Functional magnetic resonance imaging based on Chinese tasks to protect language function in epileptics

Abstract Objective To evaluate the efficacy of functional magnetic resonance imaging (fMRI) based on Chinese tasks to protect the language function in epileptics. Materials and Methods A total of 34 native Chinese patients with epilepsy were enrolled and examined with BOLD‐fMRI scan based on six Chinese tasks. The epileptics were randomly divided into the control group (n = 15) and the experimental group (n = 19). The control group underwent the hollowing and multiple subpial transection operation only based on intraoperative EEG, while the experimental group was under notification of task‐state fMRI results in addition. Whereafter, the language ability of patients was evaluated by ABC assessment. Results The brain regions related to Chinese function activated by different tasks were remarkably distinct and mainly concentrated in the temporal lobe and frontal lobe. In ontoanalysis, the activation signals of the fusiform gyrus, parahippocampal gyrus, hippocampus, and precentral gyrus were generally low or even could not be detected. Unlike ontoanalysis, group analysis showed that the main effect regions of AN and PN task were in right superior temporal gyrus. The main effect regions of FF and VFC task were in right middle temporal gyrus. The main effect region of SF task was in left superior temporal gyrus. The main effect region of VFL task was in right middle frontal gyrus. The ABC assessment score of the control group 6 months after surgery was significantly lower than that 1 week before surgery (p < .05), while there was no significant difference in the experimental group, and the score of the experimental group was higher than that of the control group. Conclusion In the surgical treatment of epilepsy, a personalized surgical plan, based on task‐state fMRI and intraoperative EEG, can be developed according to the difference of activation areas to protect the language function and improve the quality of life in postoperative patients.


| ABC assessment
The Aphasia Battery of Chinese (ABC) assessment, which was established by the Neuropsychology Department of the First Affiliated Hospital of Beijing Medical University (Gao et al., 1992), was used to quantitatively assess the language ability of patients, including verbal fluency, repetition, comprehension, naming, reading, and writing. The total score is 750. The language ability of patients was evaluated 1 week before surgery and 6 months after surgery with different versions of the battery of tests to avoid test-retest effects.

| Language task design
All tasks were arranged by the professional stimulus task design program E-prime based on previous studies (Ci et al., 2016;Pauselli et al., 2018;Wang et al., 2016) and displayed onto the screen via a projector. Language stimulus software was Visual&Audio Stimulation System for fMRI developed by Sinorad medical company. The experiment was planned to use block design, with stimulus (Task) and rest alternating. According to the characteristics of Chinese and the experimental elements of task-mode, six experimental modules were designed as follows and shown as Figure 1. All participants watched pictures through a small mirror on the head coil for visual stimulation, heard the voice through a magnetic resonance compatible speech system for auditory stimulation.

| Auditory naming (AN)
There were five cycles in AN. Each cycle included a 30 s stimulus task and two control tasks, namely the reverse voice (Reversed, Re) and "*" for 15 s, respectively. During the stimulus task, participants were asked to name the items based on description they had heard. For example, stimulus task was "quacking", whose answer should be "duck". During retask, participants only could hear the voice, but not know voice content, and were required to number "1, 2." When "*" appeared on the screen, rest started.

| Picture naming (PN)
There were five cycles in PN, and each cycle consisted of four modules. Line-like black-and-white pictures, which were common objects or animals and plants on white background (Snodgrass atlas, provided by the institute of psychology, Chinese Academy of Sciences), were displayed for 30 s. Participants were asked to name the items in the picture. Scrambled faces (SF) pictures were displayed for 15 s on the screen, scrambled pictures were displayed for 15 s, and rest pictures "*" were displayed for 15 s. Participants were asked to number "1, 2" over Scrambled faces (SF) pictures and scrambled pictures and rested when saw "*".

| Free fluency (FF)
There were five cycles in FF, and each cycle consisted of 3 modules. During stimulus task, Chinese characters were displayed for 30 s on the screen, and participants were asked to list as many words associated with the Chinese characters as possible until the next task began. For example, Chinese characters "glass" could be associated with "Windows, houses, furniture… ". In the next 15 s, participants were asked to just read and repeat (RR) the words that appeared on the screen, but not to think. For example, "science" was read and repeated once by participants. When "*" appeared, participants rested.

| Semantic fluency (SF)
There were five cycles in SF, and each cycle consisted of 3 modules.
During the first 30 s of stimulus task, participants were asked to list as many words belonging to the Chinese characters on the screen as they can. For example, the word "fruit" could be listed as "apple,

TA B L E 1
The general information of 34 patients banana, peach… ". In the next 15 s, participants were asked to read and repeat (RR) the words appeared on the screen, such as "insect".

| Verbal fluency letter (VFL)
There were five cycles in VFL, and each cycle consisted of 3 modules. According to the given initials, the initials and different finals could be matched to form Pinyin during the first 30 s of stimulus task. For example, initials "P" could be divided into "pa, po, pin…".
In the next 15 s, participants were asked to read and repeat (RR) the words appeared on the screen. When "*" appeared, participants rested.

| Verbal fluency characters (VFC)
There were five cycles in VFC, and each cycle consisted of 3 modules. According to the given Chinese characters, participants were asked to combine words with the meaning of the Chinese character during the first 30 s of stimulus task. For example, the Chinese character "water" could be grouped as "flood, sewage, boiling water…". In the next 15 s, participants were asked to read and repeat (RR). When "*" appeared, rest started.

| MRI scan
All the participants were informed about the significance of the experiment design to strive for positive coordination and trained to keep their heads still while performing the above tasks. Before examination, the participants also rested quietly to adapt to the MR scanning environment.
All data were acquired using a 3.0 T Philips NMR with an 8-channel array head coil for reception and the body coil for transmission.
The blood oxygenation level-dependent (BOLD) signals collection was performed with single-shot echo-planar imaging(EPI), and and the signal-to-noise ratio of frontotemporal lobe image also was increased.

| fMRI data analysis
The data were analyzed using SPM8 software (Wellcome by 2nd level single sample t test installed in statistic parametric mapping (SPM8). All data in the same group were calculated and determined as activation, and the activation level was defined as p < .001 after correct inspection. The activation region of language task was more than 10 voxels, otherwise, and it was identified as spike and excluded.

| Surgery treatment
A total of 34 patients were randomly grouped before surgery,

| Statistical analysis
The fMRI data were analyzed using SPM8 software described above.
Statistical analysis was carried out using the software SPSS 20.0.
The measurement data were expressed as mean ± standard deviation, and the comparison between the two groups was analyzed by t test. The counting data were expressed as percentage or ratio, and the comparison between groups was analyzed by the chi-square test. Single-factor analysis of variance was done to compare the data obtained. Significant differences were considered as p < .05.

| Comparison of general information
The 34 patients (average age 29.7 years, range 18-59) were accomplished surgery and experiments. There was no difference in age, gender, handedness, and handedness scores of the two groups (p > .05), which were comparable. The comparison of general information in the two groups is shown in Table 2.

| Ontoanalysis
The distribution of effective activation regions in 34 patients (intraobject analysis) is shown in

| Group analysis
The differences in the above distribution in 3.2 were analyzed with full factors, and the result of group analysis is shown in

| language ability
As shown in Table 5, the ABC assessment score of the control group 6 months after surgery was significantly lower than that 1 week before surgery (p < .05), while the ABC assessment score of the experimental group was 503.1 ± 76.7 6 months after surgery, which had no significant difference with that 1 week before surgery (p > .05).
Besides, although there was no difference in the ABC assessment score of the two groups 1 week before surgery and 6 months after

TA B L E 2
The comparison of general information in groups surgery (p > .05), the score of the experimental group was higher than that of the control group.

| D ISCUSS I ON
Epilepsy is one of the most common diseases of the central nervous system, with over 50 million patients worldwide. There are about 9.84 million epilepsy patients in China and about 236,000 in Sichuan province. The incidence rate is higher in rural areas, and up to 30% are intractable epilepsy (Song et al., 2017). Epilepsy not only causes harm to the physical and mental health of patients but also brings a heavy economic burden to the family and society.
Epilepsy, which is difficult to treat by drugs, preferentially occurs in the temporal lobe (Perucca & Tomson, 2011). The main treatment is surgery, such as anterior temporal lobectomy and modified hollowing at present. And it is worth noting that the temporal lobe, amygdala, and hippocampus structures play an important role in cognitive functions, and the damage of them would result in the corresponding language, memory, and other cognitive function lesions (Campo et al., 2013;Hauptman et al., 2012;Noppeney et al., 2005).
Besides, pathological changes always are inconsistent with the focus (Clarke et al., 1996). Therefore, most of patients would appear with different levels of language dysfunction after surgery, and it is urgently needed to prevent these complications in clinical.
The outcome of surgery mainly depends on the accurate localization and excision of epileptogenic focus (Ren et al., 2015). Traditional intraoperative EEG could roughly locate the epileptogenic focus because of its lower special resolution, and it could not effectively mark the functional language regions of patients in the brain (Blumer et al., 1998). In recent years, the emergence of BOLD-fMRI TA B L E 3 Global maxima activation (GMA) of 34 patients---first level (intra-object analysis) technology, whose signals are the hemodynamic response of a large number of nerve cells, provides the possibility of accurate localization of epileptogenic focus. Based on its advantages of noninvasive, high sensitivity and resolution, BOLD-fMRI technology has been used to study epileptic brain networks. In our paper, BOLD-fMRI and intraoperative EEG were combined to realize the complementary.
Our results showed that different brain regions were activated by different tasks, and the activated regions of the same task were different in each individual, and the regions of different tasks overlapped with each other (Table 3). The distribution of significantly activated regions of the same task was studied with group analysis with full factors: The main effect regions of AN and PN task were in right superior temporal gyrus. The main effect regions of FF and VFC task were in right middle temporal gyrus. The main effect region of SF task was in left superior temporal gyrus. The main effect region of VFL task was in the right middle frontal gyrus (Table 4). The distribution of significantly activated regions mainly was concentrated in the temporal lobe and frontal lobe, which was consistent with previous research (Duncan, 2010). The results of group analysis was different from those of ontoanalysis, which could not represent the distribution of individuals and was not suitable for guiding surgery. And our experience has shown that the surgical plans based on previous group analysis will damage more functional areas, which may be an important reason of side effect in routine epilepsy surgery. The usual practice is that the regions with the highest degree of activation should be protected firstly. So in the best avoiding scheme, the hollowing and transections of nerve fiber should be performed in the nonactivated regions beyond the intersection (Binder et al., 2011;Mahvash et al., 2014), and the activated regions should be preserved to the greatest extent. Among different individuals, the distribution of significantly activated regions of the same task was different, so it is necessary to combine fMRI and intraoperative EEG to reduce the damage of functional language regions and develop personalized programs.
Unlike Western languages, most Chinese characters are pictographic characters (80%). They are the most representative graphic characters and have a complex block structure composed of multiple strokes. Therefore, the process of Chinese characters also requires more brain regions to participate in (Li et al., 2018). According to the particularity of Chinese, our experiment was pertinently designed to assess the functional language regions related to viewing, listening, and speaking. This paper improved the task of semantic association by exploring the relevant brain regions where the circuit for language was located from the perspective of phonetic and semantic elements (Li et al., 2018;Limotai & Mirsattari, 2012;Tie et al., 2009). Our results showed that the activated regions of the same task were not isolated specific points, but multiple brain regions in piece-like distribution. Besides, the regions were in a relatively concentrated range (Perucca & Tomson, 2011), which performed the compound function of multiple brain regions (Table 3).
This also proved the process of Chinese characters was the coordination of multiple brain regions, which was consistent with the results of other studies (Li et al., 2018). The ABC assessment score of the control group 6 months after surgery was significantly lower than that 1 week before surgery (p < .05), while there was no significant difference in the experimental group, and the score of the experimental group was higher than that of the control group. This means that the hollowing and modified multiple subpial transections were successfully performed based on task-state fMRI and intraoperative EEG, and the language function of the experiment group was better protected.

| CON CLUS ION
Language functional task-state brain magnetic resonance imaging is a new way to explore the language damage caused by epilepsy. In the surgical treatment of epilepsy, a personalized surgical plan, based on task-state fMRI and intraoperative EEG, can be developed according to the difference of activation areas to protect the language function and improve the quality of life in postoperative patients.

ACK N OWLED G M ENTS
We wish to thank the staff in the Department of Radiology of the

CO N FLI C T O F I NTE R E S T
There are no conflicts of interest.

AUTH O R CO NTR I B UTI O N
Wang Peng and Jiang Rui conceived the ideas; Du Feizhou and Li Jianhao collected the data;Wang Peng and Du Feizhou analyzed the data; Yu Hongmei and Tang Chencheng led the writing.