Assessment of 3DTV-related fatigue with resting-state fMRI

Accepted Manuscript Assessment of 3DTV-related fatigue with resting-state fMRI Chunxiao Chen, Jing Wang, Xiong Lu, Yupin Liu, Xin Chen

PII: DOI: Reference:

S0923-5965(18)30184-X https://doi.org/10.1016/j.image.2018.02.015 IMAGE 15342

To appear in:

Signal Processing: Image Communication

Received date : 19 January 2017 Revised date : 27 February 2018 Accepted date : 27 February 2018 Please cite this article as: C. Chen, J. Wang, X. Lu, Y. Liu, X. Chen, Assessment of 3DTV-related fatigue with resting-state fMRI, Signal Processing: Image Communication (2018), https://doi.org/10.1016/j.image.2018.02.015 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Manuscript

Assessment of 3DTV-related fatigue with resting-state fMRI Chunxiao Chena, *, Jing Wanga, Xiong Lub, Yupin Liuc, Xin Chenc a

Department of Biomedical Engineering.

Nanjing University of Aeronautics & Astronautics, Jiangsu Nanjing, 211106, China b

Department of Measurement and Testing Engineering，

Nanjing University of Aeronautics & Astronautics, Jiangsu Nanjing, 211106, China c

Department of Radiology,

Guangdong Province Traditional Chinese Medical Hospital, Guangzhou, 510006, China

Abstract Objective: This paper studies the fatigue caused by prolonged watching of three-dimensional television (3DTV) by applying the amplitude of low frequency fluctuation (ALFF) and the community partitioning algorithm in brain functional network techniques to resting-state functional magnetic resonance imaging (fMRI). In this study, two specific frequency bands (Slow5: 0.01-0.027 Hz and Slow4: 0.027-0.073 Hz) were analyzed to investigate the change of brain spontaneous activity after one-hour watching of 3DTV. Meanwhile, the changes of brain modular organization were analyzed to discuss the physiological mechanism of brain fatigue during long period of watching 3DTV. Methods: Valid data of resting state fMRI from 20 dextromanual subjects were adopted before and after 1 hour watching of 3DTV/2DTV. Then, ALFF was processed using SPM8 and REST software to analyze the variation of resting-state fMRI induced by watching

*

Corresponding author. Tel.: +86 025 84891938 E-mail address:[email protected](C.X. Chen). Address: No.169 Sheng Tai West Road, Jiang Ning District, Nanjing, Jiangsu Province, 211106, P R China.

2D/3DTV. Community partitioning algorithm was used to evaluate the change of brain modular organization. Results: While watching 2DTV does not significantly change the brain spontaneous activity, watching 3DTV makes significant differences on middle frontal gyrus (BA9/BA10), right inferior occipital gyrus (BA18/19), right inferior temporal gyrus (BA20/37) and cingulate cortex (BA32). The result shows that dorsal stream was activated after watching 3DTV through analyzed modularity and community, which verifies that watching 3DTV requires the brain visual pathway to process stereoscopic display. Meanwhile, the result further reflects that the neural circuit, including visual conducting pathway, temporal lobe and basal ganglia loop, is closely related to brain fatigue caused by long period of watching 3DTV. The study has shown the potential risk of 3DTV to viewers’ health, especially on vision and recognition functions. The basal ganglia neurons play crucial roles in the physiological mechanism of brain fatigue. Therefore, prolonged 3DTV watching should be avoided particularly for physically developing teenagers. This research suggested that the ALFF technique and brain modular organization is able to reveal intrinsic changes caused by visual fatigue and provided a foundation for 3DTV fatigue assessment. Keywords: 3DTV; Resting state; ALFF; visual fatigue; recognition function; modularity;

1 Introduction 3DTV has been very popular because of its advantages in realistic and stereo sense. However, a number of important perceptual and human factors issues that arise during watching stereo displays, such as the basic of human stereopsis, interocular cross talk and high-level cognitive factor [1]. Lambooij reviewed the concept of visual fatigue and its subjective counterpart, visual discomfort, in relation to stereoscopic display technology and image generation [2]. Additionally, the uncoupling of vergence and accommodation required by 3DTV can cause both visual discomfort and mental fatigue for audiences [3-5]. These issues could lead to eye irritation, blurred vision and difficulty in concentrating, as well as decrease of sensitivity, vigilance, brain excitability and reduced efficiency [6-7]. In order to safeguard the viewers’ health and promote the development of stereoscopic industry, many scholars have performed a lot of related research in recent years. Atallah[8] concluded that visual stress and discomfort associated experimental purpose with 3D viewing is prevalent for a significant proportion of the healthy consumer population based questionnaires. Chen designed subjective questionnaire to evaluate visual fatigue caused by watching 2DTV/3DTV, and noticed significant variations of electroencephalogram (EEG) in the occipital region after watching 3DTV, which is closely related to visual function [9]. Brunnström designed five different subjective experiments to investigate the overall effects of symptoms that 3DTV viewing can induce by collecting the Simulator Sickness Questionnaire (SSQ) data [10]. Yu [11] utilized the eye movement signal as an indicator for visual fatigue caused by 3D displays and found that compared with 2D videos, 3D videos increased the blink frequency and scanning range, both proportional to visual fatigue. Cho [12] focused on the relationship between stereoscopic depth and visual discomfort caused by watching stereoscopic 3D content and results showed a positive correlation between them.

Chen [13] applied task-related fMRI to assess the visual fatigue state, and the comparison found that prolonged watching of 3DTV could result in more serious visual fatigue than 2DTV. However, current studies have not attached much importance to the brain changes during resting state caused by prolonged watching of 2D or 3DTV, which could be reflected by spontaneous fluctuations in brain region and modular organization of brain functional network. Slow fluctuations in activity are fundamental features of the resting brain, and the relative magnitudes of these fluctuations are different among different brain regions or subjects. Therefore, these characteristics may be used as markers for individual differences or dysfunctions [14]. ALFF and fractional amplitude of low frequency fluctuations (fALFF) can quantify the amplitude of these low frequency oscillations. ALFF is defined as the total power within the low frequency range (0.01-0.08 Hz), and thus indexes the strength or intensity of low frequency oscillations. FALFF is defined as ALFF divided by the total power in the entire detectable frequency range, and represents the relative contribution of specific low frequency oscillations to the whole frequency range [15]. Children with ADHD (attention-deficit hyperactivity disorder) have reduced ALFF amplitude in some brain areas and increased amplitude in others compared to controls [16]. In addition, Yan [17] has found increased amplitude in the Default Mode Network during Eyes Open vs. Eyes Closed resting periods. Changes in fALFF could also occur with aging [18]. Though both ALFF and fALFF show sensitivity mostly of signals from gray matter, ALFF is generally more reliable than fALFF [19]. Furthermore, gray matter related fluctuation amplitudes mainly occur in two frequency bands, 0.01-0.027 Hz (Slow5) and 0.027-0.073 Hz (Slow4), and these two given frequency bands contribute differently to the ALFF, suggesting that individual frequency bands could link to specific characteristics [20-22]. Slow (<0.1Hz) spontaneous fluctuations in the blood oxygen level-dependent

(BOLD) signal reveal patterns of brain functional networks [23]. Brain network community can reflect brain function module recombination and differentiation from a global perspective compare to regional measurements of ALFF. Keiichi has found the modularity of brain functional network was negatively correlated with age [24]. Felix investigated the role of dopamine in the topological organization of brain networks at rest, and concluded that dopamine plays a role in maintaining high modularity of functional brain networks [25]. Our previous fMRI studies used graph theory to detect changes of brain functional networks before and after watching 2D/3DTV [26]. In this research, we applied the ALFF technique to study the regional differences of resting brain slow fluctuation and used brain modular organization to analyze the global changes of brain functional network after watching 2D/3DTV for one hour. To examine the hypothesis about visual and mental fatigue underlying watching 3DTV, the characteristic patterns of ALFF changes in 3D group will be discussed, especially Slow5 and Slow4. We aim to decide whether the ALFF and brain network community partitioning techniques could reveal the intrinsic changes caused by visual fatigue, which may provide a foundation for 3DTV fatigue assessment. 2 Materials and methods 2.1 Subjects The major consumers of the 3DTV market are young people. In order to decrease the differences caused by age, living habits, etc., test subjects were recruited from the university. A total of 20 healthy dextromanual subjects with normal or corrected vision (10 males, 10 females, with the youngest to be 19 and oldest to be 25, the average age 22.5) were recruited from university. Eight 3D-movie clips with four levels of depth of field (1/2 in front of and 1/2 behind the focus point) were displayed to ensure that they had normal stereoscopic sensitivity by pointing out the distance of

object behind or in front of the screen according to reference system. In the meantime, they do not have medical contraindications such as severe concomitant disease, alcoholism, drug abuse, as well as psychological or intellectual problems which are likely to limit compliance. All subjects have been informed with the experimental purposes and procedures in advance, and the study has been approved by the local ethics committee. They were reported to be in good condition both physically and mentally and have great sleep quality prior to the experiment. Subjects have signed an informed consent form and received CNY300 in compensation. To avoid low blood sugar levels that may affect the experiment results, the subjects are required to have a meal about 1 to 2 hours before the experiment. 2.2 Research design and data acquisition This experiment adopted a 46 inches Haixin LED46XT39G3D LED TV with a screen ratio of 16:9 and active shutter glasses with model FPS3D02. The manufacturer recommended viewing distance is 3 meters. The stimuli is Amazing Ocean with 1920 x 1080–pixel resolution and a wide depth range from 400mm to 6300mm. The average depth of crossed disparity is 487 millimeters and uncrossed disparity is 2610 millimeters. Subjects were arranged to watch both 3DTV (3D group) and 2DTV (2D group) on two different days, given at least 48 hours apart, in a dim light room. To obtain the variation of resting-state fMRI induced by watching 2D/3DTV, each subject underwent five minutes fMRI scan during a conscious resting state before and immediately after 1-hour watching 3DTV or 2DTV under the same conditions (Pre and Post sessions, task in 3DTV or 2DTV groups, respectively). The entire scans were carried out in a quiet and comfortable environment. Subjects lay flat inside the scanner with their heads immobilized, wearing blank blinder to avoid stimuli during the resting-state fMRI scans. In addition, subjects were instructed to awake with their eyes closed,

remain as still as possible, relax and think of nothing in particular, or try to concentrate on one thing. All MRI scans were obtained on GESignaHDx3.0T MR system in Guangdong Province Traditional Chinese Medical Hospital. Functional images were collected axially using an echo-planar imaging (EPI) sequence sensitive to blood oxygen level dependent (BOLD) contrast. The acquisition parameters were as follows: repetition time (TR) = 2000ms, echo time (TE) = 30ms, flip angle (FA) = 90°, thickness/gap = 4/0mm, matrix size of a single slice = 64×64, field of view (FOV) = 240×240mm and 32 axial slices. T1-weighted images covering the whole brain were then obtained with the following parameters: 196 slices, TR = 7.796, TE = 2.984ms, FA = 12°, thickness/gap = 1/1 mm, matrix size of a single slice = 256×256, FOV = 256×256mm and 196 sagittal slices. 2.3 Data pre-processing Pre-processing of the fMRI image was carried out using SPM8 (Statistical Parametric Mapping, Version 8, Welcome Department of Cognitive Neurology). The first five time points of the functional images were discarded to reduce the impact from instability of the initial MRI signals and the participants' adaptation to the scanning environment. The remaining functional scans were corrected between slices for within-scan acquisition time differences. Then “Realignment” was used to either retrospectively or prospectively to compensate for the motion on the image; volumes were realigned to the middle volume to correct artifact in this experiment. Because the ALFF measures required a constant time course for frequency and power analysis, they could not be run on scrubbed data since volumes with excessive movement would be removed [27]. Therefore, twenty participants’ data without more than 1 mm maximum displacement in x, y, or z and 1°of angular motion during the whole fMRI scans were accepted. After head-motion correction, the functional images were registered to the corresponding T1-weighted structural images, normalized to the Montreal

Neurological Institute (MNI) standard brain with a resampling voxel of 3mm×3mm×3mm, and smoothed with a Gaussian kernel of full width at half maximum (FWHM) 8 mm. 2.4 ALFF analysis and Statistical analysis After pre-processing, the resting-state fMRI were imported into REST software package (http://www.resting-fmri.Sourceforge.net) for further processing. Firstly, linear regression was applied to remove the influences of linear trends and the waveform of each voxel was band-pass filtered (0.01 – 0.08 Hz) subsequently to reduce the influences of low-frequency drift and high-frequency noise [28]. The square root was calculated at each frequency of the power spectrum and the averaged square root was obtained across 0.01-0.08Hz at each voxel. This averaged square root was taken as the ALFF. Each voxel have a value of ALFF. Then, subject-level ALFF maps were transformed into Z-scores in MNI152 standard space to create standardized subject-level maps for group statistical analysis. The power of slow fluctuations differs among different brain regions and subjects, and thus may act as a marker of individual or group dysfunction. The study aims to research ALFF significant difference of between Pre- and Post- 3DTVgroup and 2DTV group respectively. The correction thresholds were determined by Monte Carlo simulations with the program AlphaSim in REST software package [29]. The paired t-test results were presented by REST and discussed in detail below. 2.5 Modularity analysis After pre-processing, the fMRI data sets were divided into 90 regions of interest (ROIs) based on the automated anatomical labeling (AAL) atlas. The mean time series of each region was calculated and 90×90 correlation matrix was then obtained by calculating Pearson correlation coefficients between two regional time series. Finally, hierarchical clustering algorithm was used to

analyze the changes of brain community structure before and after watching 2D/3DTV. Modularity is an important feature to the whole brain functional network, which represents the degree of a brain functional network that can be subdivided into some groups [30]. The modularity Q is defined as follows: N d  l Q    s  ( s )2  2L  s 1  L

Where N and L are the total number of modules and edges in the network, respectively. ls denotes the number of edges between nodes in module s . d s is the sum of the degrees of all nodes in module s . 3. Results We set p<0.01, p<0.05 and p<0.1, but no brain regions exhibited significant differences in the Post session when compared with the Pre session in the 2DTV group for both general ALFF maps and specific frequency bands(Slow5 and Slow4). Therefore, watching 2DTV has no significant effect on brain spontaneous activity. The results in 3D group are discussed in detail in below. 3.1 Results for general ALFF To study the variation of spontaneous brain slow fluctuations caused by watching 3DTV, paired t-test was conducted on the ALFF maps of Pre and Post sessions in the 3D group. Monte Carlo simulations can create multiple simulated null data sets based on user-specified parameters, and further create a distribution of cluster sizes, from which the cluster size corresponding to a desired corrected P can be read off. According to the Monte Carlo simulations, a corrected P-value of p<0.05 can be achieved through the combination of each voxel threshold of p<0.05 and a cluster size of at least 950 voxels. The detail explanation about Monte Carlo simulations can be found in reference

[31]. Statistical parametric images were displayed on the MNI standardized 3D brain template and anatomical indicators were used to mark brain areas with significant differences in Fig.1. The real left and right side of the brain is opposite to the figure. R-value scale was shown on the right and red indicate that 3D-Post showed significantly greater increases than 3D-Pre. Sizes, brain areas, MNI coordinates and related intensity of statistically significant clusters are also recorded in Table 1.

Fig.1. Significant differences of ALFF map in the 3D-Post as compared with the 3D-Pre. Table 1 Changed ALFF regions in the 3D-Post as compared with the 3D-Pre MNI coordinate（mm） Brain Reagions

Peak Activation

X

Y

Z

-48

6

42

Brodmann Areas

Number of Voxels

Strength(t)

Temporal_Sup_L Temporal_Mid_L 6.78

4/22

Postcentral_L Precental_L 1438 Cingulum_Ant_L Cingulum_Mid_L Frontal_Sup_Medial_L

-9

21

33

6.14

32

Frontal_Mid_L -30

42

12

5.89

6/8/9/10

Frontal_Sup_L Note: Temporal_Sup_L: left superior temporal gyrus; Temporal_Mid_L: left middle temporal gyrus; Postcentral_L: left postcentral gyrus; Precental_L: left precentral gyrus; Cingulum_Ant_L: left anterior cingulated; Cingulum_Mid_L: left middle cingulated; Frontal_Mid_L: left middle frontal gyrus;Frontal_Sup_L: left superior frontal gyrus Frontal_Sup_Medial_L: left medial superior frontal gyrus.

Compared with the 3DTV Pre sessions, higher ALFF regions were observed in temporal gyrus, parietal lobe, cingulate cortex and frontal lobe after watching 3DTV for one hour. Specifically, these seemingly separate structures are in fact linked and typically associated with effective visual and emotional processing. 3.2 Results for ALFF of Slow5 and Slow4 As frequency bands contribute differently to the ALFF and gray matter related fluctuation amplitudes mainly occurred in 0.01-0.027 Hz (Slow5) and 0.027-0.073 Hz (Slow4), another two ALFF maps for each subject in 3D group were obtained by computing the power within these two frequency bands. Paired t-tests were conducted on the ALFF maps of Pre and Post sessions in the 3D group, Slow5 and Slow4 respectively. Monte Carlo simulations were used to control multiple comparisons, with the voxel threshold level set at p<0.05 and a cluster size of at least 51 voxels to achieve the corrected P-value. Fig.2 shows ALFF differences of Slow4 between Pre and Post sessions in 3D group with MNI coordinates. The detail information of region-related MNI coordinates is shown in Table 2. Significant variations exist in right inferior temporal gyrus (BA20/37), which receives visual information from the ventral stream as the function of object recognition, and right inferior occipital gyrus (BA18/19), which is responsible for the higher order processing of visual signals and their dispatch to other parts of the brain.

Fig.2. Significant differences of Slow4 ALFF map in the 3D-Post as compared with the 3D-Pre. Table 2 Changed regions of Slow4 ALFF in the 3D-Post as compared with the 3D-Pre MNI coordinate（mm） Brain Reagions

Peak Activation

Brodmann Areas

X

Y

Z

Temporal_Inf_R

57

-66

9

4.83

20/37

Occipital_Inf_R

49

-70

-22

2.68

18/19

Number of Voxels

Strength(t) 56

Note: Temporal_ Inf_R: right inferior temporal gyrus; Occipital_Inf_R: right inferior occipital gyrus.

ALFF differences of Slow5 between Pre and Post sessions in 3D group are displayed in Fig.3 and regions exhibiting significant differences are depicted in Table 3. Compared with the 3D-Pre, lower ALFF was not found in any brain region in the 3D-Post and higher ALFF was observed in frontal lobe including left superior frontal gyrus (BA9/10) and left superior medial frontal gyrus (BA32).

Fig.3. Significant differences of Slow5 ALFF map in the 3D-Post as compared with the 3D-Pre. Table 3 Changed regions of Slow5 ALFF in the 3D-Post as compared with the 3D-Pre MNI coordinate（mm） Brain Reagions

Peak Activation

Brodmann Areas

X

Y

Z

Frontal_Sup_L

-21

48

18

3.38

9/10

Frontal_Sup_Medial_L

-9

58

17

2.16

32

Number of Voxels

Strength(t) 90

Note: Frontal_Sup_L: left superior frontal gyrus; Frontal_Sup_Medial_L: left superior medial frontal gyrus.

3.3 Results for modularity The variations of the subject modularity before and after watching 2D/3DTV are shown in Fig.4(a) and (b) using two curves respectively. As shown in Fig.4(a) and (b), the value of modularity was slightly reduced after watching 2DTV while the value of modularity was slightly increased after watching 3DTV. Paired t-test results showed that after watching 2DTV, significant differences were

noticed in the cost of 0.10, 0.14 and 0.22 (p<0.05). After watching 3DTV, significant differences were noticed in the cost of 0.86 (p<0.05). Fig.5(a) and (b) respectively represents the modular of brain functional network changes before and after watching 3DTV at a cost value of K = 0.15, which is representative of the small world regime. These brain function networks were comprised of six main modules. Each node with the same color belongs to the same module in the brain function network. Module I (marked with yellow) consists of many regions, mostly from bilateral occipital lobe and frontal lobe, which are the community of the visual conducting pathway. After watching 3DTV, the yellow module integrated more brain regions for information processing, including bilateral supplementary motor area (SMA), dorsolateral superior frontal gyrus and cingulated cortex. The basal ganglia, including putamen, caudate and globus pallidus, was differentiated and reorganized after watching 3DTV.

(a)

(b)

Fig.4.The modularity changes of brain functional network before and after watching 2D/3DTV. (a) 2D modularity value, (b) 3D modularity value (‘*’represent p<0.05)

(a)

(b) Fig.5. The community structure of brain functional network before and after watching 3DTV. (a) Before watching 3DTV, (b) After watching 3DTV 4. Discussion

In the present study, visual and mental fatigue induced by long time watching of 3DTV has been investigated by using EEG and task related fMRI. Here, however, we adopted slow fluctuations signal in resting brain to assess the influence of watching 3DTV. We have found differences between Pre and Post sessions in 3DTV group throughout the subregions of temporal gyrus, parietal lobe, cingulate cortex and frontal lobe, while no significant changes among the 2DTV group. Increased ALFF values were observed in Post sessions of 3DTV group in both Slow5 and Slow4, suggesting the existence of compensation process to maintain normal brain functional and alertness after prolonged watching task and fatigue, which is consistent with findings from previous attention workload and muscle fatigue studies [32-33]. As the difference was only found in 3DTV group, we could ascribe the fatigue or dysfunction to the stereo-vision. Moreover, related brain peak regions have also indicated the same. Typical ALFF analysis has shown significant variations among temporal gyrus, parietal lobe, cingulate cortex and frontal lobe during the one-hour watching of 3DTV. Frontal lobe includes middle frontal gyrus (BA9/BA10) and superior frontal gyrus (BA6/8), in which BA6 and BA8 form the premotor cortex and provide sensory guidance of movement. The frontal eye field in BA8 played an important role in the control of visual attention and eye movement [34]. Previous studies have found that BA9 and BA10 played a role in the planning high-level executive functions and decision-related processes combined with BA32 located in cingulate cortex [35-36]. In addition, medial frontal received and sent widespread connections as part of network with higher cognitive functions. According to the explanation on changed brain regions, we can conclude that stereo-vision requires full play of vision, three-dimensional sense as well as recognition functions to ensure the accuracy and celerity.

Therefore, prolonged watching of 3DTV will affect not only viewers’ vision, but also their recognition functions. The results supported in Slow4 and Slow5 indicate that their blood-oxygenation level signals are distinct entities and the characteristics of Slow5 are similar with typical ALFF. Results in Slow5 further demonstrate viewers’ vision and recognition dysfunction. Meanwhile, variations of ALFF in Slow4 have mainly been reflected in right inferior temporal gyrus (BA20/37) and right inferior occipital gyrus (BA18/19) cortex in occipital lobe. Previous studies have found that functional connectivity strengths at Slow4 were always weaker than those at Slow5, which agrees with our findings [37]. BA19 has been noted to receive inputs from the retina via the superior colliculus and pulvinar, which may contribute to blindsight. BA18 is responsible for the interpretation of images and it is combined with BA19 to form the extrastriate (or peristriate) cortex (BA18/BA19) [38]. For humans with normal sight, extrastriate cortex is a visual association area with feature-extracting, shape recognition, attention, and multimodal integrating functions. The extrastriate cortex receives the primary visual information and conducts advanced higher order processing and then dispatches processed information to other brain regions [39-40]. Moreover, our previous study has shown that variations in BA18 and BA19 are related to visual fatigue [14]. In addition, the inferior temporal lobe, which includes BA20 and BA37, is located in the dorsal stream and associated with “what” visual pathway [41]. They appear to be involved in high-level visual processing of complex stimuli, object perception and recognition [42]. The community structure of brain functional network describes the brain functional integration among different brain regions. Fig. 4 indicates significant changes of the modularity of the brain functional network after watching 2D/3DTV. The modularity of brain functional network increases

after watching 3DTV in order to process the more complex and unfamiliar stereo images. Fig. 5 shows that after watching 3DTV, brain network community structure has experienced significant recombination and differentiation. Specifically, the visual conducting pathway requires more brain regions to participate in information processing, including bilateral supplementary motor area (SMA), dorsolateral superior frontal gyrus and cingulated cortex. Previous studies have shown that brain regions related to visual information processing under task state are divided into different communities. Among which, the visual information is transmitted from primary visual cortex to dorsal stream pathway and ventral stream pathway; dorsal stream pathway originates from V1 (striate cortex), transmits into dorsal medial area and middle temporal area, and finally reaches inferior parietal lobule. Dorsal stream pathway, usually known as visual space pathway, participates in the process of object’s spatial location information and related motor control. Whittingstall [43] has found that dorsal stream pathway, extrastriate, parietal and prefrontal areas, as well as ventral stream pathway (temporal and lingual areas) are activated by analyzing fMRI and dMRI under the stimulation of visuospatial imagery. Meanwhile, posterior cingulate cortex serves as an anatomical hub linking activity in occipital, parietal and temporal areas. Liu [44] has studied the application of binocular parallax in stereoscopic vision. Results have indicated the importance of dorsal stream pathway in processing the stereoscopic vision and the activated brain regions include gyrus lingualis and sulci occipitales laterals. This paper reveals the importance of the visual pathway along the occipital and temporal lobe in the processing of the stereoscopic vision by utilizing brain network community partitioning before and after watching 3DTV, which agrees with the previous research. Meanwhile, basal ganglion has differentiated and recombined with certain brain regions including bilateral superior temporal gyrus. Studies have shown that brain fatigue mostly originates

from the nervous centralis’ disability to execute the cognitive task and both basal ganglion and cortex area are essential in fatigue regulation [45]. Dopamine is a neurotransmitter related to basal ganglion and is able to regulate central nervous system fatigue through the reward pathway. Its concentration decreases as the brain experiences fatigue [46]. Combining the result of ALFF and modularity, this paper found that brain fatigue caused by long period watching of 3DTV is closely related to the basal ganglion’s regulation of gyrus temporalis superior. The fact that the community of the visual pathway contains most temporal lobe indicates that the visual conducting pathway-temporal lobe- basal ganglia loop is the one that regulates brain fatigue. This study demonstrates the difference of resting brain slow fluctuations after watching 3DTV for one hour. The main findings are as follows: while watching 2DTV does not significantly change brain spontaneous activity, watching 3DTV is able to make significant differences in both vision and recognition functions due to the compensation for fatigue. Results from ALFF and the changes of brain module further reflected that the viewer’ dysfunction is related to prolonged watching of 3DTV. The basal ganglia neurons play crucial roles in the physiological mechanism of brain fatigue. 5. Conclusions The study validates the potential risk of 3DTV to viewers’ health, especially to the vision and recognition functions. Changes from both ALFF and brain module have shown that long-time watching of 3DTV can lead to dysfunction. Therefore, prolonged 3DTV watching should be avoided especially for developing teenagers. We have applied the ALFF technique to examine the hypothesis about visual and mental fatigue underlying watching 3DTV and discussed characteristic patterns of ALFF and modularity changes in the 3D group. We conclude that basal ganglia neurons play crucial roles in the physiological mechanism of brain fatigue. Meanwhile, the ALFF technique is able to

reveal the intrinsic changes caused by visual fatigue and provide a foundation for 3DTV fatigue assessment. However, Lambooij [47] found that participants with moderate binocular status (MBS) in 3D conditions showed clinically meaning full changes in fusion range, experienced more visual discomfort, and performed worse on the reading task comparing with good binocular status（MBS）. So, the classification of participants should also be considered in the subsequent experiments. Additionally, quality of 3D videos, spatial perceptual information (SI), temporal perceptual information (TI), the depth spatial indicator (DSI) and the depth temporal indicator (DTI) should also be quantified as doing brain functional research in the future work [48-49]. Acknowledgements This study was supported by the National Natural Science Foundation of China (Grant No. 61773205, 61171059).

References [1] Patterson R. Review paper: Human factors of stereo displays: An update. Journal of the Society for Information Display, 2009. 17(12): 987-996. [2] Lambooij M, IJsselsteijn W, Fortuin M, et al. Visual Discomfort and Visual Fatigue of Stereoscopic Displays: A Review. Journal of Imaging Science and Technology, 2009. 53(3): 030201-1-030201-14. [3] Watanabe H, Ujike H. 3D visual fatigue: Interaction between ranges of binocular disparity and interocular rotational difference[C], Consumer Electronics (GCCE), 2014 IEEE 3rd Global Conference on. IEEE, 2014. [4] Huang Y, Liu Y, Zou B, et al. Study of objective evaluation indicators of 3D visual fatigue based on RDS related tasks[J]. Proceedings of SPIE - The International Society for Optical Engineering, 2015, 9391:93910V-93910V-7. [5] Matthieu U, Marcus B, and Patrick L C. How visual fatigue and discomfort impact 3D-TV quality of experience: a comprehensive review of technological, psychophysical, and psychological factors. Annals of Telecommunications, 2013, 68: 641-655. [6] Chen C, Li K, Wu Q, et al. EEG-based detection and evaluation of fatigue caused by watching 3DTV[J]. Displays, 2013, 34(2):81-88. [7] Yong J J, Kim D, Sohn H, et al. Subjective and objective measurements of visual fatigue induced by excessive disparities in stereoscopic images[C], IS&T/SPIE Electronic Imaging. International Society for Optics and Photonics, 2013: 86480M-86480M-8. [8] Atallah P, Pelah A, Wilkins A. Visual stress symptoms from stereoscopic television[C]// International Conference on 3d Imaging. IEEE, 2012:1-7.

[9]Chen C, Wang J, Li K, et al. Assessment visual fatigue of watching 3DTV using EEG power spectral parameters[J]. Displays, 2014, 35:266–272. [10] Brunnstro¨m K, Wang K, Tavakoli S, et al. Symptoms analysis of 3D TV viewing based on Simulator Sickness questionnaires. Quality and User Experience, 2017, 2(1): 1-15. [11] Yu J H, Lee B H, Kim D H. EOG based eye movement measure of visual fatigue caused by 2D and 3D displays[C]//Biomedical and Health Informatics (BHI), 2012 IEEE-EMBS International Conference on. IEEE, 2012: 305-308. [12] Cho S H, Kang H B. An Analysis of Visual Discomfort Caused by Watching Stereoscopic 3D Content in Terms of Depth, Viewing Time and Display Size[J]. Journal of Imaging Science and Technology, 2015, 59(2): 20503-1-20503-16. [13] Chen C, Wang J, Li K, et al. Visual fatigue caused by watching 3DTV: an fMRI study[J]. Biomedical engineering online, 2015, 14(Suppl 1): S12. [14] Gustavo D, Adrián P A, Dante M, et al. Resting-state functional connectivity emerges from structurally and dynamically shaped slow linear fluctuations.[J]. Journal of Neuroscience the Official Journal of the Society for Neuroscience, 2013, 33(27):11239-11252. [15] Zou Q H, Zhu C Z, Yang Y, et al. An improved approach to detection of amplitude of low-frequency fluctuation (ALFF) for resting-state fMRI: Fractional ALFF[J]. Journal of Neuroscience Methods, 2008, 172(1):137-141. [16] Zang Y F, Yong H, Zhu C Z, et al. Altered baseline brain activity in children with ADHD revealed by resting-state functional MRI[J]. Brain & Development, 2007, 29(2):83-91.

[17] Chaogan Y, Dongqiang L, Yong H, et al. Spontaneous Brain Activity in the Default Mode Network Is Sensitive to Different Resting-State Conditions with Limited Cognitive Load[J]. Plos One, 2009, 4(5):e5743. [18] Hu S, Chao H A, Zhang S, et al. Changes in cerebral morphometry and amplitude of low-frequency fluctuations of BOLD signals during healthy aging: correlation with inhibitory control[J]. Brain Structure & Function, 2013, 219(3):1-12. [19] Xi-Nian Z, Adriana D M, Clare K, et al. The oscillating brain: complex and reliable.[J]. Neuroimage, 2010, 49(2):1432-1445. [20] Wei L, Duan X, Zheng C, et al. Specific frequency bands of amplitude low-frequency oscillation encodes personality[J]. Human Brain Mapping, 2014, 35(1):331–339. [21] Chen Y C, Xia W, Luo B, et al. Frequency-specific alternations in the amplitude of low-frequency fluctuations in chronic tinnitus[J]. Frontiers in Neural Circuits, 2015, 9(1):67. [22] Rongjun Yu, Yi-Ling Chien, Wang H L S, et al. Frequency-specific alternations in the amplitude of low-frequency fluctuations in schizophrenia.[J]. Human Brain Mapping, 2014, 35(2):627-637. [23] Biswal B, Zerrin Yetkin F, Haughton V M, et al. Functional connectivity in the motor cortex of resting human brain using echo ‐ planar mri[J]. Magnetic Resonance in Medicine, 1995, 34(4):537-541. [24] Onoda K, Yamaguchi S. Small-worldness and modularity of the resting-state functional brain network decrease with aging.[J]. Neuroscience Letters, 2013, 556(1):104-108. [25] Carbonell F, Naganosaito A, Leyton M, et al. Dopamine precursor depletion impairs structure and efficiency of resting state brain functional networks.[J]. Neuropharmacology, 2014, 84(1):90-100.

[26] Li J, Chen C, Liu Y, et al. Small-world brain functional network altered by watching 2D/3DTV[J]. Journal of Visual Communication & Image Representation, 2016, 38(C):433-439. [27] Power J D, Barnes K A, Snyder A Z, et al. Spurious but systematic correlations in functional connectivity MRI networks arise from subject motion.[J]. Neuroimage, 2012, 59(3):2142–2154. [28] Tian L, Ren J, Zang Y. Regional homogeneity of resting state fMRI signals predicts Stop signal task performance[J].Neuroimage, 2012, 60(1): 539-544. [29] Tian L, Jiang T, Wang Y, et al. Altered resting-state functional connectivity patterns of anterior cingulate cortex in adolescents with attention deficit hyperactivity disorder[J]. Neuroscience letters, 2006, 400(1): 39-43. [30] Newman M E, Girvan M. Finding and evaluating community structure in networks.[J]. Physical Review E Statistical Nonlinear & Soft Matter Physics, 2003, 69(2 Pt 2):026113-026113. [31] Matthew D Mithchell. Validation of the Summary ROC for Diagnostic Test Meta-Analysis: A Monte Carlo Simulation [J]. Academic Radiology,2003, 10(1):25-31. [32] Gui D, Xu S, Zhu S, et al. Resting spontaneous activity in the default mode network predicts performance decline during prolonged attention workload[J]. Neuroimage, 2015, 120:323-330. [33] Jiang Z, Wang X F, Kisiel-Sajewicz K, et al. Strengthened functional connectivity in the brain during muscle fatigue.[J]. Neuroimage, 2012, 60(1):728–737. [34] Schall J D. On the role of frontal eye field in guiding attention and saccades.[J]. Vision Research, 2004, 44(12):1453–1467. [35] Neuroscience J O C. Functional specialization within the medial frontal gyrus for perceptual go/no-go decisions based on "what," "when," and "where" related information: an fMRI study.[J]. Journal of Cognitive Neuroscience, 2005, 17(7):981 - 993.

[36] Yeung N, Ralph J, Nieuwenhuis S. Drink alcohol and dim the lights: the impact of cognitive deficits on medial frontal cortex function.[J]. Cognitive Affective & Behavioral Neuroscience, 2007, 7(4):347-355. [37] Xue S W, Li D, Weng X C, et al. Different neural manifestations of two slow frequency bands in resting fMRI: A systemic survey at regional, inter-regional, and network levels.[J]. Brain Connectivity, 2014, 4. [38] Merriam E P, Colby C L. Active Vision in Parietal and Extrastriate Cortex[J]. Neuroscientist, 2005, 11(5):484-93. [39]Parker A J. Binocular depth perception and the cerebral cortex[J]. Nature Reviews Neuroscience, 2007, 8(5): 379-391. [40]Orban G A. Higher order visual processing in macaque extrastriate cortex.[J]. Physiological Reviews, 2008, 88(1):59-89. [41] Sedda A, Scarpina F. Dorsal and ventral streams across sensory modalities[J]. Neuroscience Bulletin, 2012, 28(3):291-300. [42] Schacter, Daniel L.; Gilbert, Daniel T.; Wegner, Daniel M. (2010). Psychology (2nd ed.). New York: Worth Publishers. ISBN:9781429237192. [43] Whittingstall K, Bernier M, Houde J C, et al. Structural network underlying visuospatial imagery in humans[J]. Cortex, 2014, 56:85-98. [44] Jiajia Liu, Fei Wang, Wei Hong, et al. Preliminary fMRI of human brain activation under stereoscopic vision[J]. Journal of Beijing Normal University (Natural Science), 2013(6):640-644. [45] Lorist M M, Jolij J. Trial history effects in stroop task performance are independent of Top-Down control. PLos One, 2012, 7 (6):e39802

[46] Lazarus M, Zhili Huang, Jun Lu, et al. How do the basal ganglia regulate sleep-wake behavior. Trends Neurosci, 2012, 35(12):723~732 [47] Lambooij M, Fortuin M, IJsselsteijn W, et al. Measuring visual fatigue and visual discomfort associated with 3-D displays. Journal of the SID, 2010, 18(11): 931-943 [48] Wang K, Barkowsky M, Brunnström K, et al. Perceived 3D TV transmission quality assessment: Multi-laboratory results using Absolute Category Rating on Quality of Experience scale. IEEE Transactions on Broadcasting, 2012. 58(4): 544-557 [49] Urvoy M, Barkowsky M , Cousseau R, et al. Subjective video quality assessment database on coding conditions introducing freely available high quality 3D stereoscopic sequences. in Proc. Fourth International Workshop on Quality of Multimedia Experience, 2012. Yarra Valley, 5-7 July 2012.

Highlights (for review)

Highlights  Using the amplitude of low frequency fluctuation (ALFF) to study brain spontaneous activity induced by watching 2D/3DTV.  Applying the community partitioning algorithm in brain functional network by analyze functional magnetic resonance imaging.  This research based on resting-state fMRI reveals initial mechanism of visual and mental fatigue underlying watching 3DTV.  Prolong 3DTV watching should be avoided because of the potential risk of 3DTV to viewers’ health.