4) multiple sclerosis, (5) epilepsy, Und (6) andere
neuropsychiatric disorders.
METHODEN
Search Term and Search Strategy
References were identified through searches of PubMed and EMBASE, using an array of terms
covering brain connectivity in combination with MST. Exact search terms are specified in the
zusätzliche Informationen. Außerdem, Google Scholar was used to search for articles from
2005 and onward citing the following key papers: Kruskal (1956), a seminal paper on graph
theory; Stam et al. (2014), an extensive review on the methodological state of affairs in brain
network research which has proposed the MST as a possible solution for various methodolog-
ical issues; Tewarie et al. (2015A), a study showing the relevance and reliability of MST graph
theoretical approach for studying brain networks (Kruskal, 1956; Stam et al., 2014; Tewarie
et al., 2015A). For the paper by Kruskal, search results were further narrowed down by using
the following search terms “brain OR neuronal OR cerebral.” Searches were conducted from
inception to May 2020.
The resulting articles were reviewed for relevance on title and abstract by two independent
raters (FG and NB and/or BdR). If the article was deemed potentially relevant, the full text was
also reviewed. In case of uncertainty about whether an article was eligible for inclusion, a third
rater (EvD) coreviewed the article and was then included or excluded by consensus.
Articles were included when they met the following criteria: published in English, assessing
macroscale, whole-brain network topology through the use of one or more of these four imag-
ing techniques that are broadly used in studies on neurological and psychiatric disorders:
Netzwerkneurowissenschaften
303
l
D
Ö
w
N
Ö
A
D
e
D
F
R
Ö
M
H
T
T
P
:
/
/
D
ich
R
e
C
T
.
M
ich
T
.
/
/
T
e
D
u
N
e
N
A
R
T
ich
C
e
–
P
D
l
F
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
N
e
N
_
A
_
0
0
2
4
5
P
D
.
T
F
B
j
G
u
e
S
T
T
Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
Minimum spanning tree in neuropsychiatric pathology: Systematic review
Diameter:
The largest distance (in number of
edges) between any two nodes,
normalized for the total number of
edges. Diameter is a measure of
network efficiency. An increase in
diameter, means a decrease in global
efficiency, whereas a low diameter
indicates a more efficient information
flow between brain regions.
Leaf fraction:
The leaf fraction (Lf) is the leaf
number (L) divided by the maximum
possible leaf number. Leaf fraction is
considered measure of centrality.
Leaf number:
The number of nodes in graph with
only one connection.
Kappa:
A measure that relates to the spread
of information across the tree. Wann
kappa is low there is a low number of
highly connected nodes.
Tree hierarchy:
Tree hierarchy quantifies the trade-off
between large-scale integration in
the minimum spanning tree and the
overload of central nodes.
electroencephalography (EEG), magnetoencephalography (MEG), functional MRI (fMRT), DTI;
conducted during resting-state and without intervention; having constructed a minimum span-
ning tree and reported at least one of the following MST measures: diameter, leaf fraction, leaf
number, and kappa and tree hierarchy in a population with a neurological or psychiatric dis-
Befehl. When available, numeric values for these measures were extracted from the original
articles or Supporting Information.
MST Metrics
MST measures that were analyzed included diameter, leaf fraction, and kappa and tree hier-
archy, which describe the integration and efficiency of the network. A description of these
measurements is provided in Table 1, where Figure 1 provides a schematic overview of
MST topologies and corresponding characteristics.
A small diameter combined with a high leaf fraction characterizes a more starlike, zentral-
ized network. Im Gegensatz, a large diameter with a low leaf fraction indicates a more linelike
topology and decentralized network. A starlike network is characterized by short paths
between the most remote nodes and thus facilitates efficient transfer of information across
the network. Jedoch, the central node is burdened by a relatively large flow of information
in such a network, possibly creating a greater chance of overload. Starlike networks will also
be more vulnerable to targeted attacks to central hub nodes (Otte et al., 2015).
Meta-Analysis
Due to limited available data, kappa and tree hierarchy were excluded from quantitative anal-
ysis. The MST variables diameter and leaf fraction were included in further quantitative anal-
yses. Erste, we tested if MST variables were affected by network size. Mean values of MST
metrics from healthy control groups were used for this analysis. Due to a limited number of
data points, and the fact that some fMRI-based studies had considerably more nodes, Wir
decided to use nonparametric linear regression method according to Siegel and others, Zu
ensure robust regression (Siegel, Donner, & Engel, 2012). For EEG and MEG data, we followed
the authors’ frequency bands.
Zweite, to analyze the network deviations in various brain disorders and interpret results
from a transdiagnostic perspective, mean group effects on MST metrics were analyzed based
on comparisons between clinical and control groups. We performed both fixed-effect and
random-effects meta-analysis on the standardized difference of the mean estimates of mean
leaf fraction per study and mean diameter per study. Studies were stratified by imaging modal-
ität. MEG/EEG studies were aggregated, but stratified for each frequency band. Heterogeneity
was assessed using I^2 calculated based on Cochran’s Q. Analyses were performed using the
meta package in R 3.6.1.
Endlich, we provided a narrative review to evaluate if MST analysis holds promise for bio-
marker development of brain disease, where insufficient data were available to draw definitive
conclusions on quantitative outcomes.
ERGEBNISSE
The search resulted in 43 included studies in this review, 34 described a patient-control study
Design, Und 9 studies included children. Der 43 studies were further subdivided into papers on
neurodevelopmental disorders (N = 7), adult psychiatric disorders (N = 7), neurodegenerative
disorders (N = 15), epilepsy (N = 6), multiple sclerosis (N = 4), and other disorders (N = 4). Ein
Netzwerkneurowissenschaften
304
l
D
Ö
w
N
Ö
A
D
e
D
F
R
Ö
M
H
T
T
P
:
/
/
D
ich
R
e
C
T
.
M
ich
T
.
/
/
T
e
D
u
N
e
N
A
R
T
ich
C
e
–
P
D
l
F
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
N
e
N
_
A
_
0
0
2
4
5
P
D
T
.
F
B
j
G
u
e
S
T
T
Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
Minimum spanning tree in neuropsychiatric pathology: Systematic review
Symbol
D
Concept
Diameter
LF
Leaf fraction
Th
Tree hierarchy
κ
Kappa
Tisch 1.
Explanation of the MST measurements included in this review
Explanation
A measure of network efficiency and refers to the largest distance (in number of links) between any two
nodes and is normalized for the total number of links: D = d/M, where M is the total number of links or
maximum leaf number (M = n − 1, with n the number of nodes). An increase in diameter, means a
decrease in global efficiency, whereas a low diameter indicates a more efficient information flow
between brain regions.
A measure of centrality and is based on the leaf number; the number of nodes with only one connection.
The leaf fraction (Lf) is the leaf number (L) divided by the maximum possible leaf number: Lf = L/M.
This measure ranges between 2/M, which indicates a linelike topology and a maximum value of
M = n − 1 (n = number of nodes), which indicates a star topology. A lower value of the leaf fraction
indicates a less centralized network topology and a high leaf fraction means that communication
depends strongly on hub nodes (d.h., nodes that play a central role in the network).
Quantifies the trade-off between large-scale integration in the MST and the overload of central nodes,
calculated by Th = L/(2mBCmax) (Boersma et al., 2013), where BCmax stands for the maximum value
of the betweenness centrality among all the nodes in the MST, and BC itself is computed as the fraction
of shortest paths that go through a node. Note that nodal BC and BCmax were not considered
macroscale network characteristics and therefore excluded from analysis in this review. To assure tree
hierarchy ranges between 0 Und 1, the denominator is multiplied by 2. Th ranges between 0 Und
1, where Th approaches 0 if L = 2 (linelike topology) and M approaches infinity. For L = M (starlike
Topologie), Th approaches 0.5. A network topology that optimizes a trade-off between integration and
segregation is hypothesized when Th approaches 1 (Stam et al., 2014).
A measure of the broadness of the degree distribution or the heterogeneity of degrees and relates to the
spread of information across the tree (Stam et al., 2014). A low value of kappa indicates a low number
of highly connected nodes (hubs). High kappa values are especially seen in scale-free networks (Stam
& van Straaten, 2012).
l
D
Ö
w
N
Ö
A
D
e
D
F
R
Ö
M
H
T
T
P
:
/
/
D
ich
R
e
C
T
.
M
ich
T
.
/
/
T
e
D
u
N
e
N
A
R
T
ich
C
e
–
P
D
l
F
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
N
e
N
_
A
_
0
0
2
4
5
P
D
.
T
F
B
j
G
u
e
S
T
T
Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
Schematic depiction of three different minimum spanning trees, with a starlike, intermediate and linelike configuration from left to
Figur 1.
Rechts. The green nodes represent leaf nodes. Central nodes are depicted in orange. Diameter is depicted in red. Individual conditions and the
correlated changes in network topology as described in the discussion section are displayed, with an arrow depicting the direction of the
ändern. For neurodegenerative diseases conditions are displayed left to right from having the least shift toward a more linelike topology
(bvFTD) to the most (PDD). AD, Alzheimer’s disease; bvFTD, behavioral variant of frontotemporal dementia; DLB, dementia with Lewy bodies;
PDD, Parkinson’s disease dementia.
Netzwerkneurowissenschaften
305
Minimum spanning tree in neuropsychiatric pathology: Systematic review
Figur 2. Nonparametric linear regression for number of nodes and normalized leaf fraction (A) and diameter (B). The gray line indicates the
same regression, but excluding studies with more than 250 Knoten; in this analysis for leaf fraction (2.1) the p value is 0.023, with an intercept of
0.63 and a slope of −0.0007. For diameter the p value is 0.426, with an intercept of 0.279 and a slope of −0.001.
overview of the included studies is given in Supporting Information Table S1 and a flowchart of
the systematic selection process can be found in Supporting Information Figure S1.
Network Size and Imaging Modality Effects
To analyze network size effects, control groups from studies in adult populations were used.
Für 12 studies data were available: 5 EEG studies, 2 MEG, 2 fMRT-Studie, 2 DTI study, Und 1
combined MEG/MRI study. Regression analysis showed that leaf fraction decreased with the
number of nodes in the network (slope = −6.91 × 10−4; p = 0.00257; Figure 2A). Values for leaf
fraction in healthy controls ranged from 0.35 Zu 0.859. Diameter did not show significant cor-
relation with the number of nodes (slope = −5.05 × 10−4; p = 0.232; Figure 2B). Values for
diameter in healthy controls ranged from 0.108 Zu 0.401. Insufficient data were available to
provide a quantitative analysis of imaging modality effects, frequency band effects in
EEG/MEG studies, or effects using different connectivity measures/connection definitions
within one modality.
MST Characteristics of Neurological and Psychiatric Disorders
Data from 16 studies were available to calculate transdiagnostic, standardized effects of brain
disorders on MST metrics. Fixed-effect and random-effects meta-analysis were performed on
the standardized difference of the mean estimates of mean leaf fraction per study and mean
diameter per study. EEG and MEG studies were aggregated per frequency band. fMRI studies
and DTI studies were analyzed separately because of unknown imaging modality effects and
differences in network size compared to EEG/MEG studies. Figur 3 and Figure 4 show the
disease effect for diameter and leaf fraction, jeweils.
A significant effect is seen for delta-band diameter (SMD = 0.322, 95% CI = 0.092; 0.551)
and leaf fraction (fixed effects model: SMD = −0.295, 95% CI = −0.483; −0.107; random-
effects model SMD = −0.336, 95% CI = −0.594; -−0.078), and for diameter in gamma band
(SMD = 0.470, 95% CI = 0.087; 0.853), implying a cross-disorder effect of a shift toward a
more linelike network.
Due to overall levels of heterogeneity (siehe Abbildung 3 and Figure 4), we concluded that it is
not possible to show any generic disease effects across disorders and modalities on diameter
and leaf fraction in our meta-analysis. Because of the small number of studies per disease cat-
egory, no separate analyses for each disease category was conducted; instead these effects are
described in a qualitative matter in the following sections.
Netzwerkneurowissenschaften
306
l
D
Ö
w
N
Ö
A
D
e
D
F
R
Ö
M
H
T
T
P
:
/
/
D
ich
R
e
C
T
.
M
ich
T
.
T
/
/
e
D
u
N
e
N
A
R
T
ich
C
e
–
P
D
l
F
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
N
e
N
_
A
_
0
0
2
4
5
P
D
T
.
F
B
j
G
u
e
S
T
T
Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
Minimum spanning tree in neuropsychiatric pathology: Systematic review
l
D
Ö
w
N
Ö
A
D
e
D
F
R
Ö
M
H
T
T
P
:
/
/
D
ich
R
e
C
T
.
M
ich
T
.
/
/
T
e
D
u
N
e
N
A
R
T
ich
C
e
–
P
D
l
F
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
N
e
N
_
A
_
0
0
2
4
5
P
D
T
.
F
B
j
G
u
e
S
T
T
Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
Figur 3. Forest plots for fixed-effect and random-effects meta-analysis on the standardized difference of the mean estimates for mean diam-
eter. Studies are stratified by imaging modality. MEG/EEG studies are aggregated, but stratified for each frequency band. Low heterogeneity
indicates that the included studies agree about the magnitude and direction of effect. The p value indicates whether the calculated hetero-
geneity deviates significantly. Meta-analyses with more studies tend to have a higher power to detect significant heterogeneity. SMD, Stan-
dardized mean difference; 95%-CI, 95% confidence interval; DER, estimate of treatment effect, Zum Beispiel, log hazard ratio or risk difference;
seTE, standard error of treatment estimate; ADHD, attention-deficit hyperactivity disorder; LHON, Leber’s hereditary optic neuropathy; MCI,
mild cognitive impairment; MS, multiple sclerosis; DLB, dementia with Lewy bodies.
Netzwerkneurowissenschaften
307
Minimum spanning tree in neuropsychiatric pathology: Systematic review
l
D
Ö
w
N
Ö
A
D
e
D
F
R
Ö
M
H
T
T
P
:
/
/
D
ich
R
e
C
T
.
M
ich
T
.
/
T
/
e
D
u
N
e
N
A
R
T
ich
C
e
–
P
D
l
F
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
N
e
N
_
A
_
0
0
2
4
5
P
D
.
T
F
B
j
G
u
e
S
T
T
Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
Forest plots for fixed-effect and random-effects meta-analysis on the standardized difference of the mean estimates for mean leaf
Figur 4.
fraction. Studies are stratified by imaging modality. MEG/EEG studies are aggregated, but stratified for each frequency band. SMD, standard-
ized mean difference; 95%-CI, 95% confidence interval; DER, estimate of treatment effect, Zum Beispiel, log hazard ratio or risk difference; seTE,
standard error of treatment estimate; ADHD, attention-deficit hyperactivity disorder; MCI, mild cognitive impairment; AD, Alzheimer’s disease;
DLB, dementia with Lewy bodies; LHON, Leber’s hereditary optic neuropathy; MS, multiple sclerosis; PD, Parkinson’s disease.
Netzwerkneurowissenschaften
308
Minimum spanning tree in neuropsychiatric pathology: Systematic review
Neurodevelopmental Disorders
We included two studies on attention-deficit/hyperactivity disorder (ADHD), one study on
autism spectrum disorder (ASD), three studies on dyslexia, and one study on language-based
learning disorder, of which 6 were EEG studies and 1 was an fMRI study. An overview of
the mean MST values and standard deviation (SD) is included as Supporting Information
Table S2. The studies comprised a total number of 261 patients with neurodevelopmental
disorders.
Children with ASD showed a lower EEG alpha-band leaf fraction than healthy controls
(Zeng et al., 2017). Wang and others found a lower fMRI leaf fraction and kappa and tree
hierarchy in children with ADHD compared to normal developing children (Wang et al.,
2020). Im Gegensatz, Janssen and others found a lower EEG alpha-band diameter and higher leaf
fraction and tree hierarchy in children with ADHD than typically developing children and a
higher beta-band leaf fraction (Janssen et al., 2017).
Fraga González and others found dyslexic children to have a significantly lower leaf frac-
tion and higher diameter in EEG theta band than typically reading children (Fraga González
et al., 2016), while Xue and others found no significant differences (Xue et al., 2020). Fraga
González and others found a higher kappa in dyslexic young adults (Fraga González et al.,
2018). Infants at risk for developing a language-based learning disorder showed a higher leaf
fraction than typically developing children (Zare et al., 2016).
Adult psychiatric disorders. We included seven studies on adult psychiatric disorders, mit 483
Patienten: four studies on psychotic disorders, one study on bipolar disorder and psychotic dis-
Befehl, one study on major depressive disorder, and one study on internet addiction. Imaging
modalities included EEG (N = 4), fMRT (N = 1), and DTI (N = 2). An overview of the available
mean MST values and SD is included as Supporting Information Table S3.
Anjomshoa and others found a higher DTI diameter and lower kappa and leaf number in
schizophrenia patients than healthy controls (Anjomshoa et al., 2016). Im Gegensatz, Van Dellen
and others found no significant differences in MST topology between patients with a psychotic
disorder, individuals with subclinical psychotic symptoms, and healthy controls, neither with
fMRI nor with DTI-based networks (van Dellen et al., 2016; van Dellen et al., 2020). Krukow
and others reported findings seemingly inconsistent with the studies above, das ist, a smaller
diameter and leaf fraction in EEG delta and lower gamma band and smaller diameter in EEG
beta band. Importantly in this EEG study, all patients were treated with atypical antipsychotics
(Krukow et al., 2019). Jonak and others found a higher gamma-band hierarchy but lower beta-
band hierarchy in first-episode psychosis patients compared to patients with longer illness
Dauer, but found no differences in diameter, leaf fraction, or kappa between these groups
(Jonak et al., 2019).
Bipolar-I patients were found to have a lower fMRI leaf fraction and kappa than controls
and patients with subclinical psychosis, and a lower leaf fraction than patients with schizo-
phrenia (van Dellen et al., 2020). This study found no differences in global MST network
topology between patients on antipsychotic medication or lithium and those who did not use
medication. Li and others found a higher EEG theta-band leaf fraction for patients with major
depressive disorder than healthy controls (Li et al., 2017).
Endlich, Wang and others found that alpha- and beta-band EEG MST was more starlike in
subjects with internet addiction than controls (Wang et al., 2019): a higher kappa and lower
diameter correlated with higher addiction severity.
Netzwerkneurowissenschaften
309
l
D
Ö
w
N
Ö
A
D
e
D
F
R
Ö
M
H
T
T
P
:
/
/
D
ich
R
e
C
T
.
M
ich
T
.
/
T
/
e
D
u
N
e
N
A
R
T
ich
C
e
–
P
D
l
F
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
N
e
N
_
A
_
0
0
2
4
5
P
D
.
T
F
B
j
G
u
e
S
T
T
Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
Minimum spanning tree in neuropsychiatric pathology: Systematic review
Neurodegenerative diseases. Fifteen studies were included, umfassend 760 patients with neu-
rodegenerative disorders: seven EEG, five MEG, two fMRI and one DTI study. An overview of
the available mean MST values and SD is included as Supporting Information Table S4.
Synucleinopathies are neurodegenerative diseases characterized by the abnormal neural
accumulation of alpha-synuclein proteins. Parkinson’s disease (PD) and dementia with Lewy
bodies (DLB) are types of synucleinopathies, which may be diagnosed based on the neuroan-
atomical spreading pattern of the proteins and clinical presentation. Three studies reported
MST disturbances in synucleinopathies, which may also reflect disease progression.
Olde Dubbelink and others studied MEG recordings of de novo PD patients, 43 chronic PD
patients and 14 Kontrollen, and included a follow-up measurement after 4 Jahre (Olde
Dubbelink et al., 2014). They found lower leaf fraction and tree hierarchy in the (upper) alpha
band in PD patients compared to controls. Leaf number (theta band) and tree hierarchy (delta
band) were decreased at follow-up in the PD group. In the control group, alpha-band leaf
fraction and beta-band tree hierarchy decreased at follow-up.
Utianski and others also found a lower diameter and higher leaf fraction in EEG delta and
theta-band recordings of cognitively normal PD (PD-CN) than healthy controls. Patients with
PD and mild cognitive impairment (PD-MCI) or dementia (PDD) had a lower leaf fraction
(upper alpha band) when compared to PD-CN patients (Utianski et al., 2016). Der (lower
alpha band) diameter was also higher in the PDD compared to PD-CN patients.
Peraza and others found higher theta-band diameter and lower alpha-band leaf fraction in
EEGs of patients with PDD, DLB, and AD compared to controls, indicating more linelike net-
works in the patient groups. A classifier between AD and DLB based on MST and connectivity
values reached 80% sensitivity and 85% specificity (Peraza et al., 2018). A second study also
reported a higher diameter and lower leaf fraction in DLB compared to patients with AD and
control subjects, but only in the alpha band (van Dellen et al., 2015).
Of interest, (theta and alpha band) leaf fraction was associated with cognitive decline in
cross-sectional analyses of three EEG studies of DLB and PDD patients (Peraza et al., 2018;
Utianski et al., 2016; van Dellen et al., 2015). The severity of PD motor symptoms was asso-
ciated with lower MEG delta-band leaf number and tree hierarchy.
Taken together, these studies found disease effects in different frequency bands. A tendency
toward less integrated, more linelike MST topology was reported across studies and was asso-
ciated with clinical deterioration in synucleinopathies.
AD is the most common cause of dementia, and is subject of six EEG studies that char-
acterize MST topology. Das and Puthankattil found a higher diameter in 13 mildly cognitively
impaired AD patients compared to 20 healthy controls across a range of (delta, theta, lower
alpha, upper alpha, and beta) frequency bands and in a variety of recording protocols (Das &
Puthankattil, 2020). A lower leaf fraction was observed in lower alpha band in AD compared
to controls. Yu and others found a similarly higher diameter and lower leaf fraction and kappa
in AD patients’ alpha-band EEG recordings compared to subjects with subjective cognitive
Abfall. A lower leaf fraction and kappa in AD compared to patients with the behavioral
variant of frontotemporal dementia (bvFTD) (Yu et al., 2016). Studies by van Dellen (EEG
alpha band) and Peraza (EEG theta and alpha band) found MST characteristics of AD patients
to be in between those of the DLB patients and controls (Peraza et al., 2018; van Dellen
et al., 2015).
Five studies reported MST metrics related to at-risk states for AD dementia, with inconsis-
tent results. An fMRI study showed that compared to controls, patients with MCI had a more
Netzwerkneurowissenschaften
310
l
D
Ö
w
N
Ö
A
D
e
D
F
R
Ö
M
H
T
T
P
:
/
/
D
ich
R
e
C
T
.
M
ich
T
.
/
T
/
e
D
u
N
e
N
A
R
T
ich
C
e
–
P
D
l
F
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
N
e
N
_
A
_
0
0
2
4
5
P
D
T
.
F
B
j
G
u
e
S
T
T
Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
Minimum spanning tree in neuropsychiatric pathology: Systematic review
starlike network, while AD patients had a more linelike network (Wang et al., 2018). Two MEG
studies found regional but no global metric differences in MCI patients compared to controls
(Jacini et al., 2018; López et al., 2017), while an EEG study found a more linelike MST in the
delta band, which contributed to a classifier for MCI versus controls (Požar, Giordani, &
Kavcic, 2020).
Sorrentino and others studied MEG networks concerning insulin growth factor-1 (IGF-1),
which has been suggested as a brain atrophy marker related to the risk of developing AD
(Sorrentino et al., 2017). IGF-1 was correlated with beta-band leaf fraction, tree hierarchy,
and theta-band leaf fraction, suggesting an association with less integrated network topology.
Frontotemporal dementia (FTD) is characterized by progressive cell loss in frontal and
temporal lobes. One variant of FTD is the behavioral variant mentioned above. Two studies
report on MST topology in bvFTD. Yu and others found no MST disturbances in EEG record-
ings of 48 bvFTD patients compared to subjects with SMC, Und, as mentioned above, did find
disturbances in AD compared to bvFTD (Yu et al., 2016). Im Gegensatz, Saba and others found a
higher diameter and lower leaf fraction in resting-state fMRI recordings of patients with bvFTD
than controls (Saba et al., 2019). MST findings in FTD thus remain inconclusive.
Fraschini and others compared resting-state EEG networks of 21 patients with amyotrophic
lateral sclerosis (ALS) (a degenerative motor neuron disease affecting upper and lower motor
Neuronen) Und 16 control patients (Fraschini et al., 2016). A lower beta-band leaf fraction was
found in ALS patients compared to controls, and lower leaf fraction, kappa, and tree hierarchy
correlated with worse disability scores. Im Gegensatz, Sorrentino and others found more pro-
nounced disturbances in MEG recordings of patients with advanced stage ALS (N = 24) als
early stage ALS (N = 26) compared to healthy controls (N = 25), with a pattern of higher tree
hierarchy, kappa and leaf fraction across frequency bands, suggesting more starlike networks
with progressive ALS (Sorrentino et al., 2018).
Endlich, Jonak and others found lower leaf fraction and tree hierarchy in DTI networks of 15
Leber’s hereditary optic neuropathy (LHON) patients to 17 Kontrollen (Jonak et al., 2021). Der
more linelike MST in LHON patients correlated with illness duration.
Multiple sclerosis. Tewarie and others performed three subsequent MST studies in multiple
sclerosis (MS) using MEG recordings, while Nauta and colleagues performed a fourth analysis
partially on the same cohort. An overview of the available mean MST values and SD is
included as Supporting Information Table S5.
Erste, Tewarie and others found a frequency-specific effect on MEG networks compared to
21 early MS patients (relapsing-remitting subtype) Zu 17 Kontrollen (Tewarie et al., 2014B).
Ähnlich, diameter was lower in the theta band but higher in the upper alpha band in the
MS group, while the opposite pattern was seen for the leaf fraction. Außerdem, the upper
alpha-band kappa and hierarchy were lower in the MS group. Taken together, diese Ergebnisse
pointed toward a more starlike theta-band network but more linelike alpha- and beta-band
network in MS patients.
Zweitens, this group compared MEG data of 102 MS patients (67% relapsing-remitting
subtype, 21% secondary- Und 12% primary-progressive subtype) Und 42 Kontrollen (Tewarie
et al., 2014A). Patients showed lower leaf fraction, hierarchy, and kappa than controls in
the upper alpha band, but not in other frequency bands. A third study analyzed MEG and fMRI
recordings of 86 MS patients (around 6 years after diagnosis) Und 21 healthy controls (Tewarie
et al., 2015B). They found group differences in MEG recordings but not in fMRI data; MS
patients had a lower leaf fraction in the upper alpha band and a lower kappa in the frequencies
Netzwerkneurowissenschaften
311
l
D
Ö
w
N
Ö
A
D
e
D
F
R
Ö
M
H
T
T
P
:
/
/
D
ich
R
e
C
T
.
M
ich
T
.
T
/
/
e
D
u
N
e
N
A
R
T
ich
C
e
–
P
D
l
F
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
N
e
N
_
A
_
0
0
2
4
5
P
D
.
T
F
B
j
G
u
e
S
T
T
Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
Minimum spanning tree in neuropsychiatric pathology: Systematic review
aus 0.5 Zu 13 Hz. Außerdem, in MS patients, a more linelike topology was associated
with clinical disability (delta/theta-band leaf fraction and kappa; lower alpha-band kappa),
thalamic atrophy (theta/alpha-band leaf fraction and kappa) and cognitive functioning
(alpha-band kappa).
Nauta and others then showed in an extended sample of this cohort that a lower beta-band
diameter and lower delta-band leaf fraction predicted 15% of the variance in cognitive decline
nach 5 Jahre, independent of structural damage. Cross-sectional analyses showed that lower
tree hierarchy was especially related to worse cognition, independent of the frequency band
(Nauta et al., 2020).
MST studies in MS patients are based on studies from one research group and partially the
same cohort of respectable sample size. A complex, frequency-dependent pattern of alter-
ations in MST characteristics emerges from these studies, which relates to clinical impairments,
but a straightforward interpretation seems impossible.
Epilepsy. Six studies on MST metrics and epilepsy were included in this review, consisting of
four EEG studies, one MEG study, and one fMRI study, with a total number of 231 Patienten. Ein
overview of the available mean MST values and SD is included as Supporting Information
Table S6. All studies focused on interictal recordings. Three studies included patients with
childhood epilepsy.
Van Diessen and others found a higher MST diameter and lower leaf fraction in the delta
band in EEGs of drug-naïve children with newly diagnosed focal epilepsy than in controls (Transporter
Diessen et al., 2016). They found an opposite difference in topology in the upper alpha band,
with a lower diameter and higher leaf fraction in the focal epilepsy group. No differences
were found when comparing children with generalized epilepsy to the focal epilepsy group
or controls.
The same authors studied the effect of sleep deprivation, which lowers the seizure threshold
and increases interictal EEG abnormalities, on functional EEG networks in children with focal
epilepsy compared to age-matched controls (van Diessen et al., 2014). Alpha-band diameter
increased and the leaf fraction decreased after sleep deprivation in patients, while the opposite
pattern was seen in controls. They speculated that this shift in network organization after sleep
deprivation in epilepsy patients follows literature showing a more regular alpha-band network
organization during the ictal state; sleep deprivation may thus cause a shift toward an ictal
Netzwerkstatus. Kinney-Lang and others analyzed the EEG networks of preschool children with
epilepsy (both focal and generalized) and cognitive impairment (Kinney-Lang et al., 2019).
They found that worse performance on cognitive tasks was associated with lower (alpha/beta
band) diameter and higher leaf fraction.
Interictal MST characteristics may be associated with disease severity and treatment resis-
tanz. DeSalvo and others studied fMRI data of 40 patients with medically intractable tempo-
ral lobe epilepsy who were to undergo epilepsy surgery. They found that preoperative MST
topology differed between patients who became seizure-free after surgery as compared to
patients who would not (DeSalvo et al., 2020). Leaf fraction was 9% lower and tree hierarchy
War 10% lower in patients with ongoing seizures than in seizure-free patients, suggesting less
integrated networks in patients with worse outcomes. A similar finding was reported by Van
Dellen and colleagues, where the preoperative diameter in MEG recordings of patients with
lesional epilepsy correlated with higher seizure frequency (4–10 Hz) (van Dellen et al., 2013).
The alpha-band leaf fraction increased in patients after successful epilepsy surgery, Aber
remained unchanged after surgery in patients with ongoing seizures. A third study reported
Netzwerkneurowissenschaften
312
l
D
Ö
w
N
Ö
A
D
e
D
F
R
Ö
M
H
T
T
P
:
/
/
D
ich
R
e
C
T
.
M
ich
T
.
/
T
/
e
D
u
N
e
N
A
R
T
ich
C
e
–
P
D
l
F
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
N
e
N
_
A
_
0
0
2
4
5
P
D
.
T
F
B
j
G
u
e
S
T
T
Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
Minimum spanning tree in neuropsychiatric pathology: Systematic review
on adults with pharmacoresistant epilepsy and found that theta-band MST diameter decreased
in responders to vagal nerve stimulation as an add-on treatment, but not in nonresponders
(Fraschini et al., 2014). These findings indicate that successful epilepsy treatment is associated
with a recovery toward a more integrated (d.h., more starlike) network organization.
Other studies. We included four studies with a total of 98 patients with other disorders, das ist,
delirium (N = 2), migraine (N = 1) and meningioma (N = 1), including one EEG, two MEG,
and one fMRI study. An overview of the available mean MST values and SD is included as
Supporting Information Table S7.
One EEG and one fMRI study analyzed the MST topology related to delirium. Numan and
others studied EEG recordings of patients who had undergone cardiac surgery and compared
patients who developed hypoactive delirium to patients without delirium (Numan et al.,
2017). They found a lower alpha-band leaf fraction in patients with hypoactive delirium
compared to the nondelirious controls. A resting-state fMRI study by Van Montfort and others
similarly found a higher diameter and lower leaf fraction in brain networks of patients during
delirium, which normalized after delirium resolved (van Montfort et al., 2018). In this pilot
study of nine patients, a lower leaf fraction and tree hierarchy correlated to a longer delirium
Dauer, used as a proxy for syndrome severity. Both MST studies thus indicate a less inte-
grated network during a state of delirium.
One MEG study compared patients with migraine to controls but found no group differ-
ences in global MST metrics (Nieboer et al., 2020). Endlich, Van Nieuwenhuizen and others
found a lower theta-band maximum MST degree, but no other differences in global MST
metrics, in MEG recordings of 20 meningioma patients compared to healthy controls (Transporter
Nieuwenhuizen et al., 2018).
DISKUSSION
l
D
Ö
w
N
Ö
A
D
e
D
F
R
Ö
M
H
T
T
P
:
/
/
D
ich
R
e
C
T
.
M
ich
T
.
/
/
T
e
D
u
N
e
N
A
R
T
ich
C
e
–
P
D
l
F
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
N
e
N
_
A
_
0
0
2
4
5
P
D
T
.
F
B
j
G
u
e
S
T
T
Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
This systematic review shows that the past decade of MST analysis has not yet led to definitive
neuropsychiatric symptoms or disease biomarkers. This finding may partially explain network
size and modality differences across studies: MST leaf fraction (but not diameter) was found to
correlate with network size significantly.
Previous meta-analyses on cross-disorder data have shown general disease effects on brain
networks across different neuropsychiatric disorders (Crossley et al., 2014). For Alzheimer’s
disease and epilepsy, reviews have suggested modality-invariant disease effects on network
characteristics (Tijms et al., 2013; Van Diessen et al., 2013). We therefore assessed whether
MST characteristics show modality- and disorder-independent alterations in neuropsychiatric
disorders. Our meta-analyses suggest a cross-disorder shift toward a more linelike topology in
EEG/MEG delta and gamma band. We found no significant heterogeneity for these frequen-
cies. Trotzdem, due to the relatively small number of included studies that had small sample sizes,
heterogeneity analysis may not have been sufficiently powered.
Insufficient data were available for stratified analysis based on imaging modality, node and
connection definition, or study population. Trotzdem, this review and meta-analysis has
several implications for the methodological approach in network neuroscience and a trans-
diagnostic perspective on network alterations in brain disease. Since a straightforward, trans-
diagnostic interpretation was not possible for all disorder categories, only the disorders that
suggested a trend are discussed below.
Netzwerkneurowissenschaften
313
Minimum spanning tree in neuropsychiatric pathology: Systematic review
Network Size and Modality Effects
The effects of network size and other differences in methodology most likely hamper repro-
ducibility necessary for the development of biomarkers in the field of network neuroscience.
The comparison of MST metrics in control populations across studies showed that the MST leaf
fraction but not the diameter decreases with increasing network size. Außerdem, the range of
values for both measures was large (leaf fraction = 0.35–0.859 and diameter = 0.108–0.401),
further illustrating effects of methodological differences across studies. MST analysis was most
frequently applied to band-pass filtered neurophysiological recordings, but applications to
functional and structural MRI scans have also been reported. Some individual studies did show
substantially different metric values, suggesting that as expected, the specific processing pipe-
line before MST reconstruction impacts the estimated MST topology. Zusätzlich, for EEG and
MEG data, it is important which connectivity measure is used. Zum Beispiel, phase-based mea-
sures such as the phase-lag index might be noisier than amplitude-based measures such as the
amplitude envelop correlation, which could reduce the ability to extract consistent functional
connectivity and subsequently influence MST parameters (Colclough et al., 2016). zuletzt,
harmonization of node definitions may help to increase comparability between studies and
across modalities (Tewarie et al., 2015A).
l
D
Ö
w
N
Ö
A
D
e
D
F
R
Ö
M
H
T
T
P
:
/
/
D
ich
R
e
C
T
.
M
ich
T
.
/
/
T
e
D
u
N
e
N
A
R
T
ich
C
e
–
P
D
l
F
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
N
e
N
_
A
_
0
0
2
4
5
P
D
T
.
F
B
j
G
u
e
S
T
T
Ö
N
0
8
S
e
P
e
M
B
e
R
2
0
2
3
Neurodevelopment and Neurodegeneration
During maturation, MST topology shifts from a linelike, decentralized topology toward a more
starlike (integrated) Topologie, while the inverted age-relation is seen with aging in the fifth-sixth
Jahrzehnt (Boersma et al., 2013; Otte et al., 2015; Smit et al., 2016; van Dellen et al., 2018). Wir
found no systematic pattern of network disturbances in the scarce MST literature on neurode-
velopmental disorders. Studies on neurodegeneration suggest a shift toward a more linelike
MST topology as a general characteristic in neurophysiological recordings, which is in line
with broader findings in brain-aging literature (Fjell et al., 2014). MST characteristics in
neuroimaging modalities remain understudied. Studies in ALS and patients at risk for (AD)
dementia show inconsistent findings; the latter outcome of this review suggests that currently
used MST characteristics are not a promising predictive biomarker in at-risk groups for devel-
oping dementia.
Interessant, different subtypes of neurodegenerative disease seem to vary in damage to
MST organization. A pattern emerges of MST topology from starlike to linelike with HC <
bvFTD < AD < DLB < PDD. MST disturbances in neurophysiological recordings may thus
be a marker of disease progression and symptom severity in synucleinopathies.
Epilepsy and Attention Disorders
Preliminary evidence showing a loss of functional network integration in frequencies below
10 Hz is reported in association with focal epilepsy, which may be further provoked by sleep
deprivation. Interestingly, successful treatment seems to be associated with increased network
integration in this frequency range. Increased network integration in the upper alpha band is
found in focal childhood epilepsy. Increased network integration in frequencies above 10 Hz
may be associated with cognitive impairment in patients with childhood epilepsy.
In disorders characterized by disturbances in the cognitive domain of attention, a loss of
alpha-band network integration (lower leaf fraction, higher diameter) emerges as a recurrent
finding. It has been reported in delirium, in one study in ADHD (although the opposite finding
was also reported in ADHD), and DLB. Future studies may reveal if interventions that increase
alpha-band MST network integration and efficiency may be used to treat attention deficits,
Network Neuroscience
314
Minimum spanning tree in neuropsychiatric pathology: Systematic review
similar to early findings in epilepsy patients showing a network normalization after seizure
freedom. Of interest, van Lutterveld and others, found a higher MST maximum centrality in
the EEG alpha band in experienced meditators than in novice meditators, and leaf fraction
tended to be higher in this group (van Lutterveld et al., 2017). Applications of this and other
interventions in clinical populations are needed to test if sufficient alpha-band network inte-
gration is a prerequisite for attentional tasks, and if interventions aiming to improve these char-
acteristics specifically can be used in clinical care.
Limitations
Our approach to reviewing and interpreting MST studies’ findings across modalities has several
limitations. First, several studies did not report actual MST metric values or effect sizes, limiting
the interpretability of findings. The limited availability of quantitative data makes it premature
to conclude if there is no pattern of consistent network disturbances across disorders, or if data
on this topic is simply underpowered. Secondly, most studies solely mentioned significant
values, leading to a positive outcome bias. We suggest that reporting guidelines are needed
in network neuroscience that emphasize reporting the numeric values for network metrics as
completely as possible.
We did not include node-specific MST metrics in this review; we found no indication that
consistent findings were reported with this approach, but the macroscale MST metrics analysis
may only be less sensitive to disease-specific effects on brain networks. Other graph theoret-
ical approaches bring complementary clinical insights, including individual nodal, edge and
modular characteristics.
Finally, we aimed to gain transdiagnostic insights from different studies with variable meth-
odology in the definition of nodes and edges, imaging modalities, and frequency bands in
MEG/EEG recordings; possible confounds due to (differences in) processing pipelines are no
longer apparent in our modality-invariant summation of these studies.
We used a transdiagnostic approach to neuropsychiatric pathology. There is increasing
interest in overarching mechanisms that are a final common pathway to general factors of
psychopathology. These include the p factor for psychopathology, the characterization of
psychiatric disorders from a symptom-network perspective, and studies of general cognitive
dysfunction based on graph analysis (Borsboom, 2017; Caspi et al., 2014; Crossley et al.,
2014; Menon, 2020; van Bork et al., 2017). Another, complementary approach that may
advance the field is to look for convergence of evidence in isolating disease-specific effects
by comparing different network analysis approaches. Such within-disease approaches may for
example contribute to the development of staging or subtyping in specific pathological con-
ditions, and may help facilitate precision medicine approaches.
Conclusion
The MST approach has proven fruitful in capturing disease-related changes in brain network
topology. Harmonization of node definitions and especially network size remains a prereq-
uisite for comparing findings across studies and modalities. Empirical findings are more
consistent in neurological (in particular neurodegenerative) than psychiatric disorders and
neurodevelopmental disorders. They show that alterations in network topology are found
across disorders even after strict correction for network density effects. Most consistent (but
still preliminary) evidence was found for MST measures as markers of attention disorders,
particularly in epilepsy, and as markers of disease progression in neurodegenerative disease.
Importantly, contradicting findings within clinical populations were shown in previous
Network Neuroscience
315
l
D
o
w
n
o
a
d
e
d
f
r
o
m
h
t
t
p
:
/
/
d
i
r
e
c
t
.
m
i
t
.
t
/
/
e
d
u
n
e
n
a
r
t
i
c
e
-
p
d
l
f
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
n
e
n
_
a
_
0
0
2
4
5
p
d
t
.
f
b
y
g
u
e
s
t
t
o
n
0
8
S
e
p
e
m
b
e
r
2
0
2
3
Minimum spanning tree in neuropsychiatric pathology: Systematic review
reviews on conventional graph analysis, for example, in Alzheimer’s disease and epilepsy;
such contradictions were not found in the MST literature to date (Tijms et al., 2013; Van Diessen
et al., 2013). There is currently insufficient evidence for the use of MST metrics as sensitive and
specific biomarkers for neuropsychiatric disorders.
SUPPORTING INFORMATION
Supporting information for this article is available at https://doi.org/10.1162/netn_a_00245.
AUTHOR CONTRIBUTIONS
Edwin van Dellen: Conceptualization; Data curation; Investigation; Methodology; Supervision;
Visualization; Writing – original draft; Writing – review & editing. Bart de Rooy: Conceptualiza-
tion; Data curation; Methodology; Project administration; Resources; Visualization; Writing –
original draft; Writing – review & editing. Nicky Blomsma: Conceptualization; Data curation;
Formal analysis; Investigation; Methodology; Project administration; Visualization; Writing –
original draft; Writing – review & editing. Rick van der Spek: Conceptualization; Formal analysis;
Methodology; Software; Visualization; Writing – review & editing. Frank Gerritse: Conceptualization;
Data curation; Project administration; Writing – original draft. Prejaas Tewarie: Conceptualization;
Methodology; Writing – review & editing. Arjan Hillebrand: Conceptualization; Methodology;
Writing – review & editing. Wim Otte: Conceptualization; Methodology; Writing – review &
editing. Cornelis Jan Stam: Conceptualization; Methodology; Writing – review & editing.
FUNDING INFORMATION
Edwin van Dellen, ZonMw (https://dx.doi.org/10.13039/501100001826), Award ID: 60-
63600-98-711. Edwin van Dellen, UMC Utrecht Clinical Research Talent Fellowship.
REFERENCES
Anjomshoa, A., Dolatshahi, M., Amirkhani, F., Rahmani, F.,
Mirbagheri, M. M., & Aarabi, M. H. (2016). Structural brain net-
work analysis in schizophrenia using minimum spanning tree. In
38th annual international conference of the IEEE engineering in
medicine and biology society (EMBC) (pp. 4075–4078). Orlando,
FL: IEEE. https://doi.org/10.1109/EMBC.2016.7591622, PubMed:
28269178
Bassett, D. S., & Bullmore, E. T. (2009). Human brain networks in
health and disease. Current Opinion in Neurology, 22, 340–347.
https://doi.org/10.1097/wco.0b013e32832d93dd, PubMed:
19494774
Boersma, M., Smit, D. J. A., Boomsma, D. I., de Geus, E. J. C., delemarre-
Van de Waal, H. A., & Stam, C. J. (2013). Growing trees in child
brains: Graph theoretical analysis of electroencephalography-
derived minimum spanning tree in 5- and 7-year-old children
reflects brain maturation. Brain Connectivity, 3(1), 50–60. https://
doi.org/10.1089/brain.2012.0106, PubMed: 23106635
Borsboom, D. (2017). A network theory of mental disorders. World
Psychiatry, 16(1), 5–13. https://doi.org/10.1002/wps.20375,
PubMed: 28127906
Bullmore, E., & Sporns, O. (2009). Complex brain networks: Graph
theoretical analysis of structural and functional systems. Nature
Reviews Neuroscience, 10(3), 186–198. https://doi.org/10.1038
/nrn2575, PubMed: 19190637
Caspi, A., Houts, R. M., Belsky, D. W., Goldman-Mellor, S. J.,
Harrington, H., Israel, S., … Moffitt, T. E. (2014). The p factor: One
general psychopathology factor in the structure of psychiatric
disorders? Clinical Psychological Science, 2(2), 119. https://doi
.org/10.1177/2167702613497473, PubMed: 25360393
Colclough, G. L., Woolrich, M. W., Tewarie, P. K., Brookes, M. J.,
Quinn, A. J., & Smith, S. M. (2016). How reliable are MEG resting-
state connectivity metrics? NeuroImage, 138, 284–293. https://doi
.org/10.1016/j.neuroimage.2016.05.070, PubMed: 27262239
Crossley, N. A., Mechelli, A., Scott, J., Carletti, F., Fox, P. T.,
Mcguire, P., & Bullmore, E. T. (2014). The hubs of the human
connectome are generally implicated in the anatomy of brain
disorders. Brain, 137(8), 2382–2395. https://doi.org/10.1093
/brain/awu132, PubMed: 25057133
Das, S., & Puthankattil, S. D. (2020). Complex network analysis of
MCI-AD EEG signals under cognitive and resting state. Brain
Research, 1735. https://doi.org/10.1016/j.brainres.2020.146743,
PubMed: 32114060
De Gruttola, V. G., Clax, P., DeMets, D. L., Downing, G. J., Ellenberg,
S. S., Friedman, L., … Zeger, S. L. (2001). Considerations in the
evaluation of surrogate endpoints in clinical trials: Summary of a
National Institutes of Health workshop. Controlled Clinical Trials,
22(5), 485–502. https://doi.org/10.1016/S0197-2456(01)00153-2,
PubMed: 11578783
Network Neuroscience
316
l
D
o
w
n
o
a
d
e
d
f
r
o
m
h
t
t
p
:
/
/
d
i
r
e
c
t
.
m
i
t
.
t
/
/
e
d
u
n
e
n
a
r
t
i
c
e
-
p
d
l
f
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
n
e
n
_
a
_
0
0
2
4
5
p
d
t
.
f
b
y
g
u
e
s
t
t
o
n
0
8
S
e
p
e
m
b
e
r
2
0
2
3
Minimum spanning tree in neuropsychiatric pathology: Systematic review
De Vico Fallani, F., Latora, V., & Chavez, M. (2017). A topological
criterion for filtering information in complex brain networks.
PLoS Computational Biology, 13(1). https://doi.org/10.1371
/journal.pcbi.1005305, PubMed: 28076353
DeSalvo, M. N., Tanaka, N., Douw, L., Cole, A. J., & Stufflebeam,
S. M. (2020). Contralateral preoperative resting-state functional
MRI network integration is associated with surgical outcome in
temporal lobe epilepsy. Radiology, 294(2), 622–627. https://doi
.org/10.1148/radiol.2020191008, PubMed: 31961245
Douw, L., van Dellen, E., Gouw, A. A., Griffa, A., de Haan, W., van
den Heuvel, M., … Stam, C. J. (2019). The road ahead in clinical
network neuroscience. Network Neuroscience, 3, 969–993.
https://doi.org/10.1162/netn_a_00103, PubMed: 31637334
Ercsey-Ravasz, M., Markov, N. T., Lamy, C., VanEssen, D. C.,
Knoblauch, K., Toroczkai, Z., & Kennedy, H. (2013). A predictive
network model of cerebral cortical connectivity based on a
distance rule. Neuron, 80(1), 184–197. https://doi.org/10.1016/j
.neuron.2013.07.036, PubMed: 24094111
First, M. B., Drevets, W. C., Carter, C., Dickstein, D. P., Kasoff, L.,
Kim, K. L., … Zubieta, J.-K. (2018). Clinical applications of
neuroimaging in psychiatric disorders. The American Journal of
Psychiatry, 175(9), 915. https://doi.org/10.1176/appi.ajp.2018
.1750701, PubMed: 30173550
Fjell, A. M., McEvoy, L., Holland, D., Dale, A. M., & Walhovd, K. B.
(2014). What is normal in normal aging? Effects of aging, amyloid
and Alzheimer’s disease on the cerebral cortex and the hippo-
campus. Progress in Neurobiology, 117, 20–40. https://doi.org
/10.1016/j.pneurobio.2014.02.004, PubMed: 24548606
Fornito, A., Zalesky, A., & Breakspear, M. (2013). Graph analysis of
the human connectome: Promise, progress, and pitfalls. NeuroImage,
80, 426–444. https://doi.org/10.1016/j.neuroimage.2013.04.087,
PubMed: 23643999
Fornito, A., Zalesky, A., Pantelis, C., & Bullmore, E. T. (2012).
Schizophrenia, neuroimaging and connectomics. NeuroImage,
62, 2296–2314. https://doi.org/10.1016/j.neuroimage.2011.12
.090, PubMed: 22387165
Fraga González, G., Van der Molen, M. J. W., Žarić, G., Bonte, M.,
Tijms, J., Blomert, L., … Van der Molen, M. W. (2016). Graph
analysis of EEG resting state functional networks in dyslexic
readers. Clinical Neurophysiology, 127(9), 3165–3175. https://
doi.org/10.1016/j.clinph.2016.06.023, PubMed: 27476025
Fraga González, G., Smit, D. J. A., van der Molen, M. J. W., Tijms,
J., Stam, C. J., de Geus, E. J. C., & van der Molen, M. W. (2018).
EEG resting state functional connectivity in adult dyslexics using
phase lag index and graph analysis. Frontiers in Human Neuro-
science, 12, 341. https://doi.org/10.3389/fnhum.2018.00341,
PubMed: 30214403
Fraschini, M., Demuru, M., Hillebrand, A., Cuccu, L., Porcu, S., Di
Stefano, F., … Marrosu, F. (2016). EEG functional network
topology is associated with disability in patients with amyotro-
phic lateral sclerosis. Scientific Reports, 6. https://doi.org/10
.1038/srep38653, PubMed: 27924954
Fraschini, M., Demuru, M., Puligheddu, M., Floridia, S., Polizzi,
L., Maleci, A., … Marrosu, F. (2014). The re-organization of func-
tional brain networks in pharmaco-resistant epileptic patients
who respond to VNS. Neuroscience Letters, 580, 153–157.
https://doi.org/10.1016/j.neulet.2014.08.010, PubMed:
25123446
Jacini, F., Sorrentino, P., Lardone, A., Rucco, R., Baselice, F.,
Cavaliere, C., … Sorrentino, G. (2018). Amnestic mild cognitive
impairment is associated with frequency-specific brain network
alterations in temporal poles. Frontiers in Aging Neuroscience,
10, 400. https://doi.org/10.3389/fnagi.2018.00400, PubMed:
30574086
Janssen, T. W. P., Hillebrand, A., Gouw, A., Geladé, K., Van Mourik, R.,
Maras, A., & Oosterlaan, J. (2017). Neural network topology in
ADHD: Evidence for maturational delay and default-mode network
alterations. Clinical Neurophysiology, 128(11), 2258–2267. https://
doi.org/10.1016/j.clinph.2017.09.004, PubMed: 29028500
Jonak, K., Krukow, P., Jonak, K. E., Grochowski, C., & Karakuła-
Juchnowicz, H. (2019). Quantitative and qualitative comparison
of EEG-based neural network organization in two schizophrenia
groups differing in the duration of illness and disease burden:
Graph analysis with application of the minimum spanning tree.
Clinical EEG and Neuroscience, 50(4), 231–241. https://doi.org
/10.1177/1550059418807372, PubMed: 30322279
Jonak, K., Krukow, P., Karakuła-Juchnowicz, H., Rahnama-Hezavah,
M., Jonak, K. E., Stępniewski, A., … Grochowski, C. (2021). Aber-
rant structural network architecture in Leber’s hereditary optic
neuropathy. minimum spanning tree graph analysis application
into diffusion 7T MRI. Neuroscience, 455, 128–140. https://doi
.org/10.1016/J.neuroscience.2020.12.019, PubMed: 33359657
Kinney-Lang, E., Yoong, M., Hunter, M., Kamath Tallur, K., Shetty, J.,
McLellan, A., … Escudero, J. (2019). Analysis of EEG networks
and their correlation with cognitive impairment in preschool chil-
dren with epilepsy. Epilepsy and Behavior, 90, 45–56. https://doi
.org/10.1016/j.yebeh.2018.11.011, PubMed: 30513434
Krukow, P., Jonak, K., Karpiński, R., & Karakuła-Juchnowicz, H.
(2019). Abnormalities in hubs location and nodes centrality
predict cognitive slowing and increased performance variability
in first-episode schizophrenia patients. Scientific Reports, 9(1),
1–13. https://doi.org/10.1038/s41598-019-46111-0, PubMed:
31270391
Kruskal, J. B. (1956). On the shortest spanning subtree of a graph
and the traveling salesman problem. Proceedings of the Ameri-
can Mathematical Society, 7, 48–50. https://doi.org/10.1090
/S0002-9939-1956-0078686-7
Li, X., Jing, Z., Hu, B., Zhu, J., Zhong, N., Li, M., … Majoe, D.
(2017). A resting-state brain functional network study in MDD
based on minimum spanning tree analysis and the hierarchical
clustering. Complexity, 2017. https://doi.org/10.1155/2017
/9514369
López, M. E., Engels, M. M. A., van Straaten, E. C. W., Bajo, R., Delgado,
M. L., Scheltens, P., … Maestú, F. (2017). MEG beamformer-based
reconstructions of functional networks in mild cognitive impairment.
Frontiers in Aging Neuroscience, 9(APR), 107. https://doi.org/10
.3389/fnagi.2017.00107, PubMed: 28487647
Menon, V. (2020). Brain networks and cognitive impairment in psy-
chiatric disorders. World Psychiatry, 19(3), 309. https://doi.org/10
.1002/wps.20799, PubMed: 32931097
Nauta, I. M., Kulik, S. D., Breedt, L. C., Eijlers, A. J., Strijbis, E. M.,
Bertens, D., … Schoonheim, M. M. (2020). Functional brain
network organization measured with magnetoencephalography
predicts cognitive decline in multiple sclerosis. Multiple Sclerosis.
https://doi.org/10.1177/1352458520977160, PubMed:
33295249
Network Neuroscience
317
l
D
o
w
n
o
a
d
e
d
f
r
o
m
h
t
t
p
:
/
/
d
i
r
e
c
t
.
m
i
t
.
t
/
/
e
d
u
n
e
n
a
r
t
i
c
e
-
p
d
l
f
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
n
e
n
_
a
_
0
0
2
4
5
p
d
.
t
f
b
y
g
u
e
s
t
t
o
n
0
8
S
e
p
e
m
b
e
r
2
0
2
3
Minimum spanning tree in neuropsychiatric pathology: Systematic review
Nieboer, D., Sorrentino, P., Hillebrand, A., Heymans, M. W., Twisk,
J. W. R., Stam, C. J., & Douw, L. (2020). Brain network integration
in patients with migraine: A magnetoencephalography study.
Brain Connectivity, 10(5), 224–235. https://doi.org/10.1089
/brain.2019.0705, PubMed: 32397732
Numan, T., Slooter, A. J. C., van der Kooi, A. W., Hoekman, A. M. L.,
Suyker, W. J. L, Stam, C. J., & van Dellen, E. (2017). Functional
connectivity and network analysis during hypoactive delirium and
recovery from anesthesia. Clinical Neurophysiology, 128(6),
914–924. https://doi.org/10.1016/j.clinph.2017.02.022,
PubMed: 28402867
Olde Dubbelink, K. T. E., Hillebrand, A., Stoffers, D., Deijen, J. B.,
Twisk, J. W. R., Stam, C. J., & Berendse, H. W. (2014). Disrupted
brain network topology in Parkinson’s disease: A longitudinal
magnetoencephalography study. Brain, 137(1), 197–207.
https://doi.org/10.1093/brain/awt316, PubMed: 24271324
Otte, W. M., van Diessen, E., Paul, S., Ramaswamy, R., Subramanyam
Rallabandi, V. P., Stam, C. J., & Roy, P. K. (2015). Aging alterations in
whole-brain networks during adulthood mapped with the minimum
spanning tree indices: The interplay of density, connectivity cost
and life-time trajectory. NeuroImage, 109, 171–189. https://doi
.org/10.1016/j.neuroimage.2015.01.011, PubMed: 25585021
Peraza, L. R., Cromarty, R., Kobeleva, X., Firbank, M. J., Killen, A.,
Graziadio, S., … Taylor, J. P. (2018). Electroencephalographic
derived network differences in Lewy body dementia compared
to Alzheimer’s disease patients. Scientific Reports, 8(1). https://
doi.org/10.1038/s41598-018-22984-5, PubMed: 29545639
Požar, R., Giordani, B., & Kavcic, V. (2020). Effective differentiation
of mild cognitive impairment by functional brain graph analysis
and computerized testing. PLoS One, 15(3), e0230099. https://
doi.org/10.1371/journal.pone.0230099, PubMed: 32176709
Prim, R. C. (1957). Shortest connection networks and some gener-
alizations. Bell System Technical Journal, 36(6), 1389–1401.
https://doi.org/10.1002/j.1538-7305.1957.tb01515.x
Saba, V., Premi, E., Cristillo, V., Gazzina, S., Palluzzi, F., Zanetti,
O., … Grassi, M. (2019). Brain connectivity and information-flow
breakdown revealed by a minimum spanning tree-based analysis
of mri data in behavioral variant frontotemporal dementia. Fron-
tiers in Neuroscience, 13. https://doi.org/10.3389/fnins.2019
.00211, PubMed: 30930736
Siegel, M., Donner, T. H., & Engel, A. K. (2012). Spectral fingerprints of
large-scale neuronal interactions. Nature Reviews Neuroscience,
13, 121–134. https://doi.org/10.1038/nrn3137, PubMed: 22233726
Smit, D. J. A., De Geus, E. J. C., Boersma, M., Boomsma, D. I., &
Stam, C. J. (2016). Life-span development of brain network inte-
gration assessed with phase lag index connectivity and minimum
spanning tree graphs. Brain Connectivity, 6(4), 312–325. https://
doi.org/10.1089/brain.2015.0359, PubMed: 26885699
Sorrentino, P., Nieboer, D., Twisk, J. W. R., Stam, C. J., Douw, L., &
Hillebrand, A. (2017). The hierarchy of brain networks is related
to insulin growth factor-1 in a large, middle-aged, healthy cohort:
An exploratory magnetoencephalography study. Brain Connec-
tivity, 7(5), 321–330. https://doi.org/10.1089/brain.2016.0469,
PubMed: 28520468
Sorrentino, P., Rucco, R., Francesca, J., Trojsi, F., Anna, L., Baselice,
F., … Sorrentino, G. (2018). Brain functional networks become
more connected as amyotrophic lateral sclerosis progresses: A
source level magnetoencephalographic study. NeuroImage:
Clinical, 20, 564–571. https://doi.org/10.1016/j.nicl.2018.08
.001, PubMed: 30186760
Stam, C. J., Tewarie, P., Van Dellen, E., van Straaten, E. C. W.,
Hillebrand, A., & Van Mieghem, P. (2014). The trees and the
forest: Characterization of complex brain networks with mini-
mum spanning trees. International Journal of Psychophysiology,
92, 129–138. https://doi.org/10.1016/j.ijpsycho.2014.04.001,
PubMed: 24726900
Stam, C. J., & van Straaten, E. C. W. (2012). The organization of
physiological brain networks. Clinical Neurophysiology, 123,
1067–1087. https://doi.org/10.1016/j.clinph.2012.01.011,
PubMed: 22356937
Tewarie, P., Hillebrand, A., Schoonheim, M. M., van Dijk, B. W.,
Geurts, J. J. G., Barkhof, F., … Stam, C. J. (2014a). Functional
brain network analysis using minimum spanning trees in multiple
sclerosis: An MEG source-space study. NeuroImage, 88, 308–318.
https://doi.org/10.1016/j.neuroimage.2013.10.022, PubMed:
24161625
Tewarie, P., Schoonheim, M. M., Schouten, D. I., Polman, C. H.,
Balk, L. J., Uitdehaag, B. M. J., … Stam, C. J. (2015a). Functional
brain networks: Linking thalamic atrophy to clinical disability in
multiple sclerosis, a multimodal fMRI and MEG Study. Human
Brain Mapping, 36(2), 603–618. https://doi.org/10.1002/ hbm
.22650, PubMed: 25293505
Tewarie, P., Steenwijk, M. D., Tijms, B. M., Daams, M., Balk, L. J.,
Stam, C. J., … Hillebrand, A. (2014b). Disruption of structural and
functional networks in long-standing multiple sclerosis. Human
Brain Mapping, 35(12), 5946–5961. https://doi.org/10.1002
/hbm.22596, PubMed: 25053254
Tewarie, P., van Dellen, E., Hillebrand, A., & Stam, C. J. (2015b).
The minimum spanning tree: An unbiased method for brain net-
work analysis. NeuroImage, 104, 177–188. https://doi.org/10
.1016/j.neuroimage.2014.10.015, PubMed: 25451472
Tijms, B. M., Wink, A. M., de Haan, W., van der Flier, W. M., Stam,
C. J., Scheltens, P., & Barkhof, F. (2013). Alzheimer’s disease:
Connecting findings from graph theoretical studies of brain net-
works. Neurobiology of Aging, 34, 2023–2036. https://doi.org/10
.1016/j.neurobiolaging.2013.02.020, PubMed: 23541878
Utianski, R. L., Caviness, J. N., van Straaten, E. C. W., Beach, T. G.,
Dugger, B. N., Shill, H. A., … Hentz, J. G. (2016). Graph theory
network function in Parkinson’s disease assessed with electroen-
cephalography. Clinical Neurophysiology, 127(5), 2228–2236.
https://doi.org/10.1016/j.clinph.2016.02.017, PubMed:
27072094
van Bork, R., Epskamp, S., Rhemtulla, M., Borsboom, D., & van der
Maas, H. L. J. (2017). What is the p-factor of psychopathology?
Some risks of general factor modeling. Theory and Psychology,
27(6), 759–773. https://doi.org/10.1177/0959354317737185
van Dellen, E., Bohlken, M. M., Draaisma, L., Tewarie, P. K., Van
Lutterveld, R., Mandl, R., … Sommer, I. E. (2016). Structural brain
network disturbances in the psychosis spectrum. Schizophrenia
Bulletin, 42(3), 782–789. https://doi.org/10.1093/schbul
/sbv178, PubMed: 26644605
van Dellen, E., Börner, C., Schutte, M., van Montfort, S., Abramovic,
L., Boks, M. P., … Sommer, I. (2020). Functional brain networks in
the schizophrenia spectrum and bipolar disorder with psychosis.
Schizophrenia, 6(1), 1–9. https://doi.org/10.1038/s41537-020
-00111-6, PubMed: 32879316
Network Neuroscience
318
l
D
o
w
n
o
a
d
e
d
f
r
o
m
h
t
t
p
:
/
/
d
i
r
e
c
t
.
m
i
t
.
t
/
/
e
d
u
n
e
n
a
r
t
i
c
e
-
p
d
l
f
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
n
e
n
_
a
_
0
0
2
4
5
p
d
.
t
f
b
y
g
u
e
s
t
t
o
n
0
8
S
e
p
e
m
b
e
r
2
0
2
3
Minimum spanning tree in neuropsychiatric pathology: Systematic review
van Dellen, E., de Waal, H., van der Flier, W. M., Lemstra, A. W.,
Slooter, A. J. C., Smits, L. L., … Scheltens, P. (2015). Loss of EEG
network efficiency is related to cognitive impairment in dementia
with lewy bodies. Movement Disorders, 30(13), 1785–1793.
https://doi.org/10.1002/mds.26309, PubMed: 26179663
Van Dellen, E., De Witt Hamer, P. C., Douw, L., Klein, M., Heimans,
J. J., Stam, C. J., … Hillebrand, A. (2013). Connectivity in MEG
resting-state networks increases after resective surgery for
low-grade glioma and correlates with improved cognitive perfor-
mance. NeuroImage: Clinical, 2(1), 1–7. https://doi.org/10.1016/j
.nicl.2012.10.007, PubMed: 24179752
van Dellen, E., Sommer, I. E., Bohlken, M. M., Tewarie, P.,
Draaisma, L., Zalesky, A., … Stam, C. J. (2018). Minimum span-
ning tree analysis of the human connectome. Human Brain Map-
ping, 39(6), 2455–2471. https://doi.org/10.1002/ hbm.24014,
PubMed: 29468769
van den Heuvel, M. P., de Lange, S. C., Zalesky, A., Seguin, C., Yeo,
B. T. T., & Schmidt, R. (2017). Proportional thresholding in
resting-state fMRI functional connectivity networks and conse-
quences for patient-control connectome studies: Issues and rec-
ommendations. NeuroImage, 152, 437–449. https://doi.org/10
.1016/j.neuroimage.2017.02.005, PubMed: 28167349
Van Diessen, E., Diederen, S. J. H., Braun, K. P. J., Jansen, F. E., &
Stam, C. J. (2013). Functional and structural brain networks in
epilepsy: What have we learned? Epilepsia, 54, 1855–1865.
https://doi.org/10.1111/epi.12350, PubMed: 24032627
van Diessen, E., Otte, W. M., Braun, K. P. J., Stam, C. J., & Jansen,
F. E. (2014). Does sleep deprivation alter functional EEG net-
works in children with focal epilepsy? Frontiers in Systems Neu-
roscience, 8, 67. https://doi.org/10.3389/fnsys.2014.00067,
PubMed: 24808832
van Diessen, E., Otte, W. M., Stam, C. J., Braun, K. P. J., & Jansen,
F. E. (2016). Electroencephalography based functional networks
in newly diagnosed childhood epilepsies. Clinical Neurophysiol-
ogy, 127(6), 2325–2332. https://doi.org/10.1016/j.clinph.2016
.03.015, PubMed: 27178845
van Lutterveld, R., van Dellen, E., Pal, P., Yang, H., Stam, C. J., &
Brewer, J. (2017). Meditation is associated with increased brain
network integration. NeuroImage, 158, 18–25. https://doi.org/10
.1016/j.neuroimage.2017.06.071, PubMed: 28663069
Van Mieghem, P., & Van Langen, S. (2005). Influence of the link weight
structure on the shortest path. Physical Review E, 71(5). https://doi
.org/10.1103/PhysRevE.71.056113, PubMed: 16089608
van Montfort, S. J. T., van Dellen, E., van den Bosch, A. M. R., Otte,
W. M., Schutte, M. J. L., Choi, S. H., … Kim, J. J. (2018). Resting-
state fMRI reveals network disintegration during delirium. Neuro-
Image: Clinical, 20, 35–41. https://doi.org/10.1016/j.nicl.2018
.06.024, PubMed: 29998059
van Nieuwenhuizen, D., Douw, L., Klein, M., Peerdeman, S. M.,
Heimans, J. J., Reijneveld, J. C., … Hillebrand, A. (2018).
Cognitive functioning and functional brain networks in postoper-
ative WHO grade I meningioma patients. Journal of Neuro-
Oncology, 140(3), 605–613. https://doi.org/10.1007/s11060
-018-2987-1, PubMed: 30219943
van Wijk, B. C. M., Stam, C. J., & Daffertshofer, A. (2010). Compar-
ing brain networks of different size and connectivity density
using graph theory. PLoS One, 5(10). https://doi.org/10.1371
/journal.pone.0013701, PubMed: 21060892
Váša, F., Bullmore, E. T., & Patel, A. X. (2018). Probabilistic thresh-
olding of functional connectomes: Application to schizophrenia.
NeuroImage, 172, 326–340. https://doi.org/10.1016/j
.neuroimage.2017.12.043, PubMed: 29277403
Wang, B., Miao, L., Niu, Y., Cao, R., Li, D., Yan, P., … Xiang, J.
(2018). Abnormal functional brain networks in mild cognitive
impairment and Alzheimer’s disease: A minimum spanning tree
analysis. Journal of Alzheimer’s Disease, 65(4), 1093–1107.
https://doi.org/10.3233/JAD-180603, PubMed: 30149457
Wang, Y., Tao, F., Zuo, C., Kanji, M., Hu, M., & Wang, D. (2019).
Disrupted resting frontal–parietal attention network topology
is associated with a clinical measure in children with attention-
deficit/hyperactivity disorder. Frontiers in Psychiatry, 10. https://
doi.org/10.3389/fpsyt.2019.00300, PubMed: 31156474
Wang, Y., Zuo, C., Xu, Q., Liao, S., Kanji, M., & Wang, D. (2020).
Altered resting functional network topology assessed using graph
theory in youth with attention-deficit/hyperactivity disorder. Prog-
ress in Neuro-Psychopharmacology and Biological Psychiatry, 98.
https://doi.org/10.1016/j.pnpbp.2019.109796, PubMed: 31676467
Xue, H., Wang, Z., Tan, Y., Yang, H., Fu, W., Xue, L., & Zhao, J.
(2020). Resting-state EEG reveals global network deficiency in
dyslexic children. Neuropsychologia, 138. https://doi.org/10
.1016/j.neuropsychologia.2020.107343, PubMed: 31952981
Yu, M., Gouw, A. A., Hillebrand, A., Tijms, B. M., Stam, C. J., van
Straaten, E. C. W., & Pijnenburg, Y. A. L. (2016). Different func-
tional connectivity and network topology in behavioral variant of
frontotemporal dementia and Alzheimer’s disease: An EEG study.
Neurobiology of Aging, 42, 150–162. https://doi.org/10.1016/j
.neurobiolaging.2016.03.018, PubMed: 27143432
Zalesky, A., Fornito, A., Cocchi, L., Gollo, L. L., van den Heuvel, M. P.,
& Breakspear, M. (2016). Connectome sensitivity or specificity:
Which is more important? NeuroImage, 142, 407–420. https://doi
.org/10.1016/j.neuroimage.2016.06.035, PubMed: 27364472
Zare, M., Rezvani, Z., & Benasich, A. A. (2016). Automatic classi-
fication of 6-month-old infants at familial risk for language-based
learning disorder using a support vector machine. Clinical Neu-
rophysiology, 127(7), 2695–2703. https://doi.org/10.1016/j
.clinph.2016.03.025, PubMed: 27156833
Zeng, K., Kang, J., Ouyang, G., Li, J., Han, J., Wang, Y., … Li, X.
(2017). Disrupted brain network in children with autism spec-
trum disorder. Scientific Reports, 7(1), 1–12. https://doi.org/10
.1038/s41598-017-16440-z, PubMed: 29176705
Network Neuroscience
319
l
D
o
w
n
o
a
d
e
d
f
r
o
m
h
t
t
p
:
/
/
d
i
r
e
c
t
.
m
i
t
.
/
t
/
e
d
u
n
e
n
a
r
t
i
c
e
-
p
d
l
f
/
/
/
/
/
6
2
3
0
1
2
0
2
8
0
9
3
n
e
n
_
a
_
0
0
2
4
5
p
d
.
t
f
b
y
g
u
e
s
t
t
o
n
0
8
S
e
p
e
m
b
e
r
2
0
2
3
