You are on page 1of 11

Brain Imaging and Behavior

https://doi.org/10.1007/s11682-021-00457-0

ORIGINAL RESEARCH

Multivariate machine learning‐based language mapping in glioma


patients based on lesion topography
Nan Zhang 1,2 & Binke Yuan 3,4,5 & Jing Yan 6 & Jingliang Cheng 6 & Junfeng Lu 2 & Jinsong Wu 2,7

Received: 15 June 2019 / Revised: 11 December 2020 / Accepted: 21 January 2021


# The Author(s), under exclusive licence to Springer Science+Business Media, LLC part of Springer Nature 2021

Abstract
Diffusive and progressive tumor infiltration within language-related areas of the brain induces functional reorganization.
However, the macrostructural basis of subsequent language deficits is less clear. To address this issue, lesion topography data
from 137 preoperative patients with left cerebral language-network gliomas (81 low-grade gliomas and 56 high-grade gliomas),
were adopted for multivariate machine-learning-based lesion-language mapping analysis. We found that tumor location in the left
posterior middle temporal gyrus—a bottleneck where both dorsal and ventral language pathways travel—predicted deficits of
spontaneous speech (cluster size = 1356 mm3, false discovery rate corrected P < 0.05) and naming scores (cluster size = 1491
mm3, false discovery rate corrected P < 0.05) in the high-grade glioma group. In contrast, no significant lesion-language mapping
results were observed in the low-grade glioma group, suggesting a large functional reorganization. These findings suggest that in
patients with gliomas, the macrostructural plasticity mechanisms that modulate brain-behavior relationships depend on glioma
grade.

Keywords Low‐grade glioma . High‐grade glioma . Structural MRI . Language . Machine learning

Introduction

Neuroplasticity is an intrinsic adaptive mechanism for remod-


Nan Zhang, Binke Yuan and Jing Yan contributed equally to this work. eling functional systems after cerebral lesion (Payne and
Lomber 2001). When functional reorganization is sufficient,
* Junfeng Lu cognitive function is preserved. However when insufficient,
junfeng_lu@fudan.edu.cn especially in patients with lesions involving eloquent brain
areas, the results can be multiple cognitive deficits (Corbetta
1
Department of Neurosurgery, The First Affiliated Hospital of USTC, et al. 2015; Warren et al. 2014). Unlike acute cerebral lesion
Division of Life Sciences and Medicine, University of Science and with subsequent postlesional plasticity—such as occurs after
Technology of China, Anhui Hefei, China
traumatic brain injury or stroke—the growth of diffuse and
2
Glioma Surgery Division, Neurologic Surgery Department, Huashan progressive glioma is often accompanied with a good deal of
Hospital, Fudan University, Shanghai, China
functional reorganization (Almairac et al. 2015; Desmurget
3
Institute for Brain Research and Rehabilitation, South China Normal et al. 2007; Herbet et al. 2014, 2016). Exploring the pattern
University, Guangzhou, China
of functional reorganization in the brain following tumor
4
Key Laboratory of Brain, Cognition and Education Sciences (South growth not only broadens our understanding of the brain’s
China Normal University), Ministry of Education,
capacity to reorganize itself, it also provides essential infor-
Guangzhou, China
5
mation for functional surgical neuro-oncology (Duffau 2005;
Center for Language and Brain, Shenzhen Institute of Neuroscience,
Duffau et al. 2014; Weng et al. 2018; Wu et al. 2015; Yuan
Shenzhen, China
6
et al. 2020; Zhang et al. 2018). The majority of current imag-
Department of MRI , The First Affiliated Hospital of Zhengzhou
ing studies focus on the functional alterations, such as spon-
University , Zhengzhou, China
7
taneous or task-induced activity changes (Benzagmout et al.
Institute of Brain-Intelligence Technology , Zhangjiang Lab,
2007; Deverdun et al. 2019; Duffau 2005; Hart et al. 2019;
Shanghai, China
Brain Imaging and Behavior

Huang et al. 2018; Lu et al. 2017; Sarubbo et al. 2012; Yuan mapping to identify whether tumor location is critical for lan-
et al. 2019; Zhang et al. 2018), cognitive-related functional guage impairment in each group. Considering the potential
network reorganization (Fox and King 2018; Ghinda et al. mislocalization in mass-univariate voxel-based lesion-
2018; Yuan et al. 2020; Zhang et al. 2016), or changes in symptom mapping (VLSM) (Bates et al. 2003), we adopted
functional network topological properties (e.g., the small- a multivariate machine-learning lesion-symptom mapping
worldness) (Aerts et al. 2016). However, the structural technique (Mah et al. 2014; Rorden et al. 2007; Zhang et al.
neuroplasticity remains largely unknown. 2014). We hypothesized that the lesion topography for lan-
Several studies investigated the structural changes of the guage impairments in the LGG group would have limited
brain flowing tumor growth. Almairac and colleagues showed predictive ability because of the large functional reorganiza-
that diffuse low-grade glioma (LGG) within the insula— tion that occurs with LGGs. In contrast, we suspected that
which infiltrate slowly (generally taking years)—induced a damage to the eloquent areas in the HGG group would be
marked increase in the gray matter volume of the associated with language deficits.
contralesional homotopic insula (Almairac et al. 2018). The
structural recruitment of the contralesional homotopic areas is
consistent with other similar findings, including compensato- Materials and methods
ry redistribution of activity in contralesional homotopic areas
in cases of left dominant insular LGG (Duffau et al. 2001) and
left dominant Broca’s or Wernicke’s LGG (Duffau 2012; Participants
Duffau et al. 2014; Picart et al. 2018). Patients with LGGs
that include supratentorial lesions in the language network Patients were prospectively screened and data was collected at
can also exhibit overlapping changes in gray matter volume two centers: (1) Huashan Hospital (from 2011 to 2016) and (2)
and functional activity in the cerebellum (Zhang et al. 2018). the First Affiliated Hospital of Zhengzhou University (from
These evidences suggest structure/function coupled alter- 2014 to 2018). We pooled the data from the two centers be-
ations in the brains of glioma. cause only lesion masks were used .
A more direct way to investigate the structural Inclusion criteria: (1) Pathologically confirmed glioma
neuroplasticity following glioma growth was to assess the (based on the 2007 WHO classification system) in the left
lesion-symptom relationship, i.e., whether or not the tumor cerebral cortex (Louis et al. 2007); (2) No chemo and/or radi-
location determines the severity and domain of cognitive def- ation treatment history; (3) location of the glioma partially
icits (Banerjee et al. 2015; Deverdun et al. 2019; Kinoshita overlapped or was within language network areas proposed
et al. 2016; Mah et al. 2014; Pisoni et al. 2018, 2019; Sagberg by Fedorenko et al. (2010) ; (4) age was 18–75 years old; (5)
et al. 2019; Wang et al. 2015; Zhang et al. 2014). The neural right-handed, as confirmed by the Edinburgh Handedness
bases of diverse functions have been investigated. However, Inventory; (6) no symptoms of motor impairment, as evi-
the studies focusing on language function are still relatively denced by a grade V on the Medical Research Council
scarce. In these limited number of investigations focusing on (MRC) Scale for Muscle Strength (Paternostro-Sluga et al.
language function, limitations have to be addressed: First, the 2008); (7) Chinese Han nationality; (8) no history of brain
sample size in each lesion location is small, which is critical operation; (9) more than nine years of education; (10) no mid-
for statistical power (Sperber and Karnath 2017); Second, ma- line shift in the structural images, confirmed by the in situ
jority of existent studies focus on patients with LGGs location of midline brain structures (corpus callosum, septa
(Deverdun et al. 2019), and little attention has been paid to pellucidum, third ventricle, hypothalamus, and pineal region);
patients with high-grade glioma (HGG). Tumor grade has (11) both structural and functional images covered the whole
been shown to be a leading predictor of neuroplasticity and brain, especially the whole cerebellum; (12) good cooperation
cognitive outcomes (Yuan et al. 2019; Zhang et al. 2018). during the linguistic/cognitive evaluation; (13) no history of
However, tumor grade-related differences in lesion-language other major neurological or psychiatric disorders; and (14) no
relationship remain largely unknown. alcohol or drug abuse.
In this work, we enrolled a large cohort of patients (81 In total, 137 patients met all inclusion criteria and complet-
LGG and 56 HGG) with left-hemispheric gliomas within lan- ed the entire preoperative clinical, language, and imaging data
guage network areas, and investigated lesion-language rela- acquisition. The patients were further grouped based on the
tionship. All patients were scanned with high-resolution 3D- malignancy of brain tumors: 81 patients with grade I (n = 2)
T1 images, were confirmed not to have any midline shift, and and II (n = 79) brain tumors comprised as LGG group and 56
underwent detailed assessment of language function. patients with grade III (n = 28) and IV (n = 28) brain tumors
Sufficient numbers of the lesions in the left hemisphere and comprised the HGG group. For the two cases of grade I in
the different levels of language impairments within LGG and LGG group, one patient was diagnosed as ganglio glioma and
HGG groups ensured that we could perform lesion-language the other patient was diagnosed as neuroepithelial glioma.
Brain Imaging and Behavior

Clinical and neuropsychological examination DARTEL and lesion mask normalization Each 3D-T1 volume
without tumor area was segmented into gray matter (GM), white
All patients underwent comprehensive assessment of Karnofsky matter (WM), and cerebrospinal fluid (CSF), bone, soft tissue
performance status (KPS), language, motor, and Mini-mental and air/background using SPM12 (Wellcome Centre for
State Examination (MMSE). Language function was assessed Human Neuroimaging, https://www.fil.ion.ucl.ac.uk/spm). We
by the Aphasia Battery for Chinese speakers (ABC), which is then used DARTEL (Diffeomorphic Anatomical Registration
the Chinese standardized adaptation of the Western Aphasia using Exponentiated Lie algebra) for registration,
Battery (Gao et al. 1992; Lu et al. 2013; Wu et al. 2015; Yang normalization, and modulation (Ashburner 2007). It has been
et al. 2016; Zhang et al. 2018), and includes spontaneous speech demonstrated that DARTEL achieves a better normalization both
(SSS ; range, 0–20), comprehension (SCom; range, 0–230), repe- in healthy volunteers (Klein et al. 2009) and in patients with focal
tition (SRep ; range, 0–100), and naming (SNam ; range, 0–100) lesion (Ripolles et al. 2012) than other methods. After segmen-
scores. The Aphasia Quotient (AQ) score (range, 0–100) can be tation, a customized template was generated using the average
calculated from these items to reflect the global severity: tissue probability maps across all patients, and each patient’s

AQ ¼ SSS þSCom  23þSRep  10þSNam  10  2: segmented maps were warped into the template. This procedure
was repeated until a best study-specific template (Template_6)
was generated. A high quality DARTEL template was created,
Image acquisition which was free of contrast-enhancing effect (Supplementary
Figure S1). The images were then modulated by the Jacobian
All images were obtained using Siemens Magnetom Verio determinants to ensure conservation of regional differences in the
3.0 T MRI scanners (Siemens Medical Solutions, Erlangen, absolute amounts of gray matter (GM). Registered images were
Germany) at the two centers. For LGGs, high resolution T1- then transformed to Montreal Neurological Institute (MNI)
weighted and T2-weighted fluid-attenuated inversion recov- space. For each patient, the native tumor mask was spatially
ery (T2-FLAIR) images were acquired with the following normalized to standard MNI space by applying the deformation
parameters: T1-weighted images: axial magnetization- field estimated in segmentation. All T1 images and tumor masks
prepared rapid gradient echo (MPRAGE) sequence, repetition were well normalized judged by visual inspection. All LGG (or
time (TR) = 1,900 ms; echo time (TE) = 2.93 ms; inversion HGG) tumor masks in MNI space were then stacked together
time (TI) = 900 ms; flip angle (FA) = 90°; field of view and binarized to construct a tumor overlapping image (Fig. 1a),
(FOV) = 250 × 219 mm; matrix size = 256 × 215; slice thick- in which each voxel was identified as part of the tumor region
ness = 1 mm; voxel size = 1 × 1 × 1 mm3; slice number = 176; from at least one patient.
and scanning time = 7 min 47 s. T2-FLAIR: TR = 9,000 ms;
TE = 99 ms; TI = 2500 ms; FA = 150°; FOV = 240 × 214 mm;
matrix size = 256 × 160; slice thickness = 2 mm; voxel size =
0.9 × 1.3 × 2.0 mm3; slice number = 66; and scanning time = Multivariate lesion‐language mappings
7 min 30 s. For patients with HGGs in Huashan Hospital, only
T1-weighted sequences with contrasts (gadopentetate Multivariate lesion-language using SVR-LSM Due to the in-
dimeglumine) were acquired with the same parameters. For herent limitations of univariate voxel-based lesion-behav-
patients with HGGs in the First Affiliated Hospital of ior mapping (VLSM) analysis (Bates et al. 2003; Mah
Zhengzhou University, both pre- and post-contrast T1-weight- et al. 2014), here we adopted multivariate machine
ed images were acquired. learning-based lesion-symptom mapping, SVR-LSM
(Zhang et al. 2014). Rather than each isolated voxel,
Structural data processing SVR-LSM models the symptom relation to the entire le-
sion map using a nonlinear function, which intrinsically
Lesion mapping For each patient, manual tumor drawing on considers the intervoxel correlations. SVR-LSM has been
the 3D-T1 images was performed based on the contrast- shown to produce much higher sensitivity and specificity
enhancing tumor areas or the FLAIR hyperintense areas (ne- for detecting lesion-behavior relationships than can be
crotic areas were included but not peritumoral edema). For achieved via the univariate method (Mah et al. 2014;
3D-T1 images without glioma enhancement, T2- FLAIR im- Yourganov et al. 2016; Zhang et al. 2014). SVR has been
ages were first coregistered to the 3D-T1 images as a visual widely used in imaging-based prediction analysis (Cui
reference. After manually tracing the tumor, we created a 3D- and Gong 2018; Dosenbach et al. 2010; Yuan et al.
T1 volume without tumor area (set to 0) for each patient. This 2017; Zhang et al. 2014). For a linear model, SVR can
procedure was achieved manually using RANO criteria as a be described as
reference (Wen et al. 2010). The accuracy of the manual trac-
ing was further confirmed by a senior neurosurgeon. Y ¼ !T ðXÞ þ b ð1Þ
Brain Imaging and Behavior

Fig. 1 Lesion distributions and language performance of the 81 patients patients with a lesion at a specific voxel. b The distributions of language
with LGGs and the 56 patients with HGGs. a Lesion-frequency overlaps scores. SS: spontaneous speech; Com: comprehension. L, left; R, right
of tumor location in MNI space. The color bar represents the number of

where X is the lesion patterns across patients, Y is a Winkler et al. 2014), only voxels lesioned in ≥ 10 patients
network attribute, ð X Þ is the function transforming the were included in the analysis. Each patient’s lesion-data
lesion patterns to higher dimensional feature space, ω = vector was normalized to have a unit norm that served as
(ω0, ω1, ω2, …) is the fitting coefficient (weight) in high a direct total lesion volume control (hereafter SVR-
dimensional space, and b is the fitting error. LSM + dTLVC). For each language measure, a weight
map (the  -map) assessing the voxel-wise prediction
SVR-LSM uses libSVM (Chang and Lin 2011) (http:// contribution was created by back-projection. One thou-
www.csie.ntu.edu.tw/~cjlin/libsvm/) with epsilon-SVR sand permutation tests were performed to create a proba-
and a non-linear radial basis function (RBF) kernel. We bility map that inferred the regional lesion-behavior rela-
set gamma as 2 and cost (C) as 30 (i.e., the default pa- tionship. Significant results were those clusters of voxels
rameters in the SVR-LSM). To ensure sufficient statistical with uncorrected probability P < 0.001 and a false discov-
power (Kimberg et al. 2007; Sperber and Karnath 2017; ery rate (FDR)-corrected P < 0.05.
Brain Imaging and Behavior

Deterministic tractography Table 1 Demographic information and language scores for the LGG
and HGG groups

For each brain region showed significant prediction in SVR- LGG (n=81) HGG (n=56) χ2 or t P value
LSM analysis, deterministic tractography was performed to ex-
plore the white matter tracks passing through it. The fiber track- Age 38.86±9.57 45.11±14.72 -3 0.003
ing utilized a publicly available group-averaged tractography Sex (M/F) 47/34 34/22 0.31 a 0.76 a
atlas, i.e., HCP-1065 atlas (WU-Minn HCP Consortium; - Education 10.79±4.05 10.93±4.43 -0.19 0.85
https://pitt.box.com/shared/static/ Tumor volume (cm3) 82.68±49.27 72.55±48.31 1.07 0.28
wzozvjxxudvyytnzm84ux5h3bx7d8kf9.gz). The HCP-1065 at- WHO grade I/II = 2/79 III/IV = 28/28 - -
las was constructed from 1065 subjects’ diffusion MRI data AQ 95.21±4.86 89.68±10.18 4.25 <0.001
from the Human Connectome Project (2017 Q4, 1200-subject Spontaneous speech 18.86±1.46 17.68±2.65 3.36 0.001
release; multi-shell diffusion scheme, b-values: 1000, 2000, Comprehension 222.25±13.05 210.30±25.91 3.56 <0.001
3000s/mm2; diffusion sampling directions: 90, 90, and 90; in- Repetition 96.72±6.63 93±11.25 2.42 0.017
plane resolution: 1.25 mm). The diffusion data were recon- Naming 94.02±5.51 85.46±13.85 5.03 <0.001
structed in the MNI space using q-space diffeomorphic recon- a
Pearson’s chi-square test
struction (Yeh and Tseng 2011) to obtain the spin distribution
function (Yeh et al. 2010). The analysis was conducted using
diagnosis of aphasia (Fig. 1b). Overall, patients with HGGs
DSI Studio (http://dsi-studio.labsolver.org). For reproducibility
had significantly lower AQ scores (two sample t-test, t
problem in tractography, a combination of fiber tracking
(135) = 4.25, P = 4 × 10− 5), spontaneous speech scores (two
parameters within a working range (anisotropy threshold: 0.5–
sample t-test, t (135) = 3.36, P = 1 × 10− 3), comprehension
0.7; angular threshold: 15–90°; step size: 0.5–1.5 voxel
scores (two sample t-test, t (135) = 3.56, P = 5.3 × 10− 4), rep-
distance) was used. The identified tracks were automatically
etition scores (two sample t-test, t (135) = 2.43, P = 0.017) and
labeled (Yeh 2020). Moreover, automatic fiber tracking was
naming scores (two sample t-test, t (135) = 5.03, P = 1.5 × 10−
also performed when placing no target region. We then quan- 6
) than those with LGGs. The frequency distributions of the
tified the ratio of tracks passing through the target region and
five ABC scores are shown in Fig. 1b.
the total number of tracks of a specific white matter pathway.
To determine whether the language differences between
LGG and HGG groups were due to the age difference or any
The effects of covariates on the lesion‐language
collection site effect, we performed two-sample t-test for each
mappings
language test and added age, sex, education, tumor volume,
and collection site as covariates. After controlling for these
To investigate the reproducibility of the main findings, we
potentially confounding factors, we found that the AQ (t
regressed out patients’ age, sex, education and the imaging
(130) = 3.50, P = 6.26 × 10− 4), spontaneous speech (two sam-
collection site from each language test and reperformed the
ple t-test, t (130) = 2.43, P = 0.016), comprehension (t (130) =
SVR-LSM + dTLVC analyses.
3.19, P = 0.002), and naming (t (130) = 4.27, P = 3.67 × 10− 5)
scores were still significantly lower in the HGG group.
Results
Lesion anatomy
Language deficits in patients with LGGs and HGGs
Tumor locations for both groups were primarily distributed in
We found no significant differences in sex, education, or tu- the left frontal, temporal, insular, and subcortical regions
mor volume between the LGG and HGG groups (Table 1). (Fig. 1a). LGGs were most often in the left insula (N = 32),
The HGG group was significantly older than the LGG group while HGGs were most often in the left superior temporal
(two sample t-test, t (135) = -3, P = 0.003). We also found that gyrus (STG, N = 21).
age showed statistically significant correlations with language
scores for both the LGG and HGG groups (Supplementary Results of lesion‐language mapping by SVR-LSM +
Table S1). Education showed statistically significant correla- dTLVC
tions with language scores in the HGG group. Additionally,
we found no significant correlation between tumor volume No significant results were observed for the LGG group. In
and language scores in either LGG or HGG group (rs < 0.15). contrast, HGG tumor damage (Fig. 2a) in the left posterior
Regarding the five ABC scores, about 21 % (17/81) of middle temporal gyrus (pMTG) was associated with deficits
those with LGGs and 50 % (28/56) of those with HGGs had in spontaneous speech (cluster size = 1356 mm 3, FDR-
AQ scores less than 93.8—which is the clinical criterion for a corrected, P < 0.05) and naming (cluster size = 1491 mm3,
Brain Imaging and Behavior

Fig. 2 Multivariate lesion-language results for the HGG group. a Tumor b The left posterior middle temporal gyrus is located in the white matter
damage in the left posterior middle temporal gyrus significantly (FDR- confluence of the arcuate fasciculus (AF), inferior fronto-occipital fascic-
corrected, P < 0.05) predicted deficits in spontaneous speech (SS) and ulus (IFOF), inferior longitudinal fasciculus (ILF), middle longitudinal
naming scores. Each map shows the  -values identified by SVR- fasciculus (MdLF) and the posterior part of corticostriatal pathway. L,
LSM + DTLVC and the p-values obtained from 1000 permutation tests. left; R, right

FDR-corrected, P < 0.05). The volume of overlapped voxels more severe language impairments than those with LGGs.
of the two clusters was 1026 mm3. Tumor location in the left posterior middle temporal gyrus
Automatic fiber tracking on tractography atlas showed that was significantly associated with spontaneous speech and
five association or projection pathways passed through the naming scores in those with HGGs. These findings suggest
two identified clusters: arcuate fasciculus (the percentage of different types of structural plasticity underlie different de-
tracks passing through a target region and the total tracks of a grees of behavioral impairments in cases of LGG and HGG.
specific white matter pathway, spontaneous speech: 46/ To our knowledge, this is the first investigation that has clar-
1073 = 4.3 %; naming: 71/1073 = 6.6 %), inferior fronto- ified tumor grade-related macrostructural plasticity, based on
occipital fasciculus (spontaneous speech: 303/650 = 46.6 %; a large cohort of patients with glioma within the brain’s lan-
naming: 448/650 = 68.9 %), and inferior longitudinal fascicu- guage network.
lus (spontaneous speech: 2029/2574 = 78.8 %; naming:
2236/2574 = 86.9 %), middle longitudinal fasciculus (sponta-
neous speech: 138/142 = 97.2 %; naming: 129/142 = 90.9 %) The weak brain-symptom relationship in the LGG
and the posterior part of corticostriatal pathway (spontaneous group suggests large amounts of functional
speech: 126/906 = 13.9 %; naming: 136/906 = 15 %). reorganization
After regressing out sex, age, education and imaging cen-
ters from the language scores, no significant results were ob- The negative results from the lesion-language analyses for the
served for the LGG group and very similar results were ob- LGG group might have resulted from a great deal of function-
served for spontaneous speech and naming deficits of HGG al reorganization. For example, task-based functional MRI
group (Supplementary Figure S2). (fMRI) studies have shown that preoperative patients with
LGGs that infiltrated classical Broca’s, Wernicke’s, or insular
areas exhibited a redistribution of compensatory activations in
perilesional, remote ipsilesional, and contralesional
Discussion homotopic language-related areas. Further, they showed only
slight dysphasia (Benzagmout et al. 2007; Duffau 2005, 2012;
We performed lesion-language mapping to identify the struc- Duffau et al. 2001; Duffau and Taillandier 2015; Robles et al.
tural basis of language deficits in patients with left hemispher- 2008; Sarubbo et al. 2012). Similarly, a recent study has
ic gliomas. Most patients with LGGs (> 80 %) had nearly shown no significant difference in fMRI activation levels on
normal language function, and their language deficits could a picture naming task between preoperative patients with left
not be predicted by lesion topography. The weak brain- cerebral LGGs in the language network and control partici-
symptom relationship suggested a good amount of functional pants (Deverdun et al. 2019). Structurally, slowly progressive
reorganization of the cortex. Patients with HGGs exhibited glioma infiltration usually induces marked structural changes
Brain Imaging and Behavior

(Almairac et al. 2018; Desmurget et al. 2007; Zhang et al. Henseler et al. 2014; Pustina et al. 2016; Yourganov
2018). Moreover, even though both groups of patients in the et al. 2016), naming (Baldo et al. 2013; Harvey and
current study had tumor damage in the left pMTG, a bottle- Schnur 2015; Henseler et al. 2014; Pustina et al. 2016;
neck where both dorsal and ventral language pathways travel Yourganov et al. 2016), repetition (Butler et al. 2014;
(Fig. 2), the significant lesion-language association was only Henseler et al. 2014; Pustina et al. 2016; Yourganov
observed in the HGG group. Surgical neuro-oncological stud- et al. 2016), and phonology (Butler et al. 2014), as well
ies have shown that preoperative patients with LGGs that as poor language recovery (Bonilha et al. 2016;
infiltrate classical Broca’s, Wernicke’s, or insular areas could Fridriksson 2010; Ramsey et al. 2017). Taken together,
be resected without inducing permanent deficits (Duffau our findings suggest that damage in the left pMTG caused
2005, 2012; Lu et al. 2013; Picart et al. 2018). Thus, in cases by HGGs disrupts the signaling in dorsal and ventral lan-
of LGG, marked structural and functional changes occur as a guage processing streams and the deficits in speech pro-
means of compensating for the loss of function caused by duction cannot be compensated for due to the rapid glio-
tumor damage. This could explain the absence of cognitive ma infiltration.
deficits and the weak brain-symptom relationship that we ob-
served here.
Limitations
The macrostructural bases of language impairments
This study has several limitations. First, the sample sizes of the
in patient with HGGs
two groups were relatively small and cannot fully represent the
entire spectrum of patients. This might also affect the model
Converging evidence has shown that language processing re-
accuracy and stability (Cui and Gong 2018; Sperber et al.
cruits delocalized but structurally and functionally connected
2019). A large sample size from more centers is needed to in-
left lateralized cortico-subcortical areas (Cui et al. 2018;
crease representativeness and better test the reliability of the find-
Duffau et al. 2014; Fedorenko and Thompson-Schill 2014;
ings to independent datasets. Moreover, to ensure the statistical
Wu et al. 2015; Xu et al. 2016) and damage to these gray
power, only voxels lesioned in ≥ 10 patients were considered.
matter (Ramsey et al. 2017; Siegel et al. 2016) or the white-
The functional relevance of those voxels lesioned in < 10 patients
matter tracts that connect them (Fang et al. 2015; Griffis et al.
or intact brain regions need to be clarified in the future study.
2017; Yourganov et al. 2016) highly predict language deficits.
Second, although we speculate that the negative results for
However, our lesion-language mapping analysis showed that
LGGs might have been due to sufficient structural and func-
only the posterior MTG (pMTG) is associated with spontane-
tional compensation, we did find that 20 % of the LGG groups
ous speech and naming deficits in people with HGGs.
had aphasia. There were also aphasia HGG patients whose
Functional neuroimaging studies have shown that the
tumor spare the left pMTG. The aberrant structural and func-
left pMTG is consistently activated in phonological, se-
tional brain patterns for these patients, especially the remote
mantic, and sentence tasks (Binder 2017; Vigneau et al.
functional and structural disconnection (Boes et al. 2015;
2006). Diffusion tensor imaging studies have shown that
Foulon et al. 2018), underlying insufficient plasticity must
multiple long-range fiber pathways (Turken and Dronkers
be clarified in the future study.
2011), including dorsal (arcuate fasciculus) and ventral
Third, lesion-language mapping analysis only account for the
(inferior longitudinal fasciculus, middle longitudinal fas-
tumor infiltration effect. Gliomas cause symptoms through a
ciculus and inferior fronto-occipital fasciculus) language
combination of mass effect on the surrounding brain tissue and
pathways (Kummerer et al. 2013; Saur et al. 2008) tra-
direct infiltration of the brain tissue. Thus, longitudinal behaviour
verse this region (Fig. 2b). Additionally, a functional net-
and multimodal neuroimaging data are needed to fully under-
work study has shown that the left pMTG connects the
stand the neural mechanisms underlying glioma aphasia.
left frontoparietal network with the left perisylvian net-
work (Xu et al. 2016). Intraoperative electrocortical stim-
ulation of the left pMTG and its underlying white matter
pathways induces language deficits, which has been Conclusions
regarded it as a non-resectable area (Duffau et al. 2014;
Ius et al. 2011; Sanai et al. 2008; Wu et al. 2015). Our The present study found that tumor damage in the left pMTG,
results are also highly consistent with findings in patients which is located in the white matter confluence of dorsal and
with chronic left hemispheric stroke, which showed that ventral language pathways, was significantly associated with
damage to the left pMTG and its traversing fibers tracts the spontaneous speech and naming deficits exhibited by pa-
induced multiple language deficits (Corbetta et al. 2015; tients with HGGs. The negative results for the lesion-language
Griffis et al. 2017; Ivanova et al. 2016), including com- mapping in the LGG group suggest sufficient functional
prehension (Dronkers et al. 2004; Geva et al. 2012; compensation.
Brain Imaging and Behavior

Supplementary Information The online version contains supplementary Ashburner, J. (2007). A fast diffeomorphic image registration algorithm.
material available at https://doi.org/10.1007/s11682-021-00457-0. Neuroimage, 38(1), 95–113. doi:https://doi.org/10.1016/j.
neuroimage.2007.07.007.
Abbreviations LGG, low-grade glioma; HGG, high-grade glioma; Baldo, J. V., Arevalo, A., Patterson, J. P., & Dronkers, N. F. (2013). Grey
ABC, Aphasia Battery of Chinese; AQ, aphasia quotient; SS, spon- and white matter correlates of picture naming: evidence from a
taneous speech voxel-based lesion analysis of the Boston Naming Test. Cortex: A
Journal Devoted to the Study of the Nervous System and Behavior,
Acknowledgements The authors wish to thank Dr. Xiuyi Wang and 49(3), 658–667. https://doi.org/10.1016/j.cortex.2012.03.001.
Jianfeng Zhang for their advice in drafting the manuscript. Banerjee, P., Leu, K., Harris, R. J., Cloughesy, T. F., Lai, A., Nghiemphu,
P. L.,.. . Ellingson, B. M. (2015). Association between lesion loca-
tion and language function in adult glioma using voxel-based lesion-
Author contributions Conception and study design (JSW and JLC), data
symptom mapping. Neuroimage-Clinical, 9, 617–624. doi:https://
collection or acquisition (NZ, JFL and JY), statistical analysis (BKY),
doi.org/10.1016/j.nicl.2015.10.010.
interpretation of results (NZ, BKY, JSW and JFL), drafting the manu-
Bates, E., Wilson, S. M., Saygin, A. P., Dick, F., Sereno, M. I., Knight, R.
script work or revising it critically for important intellectual content (NZ,
T., & Dronkers, N. F. (2003). Voxel-based lesion-symptom map-
BKY and JFL) and approval of final version to be published and agree-
ping. Nature Neuroscience, 6(5), 448–450. https://doi.org/10.1038/
ment to be accountable for the integrity and accuracy of all aspects of the
nn1050.
work (All authors).
Benzagmout, M., Gatignol, P., & Duffau, H. (2007). Resection of World
Health Organization Grade II gliomas involving Broca’s area: meth-
Funding This work was supported by the Shenzhen Double Chain Grant odological and functional considerations. Neurosurgery, 61(4),
[2018]256, the Fundamental Research Funds for the Central Universities 741–752. doi:https://doi.org/10.1227/01.NEU.0000298902.69473.
(No. WK9110000133), Guangdong Key Basic Research Grant (No. 77. discussion 752 – 743.
2018B030332001), Guangdong Pearl River Talents Plan (No. Binder, J. R. (2017). Current controversies on Wernicke’s Area and its
2016ZT06S220), China Postdoctoral Science Foundation (No. role in language. Current Neurology and Neuroscience Reports,
2018M640825), National Natural Science Foundation of China (No.
17(8), 58. https://doi.org/10.1007/s11910-017-0764-8.
81401395, No.82001794), Natural Science Foundation of Anhui
Boes, A. D., Prasad, S., Liu, H. S., Liu, Q., Pascual-Leone, A., Caviness,
Province (No. 2008085QH380), Shanghai Young Talents in Health
V. S., & Fox, M. D. (2015). Network localization of neurological
(No. 2017YQ014), Key Program of Medical Science and Technique
symptoms from focal brain lesions. Brain, 138, 3061–3075. doi:
Foundation of Henan Province (No. SBGJ202002062), the Joint
https://doi.org/10.1093/brain/awv228.
Construction Program of Medical Science and Technique
Foundation of Henan Province (No. LHGJ20190156), Research Bonilha, L., Gleichgerrcht, E., Nesland, T., Rorden, C., & Fridriksson, J.
Projects of Henan Higher Education (No.18A320077), the Scientific (2016). Success of anomia treatment in aphasia is associated with
and Technological Research Projects of Henan Province preserved architecture of global and left temporal lobe structural
(No.192102310123). J.S.W is supported by Shanghai Municipal networks. Neurorehabilitation and Neural Repair, 30(3), 266–
Science and Technology Major Project (No. 2018SHZDZX03) and ZJ 279. https://doi.org/10.1177/1545968315593808.
Lab. The funding agencies took no part in the design or implementation Butler, R. A., Lambon Ralph, M. A., & Woollams, A. M. (2014).
of the research. Capturing multidimensionality in stroke aphasia: mapping principal
behavioural components to neural structures. Brain, 137(Pt 12),
3248–3266. doi:https://doi.org/10.1093/brain/awu286.
Declarations All processes strictly followed the requirements of the Chang, C.-C., & Lin, C.-J. (2011). LIBSVM: A library for support vector
Declaration of Helsinki. This study was approved and supervised by the machines. ACM Transactions on Intelligent Systems and
Ethics Committees of Huashan Hospital and First Affiliated Hospital of Technology, 2(3), 1–27. https://doi.org/10.1145/1961189.1961199.
Zhengzhou University. Written informed consent was obtained from the Corbetta, M., Ramsey, L., Callejas, A., Baldassarre, A., Hacker, C. D.,
legal guardians of all patients. Siegel, J. S.,.. . Shulman, G. L. (2015). Common behavioral clusters
and subcortical anatomy in stroke. Neuron, 85(5), 927–941. doi:
Competing financial interests The authors declare no competing finan- https://doi.org/10.1016/j.neuron.2015.02.027.
cial interests. Cui, Z., & Gong, G. (2018). The effect of machine learning regression
algorithms and sample size on individualized behavioral prediction
with functional connectivity features. Neuroimage, 178, 622–637.
doi:https://doi.org/10.1016/j.neuroimage.2018.06.001.
Cui, Z., Su, M., Li, L., Shu, H., & Gong, G. (2018). Individualized
References prediction of reading comprehension ability using gray matter vol-
ume. Cerebral Cortex, 28(5), 1656–1672. https://doi.org/10.1093/
Aerts, H., Fias, W., Caeyenberghs, K., & Marinazzo, D. (2016). Brain cercor/bhx061.
networks under attack: robustness properties and the impact of le- Desmurget, M., Bonnetblanc, F., & Duffau, H. (2007). Contrasting acute
sions. Brain, 139(Pt 12), 3063–3083. doi:https://doi.org/10.1093/ and slow-growing lesions: a new door to brain plasticity. Brain,
brain/aww194. 130(Pt 4), 898–914. doi:https://doi.org/10.1093/brain/awl300.
Almairac, F., Duffau, H., & Herbet, G. (2018). Contralesional macro- Deverdun, J., van Dokkum, L. E. H., Le Bars, E., Herbet, G., Mura, T.,
structural plasticity of the insular cortex in patients with glioma: A D’Agata, B., & Moritz Gasser, S. (2019). Language reorganization
VBM study. Neurology, 91(20), e1902–e1908. https://doi.org/10. after resection of low-grade gliomas: an fMRI task based connectiv-
1212/WNL.0000000000006517. ity study. Brain Imaging and Behavior. https://doi.org/10.1007/
Almairac, F., Herbet, G., Moritz-Gasser, S., de Champfleur, N. M., & s11682-019-00114-7.
Duffau, H. (2015). The left inferior fronto-occipital fasciculus sub- Dosenbach, N. U., Nardos, B., Cohen, A. L., Fair, D. A., Power, J. D.,
serves language semantics: a multilevel lesion study. Brain Church, J. A.,.. . Schlaggar, B. L. (2010). Prediction of individual
Structure & Function, 220(4), 1983–1995. https://doi.org/10.1007/ brain maturity using fMRI. Science, 329(5997), 1358–1361. doi:
s00429-014-0773-1. https://doi.org/10.1126/science.1194144.
Brain Imaging and Behavior

Dronkers, N. F., Wilkins, D. P., Van Valin, R. D. Jr., Redfern, B. B., & network robustness: a prospective cohort study. Neurosurgery,
Jaeger, J. J. (2004). Lesion analysis of the brain areas involved in 84(6), 1201–1213. https://doi.org/10.1093/neuros/nyy378.
language comprehension. Cognition, 92(1–2), 145–177. doi:https:// Harvey, D. Y., & Schnur, T. T. (2015). Distinct loci of lexical and se-
doi.org/10.1016/j.cognition.2003.11.002. mantic access deficits in aphasia: Evidence from voxel-based lesion-
Duffau, H. (2005). Lessons from brain mapping in surgery for low-grade symptom mapping and diffusion tensor imaging. Cortex: A Journal
glioma: insights into associations between tumour and brain plastic- Devoted to the Study of the Nervous System and Behavior, 67, 37–
ity. Lancet Neurology, 4(8), 476–486. https://doi.org/10.1016/ 58. https://doi.org/10.1016/j.cortex.2015.03.004.
S1474-4422(05)70140-X. Henseler, I., Regenbrecht, F., & Obrig, H. (2014). Lesion correlates of
Duffau, H. (2012). The “frontal syndrome” revisited: lessons from patholinguistic profiles in chronic aphasia: comparisons of syn-
electrostimulation mapping studies. Cortex: A Journal Devoted to drome-, modality- and symptom-level assessment. Brain, 137(Pt
the Study of the Nervous System and Behavior, 48(1), 120–131. 3), 918–930. doi:https://doi.org/10.1093/brain/awt374.
https://doi.org/10.1016/j.cortex.2011.04.029. Herbet, G., Lafargue, G., Bonnetblanc, F., Moritz-Gasser, S., Menjot de
Duffau, H., Bauchet, L., Lehericy, S., & Capelle, L. (2001). Functional Champfleur, N., & Duffau, H. (2014). Inferring a dual-stream model
compensation of the left dominant insula for language. Neuroreport, of mentalizing from associative white matter fibres disconnection.
12(10), 2159–2163. Brain, 137(Pt 3), 944–959. doi:https://doi.org/10.1093/brain/
Duffau, H., Moritz-Gasser, S., & Mandonnet, E. (2014). A re- awt370.
examination of neural basis of language processing: proposal of a Herbet, G., Moritz-Gasser, S., Boiseau, M., Duvaux, S., Cochereau, J., &
dynamic hodotopical model from data provided by brain stimulation Duffau, H. (2016). Converging evidence for a cortico-subcortical
mapping during picture naming. Brain and Language, 131, 1–10. network mediating lexical retrieval. Brain, 139(11), 3007–3021.
https://doi.org/10.1016/j.bandl.2013.05.011. doi:https://doi.org/10.1093/brain/aww220.
Duffau, H., & Taillandier, L. (2015). New concepts in the management of Huang, H., Lu, J., Wu, J., Ding, Z., Chen, S., Duan, L., & Zhang, H.
diffuse low-grade glioma: Proposal of a multistage and individual- (2018). Tumor tissue detection using blood-oxygen-level-dependent
ized therapeutic approach. Neuro-Oncology, 17(3), 332–342. functional mri based on independent component analysis. Scientific
https://doi.org/10.1093/neuonc/nou153. Reports, 8(1), 1223. https://doi.org/10.1038/s41598-017-18453-0.
Fang, Y., Han, Z., Zhong, S., Gong, G., Song, L., Liu, F., & Bi, Y. Ius, T., Angelini, E., Thiebaut de Schotten, M., Mandonnet, E., & Duffau,
(2015). The semantic anatomical network: Evidence from healthy H. (2011). Evidence for potentials and limitations of brain plasticity
and brain-damaged patient populations. Human Brain Mapping, using an atlas of functional resectability of WHO grade II gliomas:
36(9), 3499–3515. https://doi.org/10.1002/hbm.22858. towards a “minimal common brain”. Neuroimage, 56(3), 992–1000.
doi:https://doi.org/10.1016/j.neuroimage.2011.03.022.
Fedorenko, E., Hsieh, P. J., Nieto-Castanon, A., Whitfield-Gabrieli, S., &
Ivanova, M. V., Isaev, D. Y., Dragoy, O. V., Akinina, Y. S.,
Kanwisher, N. (2010). New method for fMRI investigations of lan-
Petrushevskiy, A. G., Fedina, O. N., & Dronkers, N. F. (2016).
guage: defining ROIs functionally in individual subjects. Journal of
Diffusion-tensor imaging of major white matter tracts and their role
Neurophysiology, 104(2), 1177–1194. https://doi.org/10.1152/jn.
in language processing in aphasia. Cortex: A Journal Devoted to the
00032.2010.
Study of the Nervous System and Behavior, 85, 165–181. https://doi.
Fedorenko, E., & Thompson-Schill, S. L. (2014). Reworking the lan-
org/10.1016/j.cortex.2016.04.019.
guage network. Trends in Cognitive Sciences, 18(3), 120–126.
Kimberg, D. Y., Coslett, H. B., & Schwartz, M. F. (2007). Power in
https://doi.org/10.1016/j.tics.2013.12.006.
Voxel-based lesion-symptom mapping. Journal of Cognitive
Foulon, C., Cerliani, L., Kinkingnehun, S., Levy, R., Rosso, C.,
Neuroscience, 19(7), 1067–1080. https://doi.org/10.1162/jocn.
Urbanski, M., ... Thiebaut De Schotten, M. (2018). Advanced lesion
2007.19.7.1067.
symptom mapping analyses and implementation as BCBtoolkit.
Kinoshita, M., Nakajima, R., Shinohara, H., Miyashita, K., Tanaka, S.,
Gigascience, 7(3). https://doi.org/10.1093/gigascience/giy004.
Okita, H., & Hayashi, Y. (2016). Chronic spatial working memory
Fox, M. E., & King, T. Z. (2018). Functional connectivity in adult brain deficit associated with the superior longitudinal fasciculus: a study
tumor patients: a systematic review. Brain Connectivity, 8(7), 381– using voxel-based lesion-symptom mapping and intraoperative di-
397. https://doi.org/10.1089/brain.2018.0623. rect stimulation in right prefrontal glioma surgery. Journal of
Fridriksson, J. (2010). Preservation and modulation of specific left hemi- Neurosurgery, 125(4), 1024–1032. https://doi.org/10.3171/2015.
sphere regions is vital for treated recovery from anomia in stroke. 10.JNS1591.
The Journal of Neuroscience, 30(35), 11558–11564. https://doi.org/ Klein, A., Andersson, J., Ardekani, B. A., Ashburner, J., Avants, B.,
10.1523/JNEUROSCI.2227-10.2010. Chiang, M. C.,.. . Parsey, R. V. (2009). Evaluation of 14 nonlinear
Gao, S., Zhu, Y., Shi, S., Peng, Y., et al. (1992). Standard Aphasia Battery deformation algorithms applied to human brain MRI registration.
of Chinese. [Standard Aphasia Battery of Chinese.]. Chinese Mental Neuroimage, 46(3), 786–802. doi:https://doi.org/10.1016/j.
Health Journal, 6(3), 125–128. neuroimage.2008.12.037.
Geva, S., Baron, J. C., Jones, P. S., Price, C. J., & Warburton, E. A. Kummerer, D., Hartwigsen, G., Kellmeyer, P., Glauche, V., Mader, I.,
(2012). A comparison of VLSM and VBM in a cohort of patients Kloppel, S.,.. . Saur, D. (2013). Damage to ventral and dorsal lan-
with post-stroke aphasia. NeuroImage: Clinical, 1(1), 37–47. https:// guage pathways in acute aphasia. Brain, 136(Pt 2), 619–629. doi:
doi.org/10.1016/j.nicl.2012.08.003. https://doi.org/10.1093/brain/aws354.
Ghinda, D. C., Wu, J. S., Duncan, N. W., & Northoff, G. (2018). How Louis, D. N., Ohgaki, H., Wiestler, O. D., Cavenee, W. K., Burger, P. C.,
much is enough-Can resting state fMRI provide a demarcation for Jouvet, A., & Kleihues, P. (2007). The 2007 WHO classification of
neurosurgical resection in glioma? Neuroscience and Biobehavioral tumours of the central nervous system. Acta Neuropathologica,
Reviews, 84, 245–261. https://doi.org/10.1016/j.neubiorev.2017.11. 114(2), 97–109. https://doi.org/10.1007/s00401-007-0243-4.
019. Lu, J., Wu, J., Yao, C., Zhuang, D., Qiu, T., Hu, X., & Zhou, L. (2013).
Griffis, J. C., Nenert, R., Allendorfer, J. B., & Szaflarski, J. P. (2017). Awake language mapping and 3-Tesla intraoperative MRI-guided
Damage to white matter bottlenecks contributes to language impair- volumetric resection for gliomas in language areas. Journal of
ments after left hemispheric stroke. NeuroImage: Clinical, 14, 552– Clinical Neuroscience, 20(9), 1280–1287. https://doi.org/10.1016/
565. https://doi.org/10.1016/j.nicl.2017.02.019. j.jocn.2012.10.042.
Hart, M. G., Romero-Garcia, R., Price, S. J., & Suckling, J. (2019). Lu, J., Zhang, H., Hameed, N. U. F., Zhang, J., Yuan, S., Qiu, T., & Wu,
Global effects of focal brain tumors on functional complexity and J. (2017). An automated method for identifying an independent
Brain Imaging and Behavior

component analysis-based language-related resting-state network in Saur, D., Kreher, B. W., Schnell, S., Kummerer, D., Kellmeyer, P., Vry,
brain tumor subjects for surgical planning. Scientific Reports, 7(1), M. S., & Weiller, C. (2008). Ventral and dorsal pathways for lan-
13769. https://doi.org/10.1038/s41598-017-14248-5. guage. Proceedings of the National Academy of Sciences of the
Mah, Y. H., Husain, M., Rees, G., & Nachev, P. (2014). Human brain United States of America, 105(46), 18035–18040. https://doi.org/
lesion-deficit inference remapped. Brain, 137(Pt 9), 2522–2531. 10.1073/pnas.0805234105.
doi:https://doi.org/10.1093/brain/awu164. Siegel, J. S., Ramsey, L. E., Snyder, A. Z., Metcalf, N. V., Chacko, R. V.,
Paternostro-Sluga, T., Grim-Stieger, M., Posch, M., Schuhfried, O., Weinberger, K., & Corbetta, M. (2016). Disruptions of network
Vacariu, G., Mittermaier, C., & Fialka-Moser, V. (2008). connectivity predict impairment in multiple behavioral domains af-
Reliability and validity of the Medical Research Council (MRC) ter stroke. Proceedings of the National Academy of Sciences of the
scale and a modified scale for testing muscle strength in patients United States of America, 113(30), E4367–E4376. https://doi.org/
with radial palsy. Journal of Rehabilitation Medicine, 40(8), 665– 10.1073/pnas.1521083113.
671. https://doi.org/10.2340/16501977-0235. Sperber, C., & Karnath, H. O. (2017). Impact of correction factors in
Payne, B. R., & Lomber, S. G. (2001). Reconstructing functional systems human brain lesion-behavior inference. Human Brain Mapping,
after lesions of cerebral cortex. Nature Reviews Neuroscience, 2(12), 38(3), 1692–1701. https://doi.org/10.1002/hbm.23490.
911–919. https://doi.org/10.1038/35104085. Sperber, C., Wiesen, D., & Karnath, H. O. (2019). An empirical evalua-
Picart, T., Herbet, G., Moritz-Gasser, S., & Duffau, H. (2018). Iterative tion of multivariate lesion behaviour mapping using support vector
surgical resections of diffuse glioma with awake mapping: How to regression. Human Brain Mapping, 40(5), 1381–1390. doi:https://
deal with cortical plasticity and connectomal constraints? doi.org/10.1002/hbm.24476.
Neurosurgery. https://doi.org/10.1093/neuros/nyy218. Turken, A. U., & Dronkers, N. F. (2011). The neural architecture of the
Pisoni, A., Mattavelli, G., Casarotti, A., Comi, A., Riva, M., Bello, L., & language comprehension network: converging evidence from lesion
Papagno, C. (2018). Object-action dissociation: A voxel-based le- and connectivity analyses. Frontiers in Systems Neuroscience, 5, 1.
sion-symptom mapping study on 102 patients after glioma removal. https://doi.org/10.3389/fnsys.2011.00001.
NeuroImage: Clinical, 18, 986–995. https://doi.org/10.1016/j.nicl. Vigneau, M., Beaucousin, V., Herve, P. Y., Duffau, H., Crivello, F.,
2018.03.022. Houde, O.,.. . Tzourio-Mazoyer, N. (2006). Meta-analyzing left
Pisoni, A., Mattavelli, G., Casarotti, A., Comi, A., Riva, M., Bello, L., & hemisphere language areas: phonology, semantics, and sentence
Papagno, C. (2019). The neural correlates of auditory-verbal short- processing. Neuroimage, 30(4), 1414–1432. doi:https://doi.org/10.
term memory: a voxel-based lesion-symptom mapping study on 103 1016/j.neuroimage.2005.11.002.
patients after glioma removal. Brain Structure & Function, 224(6),
Wang, Y. Y., Qian, T. Y., You, G., Peng, X. X., Chen, C., You, Y. P., &
2199–2211. https://doi.org/10.1007/s00429-019-01902-z.
Jiang, T. (2015). Localizing seizure-susceptible brain regions asso-
Pustina, D., Coslett, H. B., Turkeltaub, P. E., Tustison, N., Schwartz, M.
ciated with low-grade gliomas using voxel-based lesion-symptom
F., & Avants, B. (2016). Automated segmentation of chronic stroke
mapping. Neuro-Oncology, 17(2), 282–288. https://doi.org/10.
lesions using LINDA: Lesion identification with neighborhood data
1093/neuonc/nou130.
analysis. Human Brain Mapping, 37(4), 1405–1421. https://doi.org/
Warren, D. E., Power, J. D., Bruss, J., Denburg, N. L., Waldron, E. J.,
10.1002/hbm.23110.
Sun, H., & Tranel, D. (2014). Network measures predict neuropsy-
Ramsey, L. E., Siegel, J. S., Lang, C. E., Strube, M., Shulman, G. L., &
chological outcome after brain injury. Proceedings of the National
Corbetta, M. (2017). Behavioural clusters and predictors of perfor-
Academy of Sciences of the United States of America, 111(39),
mance during recovery from stroke. Nature Human Behaviour, 1.
14247–14252. https://doi.org/10.1073/pnas.1322173111.
https://doi.org/10.1038/s41562-016-0038.
Ripolles, P., Marco-Pallares, J., de Diego-Balaguer, R., Miro, J., Falip, Wen, P. Y., Macdonald, D. R., Reardon, D. A., Cloughesy, T. F.,
M., Juncadella, M.,.. . Rodriguez-Fornells, A. (2012). Analysis of Sorensen, A. G., Galanis, E., & Chang, S. M. (2010). Updated
automated methods for spatial normalization of lesioned brains. response assessment criteria for high-grade gliomas: response as-
Neuroimage, 60(2), 1296–1306. doi:https://doi.org/10.1016/j. sessment in neuro-oncology working group. Journal of Clinical
neuroimage.2012.01.094. Oncology, 28(11), 1963–1972. https://doi.org/10.1200/JCO.2009.
Robles, S. G., Gatignol, P., Lehericy, S., & Duffau, H. (2008). Long-term 26.3541.
brain plasticity allowing a multistage surgical approach to World Weng, H. H., Noll, K. R., Johnson, J. M., Prabhu, S. S., Tsai, Y. H.,
Health Organization Grade II gliomas in eloquent areas. Journal of Chang, S. W., & Liu, H. L. (2018). Accuracy of presurgical func-
Neurosurgery, 109(4), 615–624. https://doi.org/10.3171/JNS/2008/ tional MR imaging for language mapping of brain tumors: a system-
109/10/0615. atic review and meta-analysis. Radiology, 286(2), 512–523. https://
Rorden, C., Karnath, H. O., & Bonilha, L. (2007). Improving lesion- doi.org/10.1148/radiol.2017162971.
symptom mapping. Journal of Cognitive Neuroscience, 19(7), Winkler, A. M., Ridgway, G. R., Webster, M. A., Smith, S. M., &
1081–1088. https://doi.org/10.1162/jocn.2007.19.7.1081. Nichols, T. E. (2014). Permutation inference for the general linear
Sagberg, L. M., Iversen, D. H., Fyllingen, E. H., Jakola, A. S., Reinertsen, model. Neuroimage, 92, 381–397. doi:https://doi.org/10.1016/j.
I., & Solheim, O. (2019). Brain atlas for assessing the impact of neuroimage.2014.01.060.
tumor location on perioperative quality of life in patients with Wu, J., Lu, J., Zhang, H., Zhang, J., Yao, C., Zhuang, D., & Zhou, L.
high-grade glioma: A prospective population-based cohort study. (2015). Direct evidence from intraoperative electrocortical stimula-
Neuroimage: Clinical, 21, 101658. https://doi.org/10.1016/j.nicl. tion indicates shared and distinct speech production center between
2019.101658. Chinese and English languages. Human Brain Mapping, 36(12),
Sanai, N., Mirzadeh, Z., & Berger, M. S. (2008). Functional outcome 4972–4985. https://doi.org/10.1002/hbm.22991.
after language mapping for glioma resection. The New England Xu, Y., Lin, Q., Han, Z., He, Y., & Bi, Y. (2016). Intrinsic functional
Journal of Medicine, 358(1), 18–27. https://doi.org/10.1056/ network architecture of human semantic processing: Modules and
NEJMoa067819. hubs. Neuroimage, 132, 542–555. doi:https://doi.org/10.1016/j.
Sarubbo, S., Le Bars, E., Moritz-Gasser, S., & Duffau, H. (2012). neuroimage.2016.03.004.
Complete recovery after surgical resection of left Wernicke’s area Yang, M., Li, J., Li, Y., Li, R., Pang, Y., Yao, D., & Chen, H. (2016).
in awake patient: a brain stimulation and functional MRI study. Altered intrinsic regional activity and interregional functional con-
Neurosurgical Review, 35(2), 287–292. https://doi.org/10.1007/ nectivity in post-stroke aphasia. Scientific Reports, 6, 24803. https://
s10143-011-0351-4 discussion 292. doi.org/10.1038/srep24803.
Brain Imaging and Behavior

Yeh, F. C. (2020). Shape analysis of the human association pathways. Yuan, B., Zhang, N., Yan, J., Cheng, J., Lu, J., & Wu, J. (2020). Tumor
Neuroimage, 223, 117329. doi:https://doi.org/10.1016/j. grade-related language and control network reorganization in pa-
neuroimage.2020.117329. tients with left cerebral glioma. Cortex: A Journal Devoted to the
Yeh, F. C., & Tseng, W. Y. (2011). NTU-90: a high angular resolution Study of the Nervous System and Behavior, 129, 141–157. https://
brain atlas constructed by q-space diffeomorphic reconstruction. doi.org/10.1016/j.cortex.2020.04.015.
Neuroimage, 58(1), 91–99. doi:https://doi.org/10.1016/j. Zhang, H., Shi, Y., Yao, C., Tang, W., Yao, D., Zhang, C., & Song, Z.
neuroimage.2011.06.021. (2016). Alteration of the intra- and cross- hemisphere posterior de-
Yeh, F. C., Wedeen, V. J., & Tseng, W. Y. (2010). Generalized q- fault mode network in frontal lobe glioma patients. Scientific
sampling imaging. IEEE Transactions on Medical Imaging, 29(9), Reports, 6, 26972. https://doi.org/10.1038/srep26972.
1626–1635. https://doi.org/10.1109/TMI.2010.2045126. Zhang, N., Xia, M., Qiu, T., Wang, X., Lin, C. P., Guo, Q., & Zhou, L.
Yourganov, G., Fridriksson, J., Rorden, C., Gleichgerrcht, E., & Bonilha, (2018). Reorganization of cerebro-cerebellar circuit in patients with
L. (2016). Multivariate connectome-based symptom mapping in left hemispheric gliomas involving language network: A combined
post-stroke patients: networks supporting language and speech. structural and resting-state functional MRI study. Human Brain
The Journal of Neuroscience, 36(25), 6668–6679. https://doi.org/ Mapping, 39(12), 4802–4819. https://doi.org/10.1002/hbm.24324.
10.1523/JNEUROSCI.4396-15.2016. Zhang, Y., Kimberg, D. Y., Coslett, H. B., Schwartz, M. F., & Wang, Z.
Yuan, B., Fang, Y., Han, Z., Song, L., He, Y., & Bi, Y. (2017). Brain (2014). Multivariate lesion-symptom mapping using support vector
hubs in lesion models: Predicting functional network topology with regression. Human Brain Mapping, 35(12), 5861–5876. https://doi.
lesion patterns in patients. Scientific Reports, 7(1), 17908. https:// org/10.1002/hbm.22590.
doi.org/10.1038/s41598-017-17886-x.
Yuan, B., Zhang, N., Yan, J., Cheng, J., Lu, J., & Wu, J. (2019). Resting-
state functional connectivity predicts individual language impair- Publisher’s note Springer Nature remains neutral with regard to jurisdic-
ment of patients with left hemispheric gliomas involving language tional claims in published maps and institutional affiliations.
network. Neuroimage: Clinical, 24, 102023. https://doi.org/10.
1016/j.nicl.2019.102023.

You might also like