Multimodal characterization of the semantic N400 response within a rapid evaluation brain vital sign framework

Multimodal characterization of the semantic N400 response within a rapid evaluation brain vital... Background: For nearly four decades, the N400 has been an important brainwave marker of semantic processing. It can be recorded non-invasively from the scalp using electrical and/or magnetic sensors, but largely within the restricted domain of research laboratories specialized to run specific N400 experiments. However, there is increas- ing evidence of significant clinical utility for the N400 in neurological evaluation, particularly at the individual level. To enable clinical applications, we recently reported a rapid evaluation framework known as “brain vital signs” that successfully incorporated the N400 response as one of the core components for cognitive function evaluation. The current study characterized the rapidly evoked N400 response to demonstrate that it shares consistent features with traditional N400 responses acquired in research laboratory settings—thereby enabling its translation into brain vital signs applications. Methods: Data were collected from 17 healthy individuals using magnetoencephalography (MEG) and electroen- cephalography (EEG), with analysis of sensor-level effects as well as evaluation of brain sources. Individual-level N400 responses were classified using machine learning to determine the percentage of participants in whom the response was successfully detected. Results: The N400 response was observed in both M/EEG modalities showing significant differences to incongruent versus congruent condition in the expected time range (p < 0.05). Also as expected, N400-related brain activity was observed in the temporal and inferior frontal cortical regions, with typical left-hemispheric asymmetry. Classification robustly confirmed the N400 effect at the individual level with high accuracy (89%), sensitivity (0.88) and specificity (0.90). Conclusion: The brain vital sign N400 characteristics were highly consistent with features of the previously reported N400 responses acquired using traditional laboratory-based experiments. These results provide important evidence supporting clinical translation of the rapidly acquired N400 response as a potential tool for assessments of higher cognitive functions. Keywords: N400, ERP, MEG, Semantic language, Clinical application *Correspondence: rdarcy@sfu.ca Faculty of Applied Science, Simon Fraser University, Burnaby, BC, Canada Full list of author information is available at the end of the article © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/ publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 2 of 11 likely responsible for N400 [24], and these results are also Background supported by findings from lesion studies [25]. Measurements of brainwave activity through event- Further to its functional relevance as an indicator related potentials (ERPs) are becoming increasingly use- of neural processing in healthy individuals, the N400 ful in providing objective, physiology-based measures of response has also shown significant potential as a diag - brain function [1]. ERPs are derived from electroenceph- nostic and prognostic tool in clinical populations [4, 17, alography (EEG), and can provide information about cor- 26–33]. Studies in brain-injured patients with disor- tical electrical activity corresponding to different aspects ders of consciousness showed that the N400 response of neural processing [2, 3]. In particular, higher order was correlated with functional recovery [4]. Moreo- cognitive functions like semantic processing indexed by ver, changes in N400 response also predicted cognitive the N400 ERP are among the most promising responses decline in patients as they progressed from mild cogni- for emerging clinical applications [4–7]. The N400 tive impairment (MCI) to dementia [5, 31]. Yet despite response was first described when Kutas and Hillyard these promising findings, the use of the N400 ERP presented participants with visual sentences that either beyond the research setting has been hindered by two had a semantically related (i.e. congruent) or semantically main challenges: First, given that ERPs are produced unrelated (i.e. incongruent) ending [8]. It was observed as by averaging the neural response signals across a large a negative deflection of the incongruent relative to con - number of trials, traditional N400 studies require pro- gruent condition waveforms which peaked at approxi- longed testing paradigms [1, 34]. These paradigms are mately 400  ms latency following stimulus presentation, particularly problematic in clinical populations due to and the authors suggested that this differential was a neu - fluctuations in vigilance levels and lack of capability or ral marker of semantic language processing. motivation [30, 35]. In addition, rather than measuring In the 38  years since its initial report, the N400 only a single brain response in clinical populations (e.g. response has been studied extensively using a variety of sensation, attention, or language), there are now calls for stimulus paradigms in various healthy and clinical pop- concurrent evaluations of a spectrum of brain responses ulations [9–14]. While the initial N400 work utilized which provide a more complete profile of brain function sentence-based stimuli, subsequent studies showed that [34]. This is particularly crucial in longitudinal monitor - prime-target word pairs also successfully elicited this ing of brain function changes in clinical populations [36]. response [15, 16]. Additionally, non-language-based Under these circumstances, the traditional ERP test- stimuli such as mental arithmetic and action sequences ing paradigms may require hours to evaluate, which is have also been shown to produce the N400 response [17], impractical within most clinical settings. and the strength of this response has been found to be To assess the N400 response within a short testing time correlated with various stimulus properties [18]. Oth- while providing information about other brain function ers have demonstrated overlapping features in the tem- indicators, our group has been undertaking systematic poral and spatial characteristics of the N400 response development of rapid evaluation techniques in recent when elicited using language- as well as non-language- years. We previously demonstrated the successful evalu- based stimuli [17], with the spectral content in particu- ation of the N400 response in 100 healthy individuals lar demonstrating potential in distinguishing between using a point-of-care enabled device [34], then employed different neural processes [19]. In fact, one of the key this device to track the progress of rehabilitation therapy spectral features of the N400 response has been shown in a brain-injured patient [6]. More recently, we demon- to be a reduction in beta band oscillations when process- strated a rapid evaluation platform known as the ‘brain ing incongruent relative to congruent stimuli in semantic vital sign’ framework [37], which enables the rapid language paradigms [20]. assessment of several brain function indicators including The cortical generators of the N400 response have been the N400 (semantic language), N100 (sensory processing) investigated using numerous noninvasive imaging modal- [38] and P300 (attention orienting) [39]. The brain vital ities, such as functional magnetic resonance imaging sign framework employs a portable, low-density EEG sys- (fMRI), electroencephalography (EEG), as well as magne- tem, with automated, user-friendly software for easy clin- toencephalography (MEG). Results have revealed wide- ical applications. The testing paradigm utilizes a short, spread cortical activations across the left temporal lobe, 5-min auditory stimulus sequence in which tone and along with smaller areas of activity in the right temporal word stimuli are interlaced to maximize the number of as well as bilateral inferior frontal and parietal regions trials and signal-to-noise ratio. Results in healthy adults [11, 21–23]. Specifically, areas of the bilateral temporal showed that, not only were the target responses success- cortices (Brodmann Areas [BA] 20/21/22) and left infe- fully elicited at the individual level, but the platform also rior frontal gyrus (BA 45/47) have been shown to be key captured expected age-related changes in attention and cortical regions within the distributed language network Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 3 of 11 cognition that were undetected using conventional clini- 5 sec cal screening measures [37]. Although the rapid evaluation brain vital sign frame- work showed initial promise as a potential avenue for clinical application of the N400 ERP, the component characteristics of this rapidly elicited N400 (rN400) 5 min response have not yet been described. Given the short, complex stimulus paradigm, it is crucial to characterize Color represents type of stimulus Tone Word this response with respect to its spatiotemporal, spectral, Fig. 1 Illustration of auditory stimulus sequence of the brain vital and neuroanatomical features, and compare them with sign framework. Blocks of five tones and two words repeated 60 times for a total scan time of about 5 min. Words represent known N400 characteristics reported in studies using prime-target pairs, containing both semantic congruent (pink– more conventional approaches over the last few decades. orange) and incongruent (pink–blue) pairs. Tones (standard = green The current study utilized MEG with simultaneous EEG and deviant = black) elicit sensory (N100) and attention (P300) to investigate the temporal, spatial, spectral, and neuro- measures anatomical characteristics of the rN400 response elicited within the brain vital sign framework. We hypothesized that the rN400 response will exhibit features consistent with known characteristics of the N400 response, includ- semantically linked (incongruent condition, 50%, e.g. doc- ing: (1) increased ERP negativity and MEG signal power tor-egg) to generate the differential processing measures. for the incongruent relative to congruent condition dur- Words in both groups were balanced for characteristics ing the 300–500 ms post-stimulus interval; (2) decreased such as word frequency and length, and the words in the beta- band power for the incongruent relative to congru- semantically linked group had a minimum Cloze prob- ent condition during the same interval; and (3) increased ability of 0.8 [40]. The stimuli were recorded in a male activation of temporal and frontal cortices (BA 20, 21, 22, voice and root-mean-square normalized using Audacity 45 and 47) for processing of incongruent relative to con- software. The stimulus sequence contained 30 trials each gruent stimuli. of the congruent and incongruent conditions. Methods Participant details MEG and EEG data acquisition Seventeen (17) right-handed healthy participants with no A 151-channel CTF MEG (MEG International Services history of neurological problems or psychoactive medi- Limited, Canada) was used with concurrent 3-channel cation were recruited (22.6 ± 2.4  years, 10 males). Par- EEG, both recorded in a magnetically shielded room with ticipants were undergraduate or graduate students, had the participants in the supine position. Data were sam- normal hearing, normal or corrected-to-normal vision, pled at 1200 Hz using axial gradiometers (5-cm baseline) and were fluent in English. The study was approved by with synthetic 3rd order gradients employed for noise ethics boards at Fraser Health Authority and Simon cancellation. Continuous head position monitoring was Fraser University, and all participants provided written undertaken by three head position indicator coils located informed consent. at fiducial points (HPI, positioned at nasion, left and right pre-auricular points). EEG recordings utilized Ag/AgCl Auditory stimuli scalp electrodes placed at Fz, Cz and Pz locations, with As introduced elsewhere [37], the rapid assessment impedances kept below 5  kOhms. Four additional elec- framework utilizes a compressed auditory stimulus trodes were placed on the head corresponding to refer- sequence with interlaced tones and words to elicit brain ence (left mastoid), ground (forehead), horizontal (outer responses across four different functional domains— canthus of left eye) and vertical (supra-orbital ridge of auditory sensation (N100 ERP), attention (P300 ERP), left eye) electro-occulogram (EOG). To facilitate the and semantic language (N400 ERP)—in approximately alignment of MEG scanner and head coordinate systems, 5  min (Fig.  1). The sequence comprised 60 blocks, with the shape of the participants’ head and the 3-dimensional each block containing five tones and two words repre - position of HPI coils and EEG/EOG electrodes were senting a prime-target pair. Semantic language process- recorded using a Polhemus electromagnetic digitization ing responses were derived from conditionally averaging system prior to data collection (Polhemus Incorporated, the trials corresponding to the target word in the pair. USA). Auditory stimulation was presented binaurally Semantically linked words (congruent condition, 50%, using insert earphones, and participants were instructed e.g. doctor-nurse) were contrasted with words not to maintain visual fixation on a crosshair displayed on Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 4 of 11 the overhead screen (white cross on black background) and incongruent conditions across participants in each throughout the session. frequency [43]. This entailed the calculation of T-statistic for each time point and frequency between the congru- Data preprocessing ent and incongruent conditions in the 800  ms following Raw data for both MEG and EEG were first visually stimulus presentation. Thereafter, 1000 permutations inspected, and artifactual channels removed from further were undertaken and new T-statistic calculated for every analysis. Data were then down-sampled to 300 Hz, notch permutation leading to a null distribution against which filtered to remove frequencies corresponding to power the significance of the true T-statistic was assessed (with line (60 Hz) with its harmonics as well as HPI coils, and p < 0.05 considered to be significant). low-pass filtered to 100  Hz. Data from 2 of the 17 par - ticipants were excluded from subsequent analysis due to Neuroanatomical effects poor quality. Source level analysis was performed using SPM8 (Wel- come Trust Centre for Neuroimaging, UK) with the for- MEG analysis ward and inverse modeling steps elaborated in previously Following band-pass filtering (0.5–45  Hz), independent published work [46]. Source analysis for localizing neural component analysis (ICA) was performed with runica generators of the semantic language process was under- algorithm in EEGLAB [41] in order to remove artifact taken using minimum norm estimates (MNE) to main- from ocular, cardiac, and muscular sources. tain consistency with prior N400 studies in MEG [24, 44]. Group constraints were employed during inversion [47], Temporal effects and source reconstruction was based on trial-averaged Since head position within the MEG helmet can vary data within the entire frequency range (0.5–45  Hz) and across participants, global field power (GFP) was uti - active epoch (0–900  ms relative to stimulus presenta- lized to provide a measure of the overall activity across tion). Source-level contrast images were derived using all channels [42]. Individual-level GFP was computed for data in the 0.5–45  Hz frequency range and previously the congruent and incongruent conditions using trial- identified window of 300–500  ms. Statistical modeling averaged event-related fields. A bootstrapping approach employed a general linear model (GLM) with T-contrasts was utilized to determine time intervals of significant dif - [48]. ference between conditions, in which the GFP signals at each time point were permuted between the congruent EEG analysis and incongruent conditions across all subjects [43]. Using To facilitate future translation into point-of-care enabled this approach, the interval of significance was identified platforms, concurrently collected EEG data were also to be 300–500 ms and used as the window of interest in analyzed to extract ERPs. Contamination from ocular subsequent analyses, consistent with prior literature [44, sources was removed from the EEG signal using an adap- 45]. The mean GFP value in this time interval was then tive filtering approach [49]. For this process, the recorded calculated for each condition (congruent and incongru- EOG signals were used as reference inputs and processed ent) and participant, and compared using paired t test at using finite impulse response filters (m = 3), followed by the group level. recursive least squares-based removal from the EEG sig- nal (λ = 0.9999). Subsequent to artifact removal, standard Spectral effects analysis steps including filtering (1–10 Hz), segmentation Sensor level time–frequency analysis was undertaken by (− 200 to 900 ms) and conditional averaging were under- convolution of the data with Morlet wavelets (6 cycles) taken to generate ERPs [1, 2]. The mean value of the ERP using the continuous wavelet transform function in waveform at the Cz electrode site in the 300–500 ms time MATLAB (The Mathworks Inc., USA). The coefficients interval was calculated for each condition and partici- corresponding to 0.5–45  Hz frequency in the − 200 pant, and compared using paired t test at the group level. to 900  ms time window relative to stimulus onset were extracted, and log power was computed as the square of Individual‑level analysis the absolute value of the coefficients. To better under - To evaluate reliability of the rN400 ERP at the indi- stand the event-related spectral changes, the mean log vidual level, a machine learning-based approach was power in the baseline period (− 100 to 0  ms) was sub- undertaken using a two-category support vector tracted from the log power in the post-stimulus period machine (SVM) classifier following previously pub- for every trial within the frequency band. Significance lished methods [37, 50]. Briefly, an SVM classifier with was assessed using a bootstrapping approach by permut- a radial kernel was trained to distinguish between the ing the trial-averaged wavelet power in the congruent congruent and incongruent condition waveforms using Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 5 of 11 single-run, trial-averaged data from all three electrode Results sites. During each session, 90% of the available data Temporal and spectral effects in MEG were randomly selected to train the classifier, while Sensor-level GFP demonstrated differential processing the remaining 10% were used for testing classifica- of the target word depending upon whether they were tion accuracy. This procedure was repeated 10 times semantically related (congruent condition) or semanti- under tenfold cross-validation, such that the classifier cally unrelated (incongruent condition) to the first word. was trained and tested on all available data. Results In particular, in the 300–500  ms post-stimulus interval, were averaged across all sessions, and measures were there was increased power for the incongruent relative to derived from the confusion matrix corresponding congruent condition (p < 0.05, Fig. 2a, b). In addition, the to accuracy, sensitivity, and specificity. To further processing of incongruent words resulted in a significant assess the reliability of the analysis, results were veri- reduction in beta band power relative to the processing fied using non-parametric permutation statistics [34, of congruent words (p < 0.05, Fig.  2c). This decrease was 51]. In short, this involved randomly redistributing observed in the 335–440 ms time interval, overlapping in the congruent and incongruent class labels among all time with the N400 response. Although there appeared datasets and performing the same classification proce- to be some differences also present in other frequency dures. This process was repeated 1000 times, and the bands, none of them were statistically significant. resulting accuracies were used to create a null distribu- tion against which the true classification accuracy was Temporal effects in EEG compared. Probabilities less than 0.05 were deemed to ERP waveforms exhibited greater negativity in the incon- be significant for SVM classification outcome. gruent relative to congruent condition occurring within Fig. 2 Sensor-level MEG results showing differential processing in incongruent compared to congruent condition. a Grand-averaged GFP demonstrating increased power for incongruent relative to congruent condition. Shaded region denotes window of interest (300–500 ms). b Mean GFP averaged across the time window specified in part A, calculated for each subject and presented as mean ± SEM across subjects. *p < 0.05. c Time–frequency wavelet spectral power averaged over all MEG channels. Colour bar represents log power values Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 6 of 11 the 300–500  ms interval, which was maximal at the Pz electrode (p < 0.05, Fig. 3a–c). The trained SVM classifier successfully distinguished between the congruent and incongruent conditions with 88.89% accuracy, 88% sen- sitivity, and 90% specificity. All classification results were verified to be statistically significant through permuta - tion analysis (p < 0.05). Neuroanatomical effects in MEG Differential processing of incongruent words was source- localized to the inferior frontal, inferior parietal, and tem- poral regions (incongruent > congruent contrast, p < 0.005, k = 20). Key areas included left inferior, middle and supe- rior temporal gyri (BA 20, 21 and 22) and regions encom- Fig. 4 Source localization results. Top: Incongruent word processing activates a left-lateralized distributed region of cortex passing both the anterior and posterior portions of the including temporal, inferior frontal and inferior parietal areas left inferior frontal gyrus (BA 45, 47). Additionally, areas (incongruent > congruent contrast, p < 0.005unc.). Bottom: No of the right temporal and inferior frontal gyri were also suprathreshold clusters were identified for the reverse contrast activated. In comparison, no suprathreshold clusters were (congruent > incongruent). Color bar represents T-statistic values observed for the reverse contrast of congruent > incongru- ent (Fig. 4 bottom panel). Fig. 3 ERP results demonstrating differential processing of semantic congruence and incongruence. a–c Grand-averaged ERP waveforms at the Fz, Cz, and Pz electrode sites, respectively. Shaded regions denote windows of interest (300–500 ms). d Mean ERP amplitudes averaged over the windows of interest, calculated for each subject and presented as Mean ± SEM across subjects. *p < 0.05 Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 7 of 11 Discussion Results showed that sensor-level GFP exhibited increased Main findings activity in the incongruent relative to congruent condi- This study employed MEG with concurrent EEG to inves - tion, peaking at similar latencies relative to rN400 ERP tigate the temporal, spectral, and neuroanatomical char- (Fig.  2a, b). It is important to note that polarity differ - acteristics of the rapidly elicited N400 response (rN400) ences between the two modalities may be accounted for generated through the brain vital sign framework. Using given that GFP is a power measure and is thus always a compressed auditory stimulus sequence comprising non-negative, whereas ERP can be either positive or both tones and prime-target word pairs, we demon- negative. strated that the resulting rN400 response exhibited fea- While the present study targeted the semantic process- tures consistent with characteristics previously reported ing effect indexed by the N400 and accordingly focused for the N400 response in semantic language paradigms on the 300–500 ms window of interest to be concordant [17, 18]. In particular, we found that: (1) the sensor-level with previous literature [18, 44], other temporal differ - temporal characteristics showed rN400 ERP in the incon- ences between the two conditions were also present at gruent relative to congruent condition, peaking approxi- earlier latencies within the ERP/ERF traces. These effects mately 300–500 ms after stimulus presentation and with may be related to processes in support of semantic lan- concomitant changes in GFP (Hypothesis 1); (2) a signifi - guage comprehension such as phonological matching cant decrease in beta-band spectral power was observed [52], letter-string processing [45] or detection of mis- during the same interval in the incongruent relative to match based on predicted input [53]. These earlier effects congruent condition (Hypothesis 2); and (3) source local- may be further explored in future studies. ization analysis showed that rN400 processes activated cortical regions spanning the temporal, inferior frontal, Hypothesis 2: spectral effects and parietal regions known to be associated with the Time–frequency results demonstrated a significant N400 response (Hypothesis 3). These main findings are decrease in beta band power in the incongruent condi- summarized in Table 1. tion relative to the congruent (Fig.  2c). These spectral changes occurred over the same time interval as the Hypothesis 1: temporal effects rN400 response, and provide further confirmatory evi - The sensor-level temporal effects showed a robust rN400 dence of the processing differences between the two ERP for processing the incongruent relative to the con- conditions. A previous MEG study reported similar beta- gruent words (Fig.  3), consistent with previous findings band power reductions, and source-localized this effect based on sentences and semantic prime-target word pairs to the left inferior frontal gyrus and temporal regions, within auditory and visual modalities [17]. The response with the authors postulating that the observed N400 in the present study was observed to be maximal at the effects may have represented a dynamic communica - parietal (Pz) electrode location, also consistent with prior tion link between these regions [20]. Additionally, beta works suggesting a centro-parietal scalp distribution for band power suppression has also previously been associ- the N400 ERP [9]. Importantly, these findings were also ated with increased level of cortical processing across a supported by our concurrent results using MEG which diverse range of experimental paradigms, such as motor measures the magnetic counterpart of the rN400 ERP. movement [54], working memory [55] and information Table 1 Comparison of  the  features of  interest between  the  N400 response elicited using traditional approaches and the rN400 response elicited under the rapid assessment brain vital sign framework Modality Feature of interest Traditional approach N400 Rapid framework (rN400) EEG Peak amplitude (cong. vs. incong.) ERP: |V | > |V | ERP: |V | > |V | incong cong incong cong Peak latency (ms) ~ 400 ms 420 ms Scalp topographyCentro-parietal maxima Max at parietal (Pz) c,d MEG Amplitude difference (cong. vs. incong.) ∆ ∆ 300–500 ms 300–500 ms Spectral effects ⇓ beta-band power ⇓ beta-band power c,d,f Cortical activation ⇑ IFG, TL, IPL ⇑ IFG, TL, IPL Effects are based on comparison of the incongruent condition with the congruent condition data. EEG-based features include peak amplitude (V ), peak latency (ms), and scalp topography. MEG-based features include amplitude difference during the 300–500 ms window (∆ ), spectral effects, and cortical activations. 300–500ms Cong. congruent condition, incong. incongruent condition, IFG inferior frontal gyrus, TL temporal lobe (superior, middle and inferior temporal gyri), IPL inferior parietal lobule. Only statistically significant features are shown a b c d e f Kutas and Federmeier [17], Lau et al. [9], Halgren et al. [44], Maess et al. [24], Wang et al. [20], Helenius et al. [23] Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 8 of 11 retrieval [56]. In light of these findings, the reduction in functional integrity as well as for tracking rehabilitation beta band power observed in the current study may be progress. Beukema and colleagues reported the impor- interpreted as a potential reflection of increased process - tance of including N400 in assessments of DOC patients ing for the incongruent relative to congruent conditions [7], while Steppacher et  al. demonstrated the N400 as a within the relevant brain regions. It should also be noted crucial tool for assessing information processing abilities that, although reduced power is visually observed for the that are predictive of eventual recovery in DOC patients theta frequency band in the current study, this effect was [4]. Similarly, the N400 response has also been utilized not statistically significant. to track rehabilitation progress in traumatic brain injury [6] and for assessments of stroke patients [28]. Moreo- Hypothesis 3: neuroanatomical effects ver, the N400 response has been found to be abnormal in Our results showed left-lateralized activations in the Alzheimer’s disease [60], and was identified as a promis - temporal cortices (BA 20, 21, 22) as well as inferior fron- ing marker in differentially identifying MCI patients who tal gyri (BA 44, 45) (Fig. 4 upper panel). This is in agree - may transition to dementia [5]. These demonstrations in ment with prior works using fMRI and EEG, confirming clinical populations, combined with the excellent reliabil- the left temporal lobe as the largest source of the N400 ity and stability of N400 effects [61] provide an impetus effect, with a smaller contribution from the right tempo - for clinical integration of this promising response. The ral areas [21]. In addition, other EEG based works have present study makes N400 assessments clinically acces- identified contributions from the left perisylvian cortex sible by balancing the need for rapid assessments in [11], and bilateral inferior frontal gyri [22]. MEG based clinical settings with the inherent desire for high qual- source localization has largely confirmed these findings, ity data while retaining the key known features of the and suggested contributions from cortical areas includ- N400 response. Our results demonstrated that the rap- ing the left superior and middle temporal gyri as well as idly elicited N400 response through the brain vital sign the inferior parietal and frontal areas [23, 44]. The con - framework exhibit many of the similar characteristics verging neuroimaging results and theoretical models [9, compared to traditional N400 paradigms [9, 17, 62]. 57, 58] have led to increasing consensus that semantic Additionally, the robust identification of the N400 language processing is supported by a left lateralized net- effect at the individual level using automated expert- work of brain regions [9, 24, 44]. Our results are consist- independent machine learning approaches provides ent with these previous findings, as more left-lateralized additional support for clinical application of this rapid activations were observed in both the temporal and infe- assessment technique. The 89% hit rate in the present rior frontal regions. In addition to the left hemisphere study is quite comparable to previous reports—with activity, the right hemisphere activations observed in the prior machine learning based analysis reporting results current study were also in line with other studies using in the 86–92% range [34, 37] and other analytical tech- auditory stimuli [59]. niques also reporting observable N400 effects in similar The lack of suprathreshold clusters in the congru - proportions of healthy participants [7, 30]. ent > incongruent contrast (Fig.  4 lower panel) is also consistent with previous literature. MEG studies of N400 Caveats have shown largely overlapping areas of activation in Despite the promising findings in this study, two main both congruent and incongruent conditions, with greater limitations should also be noted. As this is the first study extent of activations in the incongruent condition due to characterizing the rapidly elicited rN400 response within increased demands associated with incongruent stimulus the brain vital sign framework, the focus was on exam- [24]. Similarly, fMRI results showed increased hemody- ining its spatiotemporal and neuroanatomical effects and namic activity for the incongruent condition compared comparing them with known features of the traditional to congruent [21]. Together, these hemodynamic and N400 response. However, given the myriad of language- electromagnetic results support our findings regarding and non-language-based experimental paradigms in lack of suprathreshold clusters in the congruent > incon- which the N400 response has previously been described, gruent contrast. it is not feasible to compare the rN400 response to every other traditional paradigm in one study. Rather, the cur- Clinical implications rent study focused on comparisons with language-based Beyond the extensive laboratory based evaluations of paradigms, and utilized response features and character- N400, clinical applications are increasingly utilizing istics that have been identified as commonalities across the N400 response in a variety of patient populations. different studies in order to account for variable modali - The N400 is being particularly studied in disorders of ties and experimental parameters (e.g. experimental consciousness (DOC) as a potential marker of residual condition, stimulus duration and type, inter-stimulus Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 9 of 11 BC V3V 1Z2, Canada. ImageTech Lab, Surrey Memorial Hospital, 13750 96 Av, interval) [9, 17, 62]. Nonetheless, future studies may Surrey, BC V3V 1Z2, Canada. be conducted to examine more detailed comparisons between the brain vital sign rN400 response and tradi- Acknowledgements The authors would like to acknowledge the volunteers for their participation tional N400 responses. Additionally, as the first study of in the study. We also thank Matt Courtemanche for assistance with data col- rN400 response, the current study utilized a distributed lection as well as Mary-Carmen Graham and Qun Gao for administrative and source modeling approach for source localization to be technical support. consistent with previous MEG studies of N400 [24, 44]. Competing interests However, given the inherent limitations of this approach Some authors are associated with HealthTech Connex Inc. This may qualify in biasing sources towards the cortical surface, future them to financially benefit from the commercialization of the NeuroCatch platform capable of measuring brain vital signs. studies are needed to confirm these results using alter - nate source localization techniques such as spatial filter - Availability of data and materials ing using beamformer [63]. The data that support the findings of this study are available from the cor - responding author upon reasonable request. Consent for publication Conclusion Not applicable. In this study, we investigated the spatiotemporal and Ethics approval and consent to participate neuroanatomical features of the N400 response as elic- The study was approved by ethics boards at Fraser Health Authority and ited by the rapid assessment brain vital signs framework. Simon Fraser University, and all participants provided written informed consent. Using both MEG and EEG, our results showed that the rapidly elicited N400 response exhibits characteristics Funding consistent with those reported in traditional semantic This work was partly supported by a grant from Mathematics of Information Technology And Complex Systems (MITACS, grant #IT03240). This study was language-based N400 paradigms. These characteristics also supported by a grant from the Surrey Memorial Hospital Foundation include temporal features showing maximal response and a grant from Natural Sciences and Engineering Research Council (NSERC, within 300–500  ms latency; topographic scalp distribu- grant # RGPIN-2015-04018), awarded to TC. SGH was supported by the Multi-Year Funding scholarship from Simon Fraser University and CL holds a tion demonstrating maximal response at the posterior Canadian Institutes of Health Research (CIHR) CGS Doctoral scholarship (grant Pz electrode; spectral effects showing reduction in beta # GSD-140381). band power; and source localization to left-lateralized temporal and inferior frontal areas. With the increas- Publisher’s Note ing use of the N400 response in patient assessments for Springer Nature remains neutral with regard to jurisdictional claims in pub- lished maps and institutional affiliations. neurological conditions such as dementia and traumatic brain injury, the convergent M/EEG results of the cur- Received: 3 March 2018 Accepted: 26 May 2018 rent study provide further support for the possibility of translating the N400 response from research to clinical settings through a rapid assessment framework for evalu- ating cognitive functions. References 1. Luck SJ. An introduction to the event-related potential technique. Cam- Abbreviations bridge: MIT Press; 2014. ERP: event related potential; MEG: magnetoencephalography; BA: brodmann 2. Gawryluk JR, D’Arcy RCN. Electroencephalography: basic concepts and area; rN400: rapidly elicited N400; GFP: global field power; GLM: general linear brain applications. In: Handbook of physics in medicine and biology. New model; IFG: inferior frontal gyrus; TL: temporal lobe; EEG: electroencephalogra- York: CRC Press; 2010. pp. 24-1–14. phy; fMRI: functional magnetic resonance imaging; MCI: mild cognitive impair- 3. Duncan CC, et al. Event-related potentials in clinical research: guidelines ment; ICA: independent component analysis; MNE: minimum norm estimate; for eliciting, recording, and quantifying mismatch negativity, P300, and SVM: support vector machine; IPL: inferior parietal lobule; DOC: disorders of N400. Clin Neurophysiol. 2009;120(11):1883–908. consciousness. 4. Steppacher I, Eickhoff S, Jordanov T, Kaps M, Witzke W, Kissler J. N400 predicts recovery from disorders of consciousness. Ann Neurol. Author contributions 2013;73(5):594–602. Conceptualization and study design: SGH, CL, and RD. Literature search: SGH, 5. Olichney JM, Yang JC, Taylor J, Kutas M. Cognitive event-related poten- CL, and RD. Data collection: SGH, CL, SF, and TC. Analysis planning: SGH and tials: biomarkers of synaptic dysfunction across the stages of Alzheimer’s CL. Data analysis: SGH and CL. Result presentation: SG, CL, and RD. Analysis disease. J Alzheimers Dis. 2011;26(Suppl 3):215–28. outcome verification: SG, CL, XS, and RD. Result interpretation: All authors. 6. Fleck-Prediger C et al. Clinical applications of the Halifax conscious- Manuscript preparation: SG, CL, and RD. Critical editing and approval of sub- ness scanner: tracking recovery in a severely brain injured patient. mission: All authors. All authors read and approved the final manuscript. International Brain Injury Association, International NeuroTrauma Let- ter, 2015. http://www.inter natio nalbr ain.org/artic les/int-issue -37/ Author details 7. Beukema S, et al. A hierarchy of event-related potential markers of Faculty of Applied Science, Simon Fraser University, Burnaby, BC, Canada. auditory processing in disorders of consciousness. NeuroImage Clin. Health Science and Innovation, Surrey Memorial Hospital, Fraser Health 2016;12:359–71. Authority, Surrey, BC, Canada. HealthTech Connex Inc, Surrey, BC, Canada. 8. Kutas M, Hillyard SA. Reading senseless sentences: brain potentials reflect Surrey NeuroTech Lab, Surrey Memorial Hospital, 13750 96 Avenue, Surrey, semantic incongruity. Science. 1980;207(4427):203–5. Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 10 of 11 9. Lau EF, Phillips C, Poeppel D. A cortical network for semantics:(de) con- 33. Wilson KR, O’Rourke H, Wozniak LA, Kostopoulos E, Marchand Y, Newman structing the N400. Nat Rev Neurosci. 2008;9(12):920–33. AJ. Changes in N400 topography following intensive speech language 10. Connolly JF, Phillips NA, Stewart SH, Brake WG. Event-related potential therapy for individuals with aphasia. Brain Lang. 2012;123(2):94–103. sensitivity to acoustic and semantic properties of terminal words in 34. Sculthorpe-Petley L, et al. A rapid event-related potential (ERP) sentences. Brain Lang. 1992;43(1):1–18. method for point-of-care evaluation of brain function: develop- 11. D’Arcy RCN, Connolly JF, Service E, Hawco CS, Houlihan ME. Separating ment of the Halifax Consciousness Scanner. J Neurosci Methods. phonological and semantic processing in auditory sentence processing: 2015;245:64–72. a high-resolution event-related brain potential study. Hum Brain Mapp. 35. Wang JT, Young GB, Connolly JF. Prognostic value of evoked responses 2004;22(1):40–51. and event-related brain potentials. Can J Neurol Sci. 2004;31(4):438–50. 12. D’Arcy RCN, Service E, Connolly JF, Hawco CS. The influence of increased 36. Sergent C, et al. Multidimensional cognitive evaluation of patients with working memory load on semantic neural systems: a high-resolution disorders of consciousness using EEG: a proof of concept study. NeuroIm- event-related brain potential study. Cognit Brain Res. 2005;22(2):177–91. age Clin. 2016;13:455–69. 13. Holcomb PJ, Neville HJ. Auditory and visual semantic priming in lexical 37. Ghosh Hajra S, et al. Developing brain vital signs: initial framework decision: a comparison using event-related brain potentials. Lang Cognit for monitoring brain function changes over time. Front Neurosci. Process. 1990;5(4):281–312. 2016;10:211. 14. Neville HJ, Mills DL, Lawson DS. Fractionating language: different neural 38. Davis PA. Eec ff ts of acoustic stimuli on the waking human brain. J Neuro - subsystems with different sensitive periods. Cereb Cortex (New York, NY, physiol. 1939;2(6):494–9. 1991). 1992;2(3):244–58. 39. Sutton S, Tueting P, Zubin J, John ER. Information delivery and the sen- 15. Bentin S, McCarthy G, Wood CC. Event-related potentials, lexical deci- sory evoked potential. Science. 1967;155(3768):1436–9. sion and semantic priming. Electroencephalogr Clin Neurophysiol. 40. Taylor WL. ‘Cloze procedure’: a new tool for measuring readability. J Bull. 1985;60(4):343–55. 1953;30(4):415–33. 16. Rugg MD. The effects of semantic priming and word repetition on event- 41. Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of related potentials. Psychophysiology. 1985;22(6):642–7. single-trial EEG dynamics including independent component analysis. J 17. Kutas M, Federmeier KD. Thirty years and counting: finding meaning in Neurosci Methods. 2004;134(1):9–21. the N400 component of the event-related brain potential (ERP). Annu 42. Skrandies W. Global field power and topographic similarity. Brain Topogr. Rev Psychol. 2011;62:621–47. 1990;3(1):137–41. 18. Kutas M, Federmeier KD. Electrophysiology reveals semantic memory use 43. Maris E, Oostenveld R. Nonparametric statistical testing of EEG-and MEG- in language comprehension. Trends Cognit Sci. 2000;4(12):463–70. data. J Neurosci Methods. 2007;164(1):177–90. 19. Roehm D, Bornkessel-Schlesewsky I, Schlesewsky M. The internal struc- 44. Halgren E, et al. N400-like magnetoencephalography responses modu- ture of the N400: frequency characteristics of a language related ERP lated by semantic context, word frequency, and lexical class in sentences. component. In: Chaos and complexity letters, vol. 2; 2007. pp. 365–95. Neuroimage. 2002;17(3):1101–16. 20. Wang L, et al. Beta oscillations relate to the N400 m during language 45. Pylkkänen L, Marantz A. Tracking the time course of word recognition comprehension. Hum Brain Mapp. 2012;33(12):2898–912. with MEG. Trends Cognit Sci. 2003;7(5):187–9. 21. Van Petten C, Luka BJ. Neural localization of semantic context 46. Liu CC, Ghosh Hajra S, Cheung TPL, Song X, D’Arcy RCN. Spontaneous effects in electromagnetic and hemodynamic studies. Brain Lang. blinks activate the precuneus: characterizing blink-related oscillations 2006;97(3):279–93. using magnetoencephalography. Front Hum Neurosci. 2017;11:489. 22. Haan H, Streb J, Bien S, Rösler F. Individual cortical current density recon- 47. Litvak V, Friston K. Electromagnetic source reconstruction for group stud- structions of the semantic N400 effect: using a generalized minimum ies. Neuroimage. 2008;42(4):1490–8. norm model with different constraints (L1 and L2 norm). Hum Brain 48. Friston KJ, Holmes AP, Worsley KJ, Poline J, Frith CD, Frackowiak RSJ. Sta- Mapp. 2000;11(3):178–92. tistical parametric maps in functional imaging: a general linear approach. 23. Helenius P, Salmelin R, Service E, Connolly JF, Leinonen S, Lyytinen H. Hum Brain Mapp. 1994;2(4):189–210. Cortical activation during spoken-word segmentation in nonreading- 49. He P, Wilson G, Russell C. Removal of ocular artifacts from electro-enceph- impaired and dyslexic adults. J. Neurosci. 2002;22(7):2936–44. alogram by adaptive filtering. Med Biol Eng Comput. 2004;42(3):407–12. 24. Maess B, Herrmann CS, Hahne A, Nakamura A, Friederici AD. Local- 50. Parvar H, Sculthorpe-Petley L, Satel J, Boshra R, D’Arcy RCN, Trappenberg izing the distributed language network responsible for the N400 TP. Detection of event-related potentials in individual subjects using sup- measured by MEG during auditory sentence processing. Brain Res. port vector machines. Brain Inform. 2014;2(1):1–12. 2006;1096(1):163–72. 51. P. Golland, B. Fischl. Permutation tests for classification: towards statistical 25. Dronkers NF, Wilkins DP, Van Valin RD, Redfern BB, Jaeger JJ. Lesion analy- significance in image-based studies. In: Biennial international conference sis of the brain areas involved in language comprehension. Cognition. on information processing in medical imaging. Berlin: Springer; 2003 pp. 2004;92(1):145–77. 330–341 26. Balconi MFAU, Arangio RFAU, Guarnerio C. Disorders of consciousness 52. Connolly JF, Phillips NA. Event-related potential components reflect and N400 ERP measures in response to a semantic task. J Neuropsychia- phonological and semantic processing of the terminal word of spoken try Clin Neurosci. 2013;25:237. sentences. J Cognit Neurosci. 1994;6(3):256–66. 27. Connolly JF, D’Arcy RCN. Innovations in neuropsychological assess- 53. Dikker S, Rabagliati H, Farmer TA, Pylkkanen L. Early occipital sensi- ment using event-related brain potentials. Int J Psychophysiol. tivity to syntactic category is based on form typicality. Psychol Sci. 2000;37(1):31–47. 2010;21(5):629–34. 28. D’Arcy RCN, et al. Electrophysiological assessment of language function 54. Alegre M, Gurtubay IG, Labarga A, Iriarte J, Malanda A, Artieda J. Alpha following stroke. Clin Neurophysiol. 2003;114(4):662–72. and beta oscillatory changes during stimulus-induced movement para- 29. Marchand Y, D’Arcy RCN, Connolly JF. Linking neurophysiological and digms: effect of stimulus predictability. NeuroReport. 2003;14(3):381–5. neuropsychological measures for aphasia assessment. Clin Neurophysiol. 55. Tallon-Baudry C. Oscillatory synchrony and human visual cognition. J 2002;113(11):1715–22. Physiol. 2003;97(2):355–63. 30. Neumann N, Kotchoubey B. Assessment of cognitive functions in severely 56. Pesonen M, Björnberg CH, Hämäläinen H, Krause CM. Brain oscillatory paralysed and severely brain-damaged patients: neuropsychological and 1–30 Hz EEG ERD/ERS responses during the different stages of an audi- electrophysiological methods. Brain Res Protoc. 2004;14(1):25–36. tory memory search task. Neurosci Lett. 2006;399(1):45–50. 31. Olichney JM, et al. Patients with MCI and N400 or P600 abnormalities are 57. Friederici AD. Towards a neural basis of auditory sentence processing. at very high risk for conversion to dementia. Neurology. 2008;70(19 Pt Trends Cognit Sci. 2002;6(2):78–84. 2):1763–70. 58. Hickok G, Poeppel D. Neural basis of speech perception. Hum Audit Syst 32. Wang K, Cheung EFC, Gong Q, Chan RCK. Semantic processing dis- Fundam Org Clin Disord. 2015;129:149–60. turbance in patients with schizophrenia: a meta-analysis of the N400 59. Friederici AD, Ruschemeyer SA, Hahne A, Fiebach CJ. The role of left component. PLoS ONE. 2011;6(10):e25435. inferior frontal and superior temporal cortex in sentence comprehension: Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 11 of 11 localizing syntactic and semantic processes. Cereb Cortex (New York, NY, 62. Lau EF, Weber K, Gramfort A, Hämäläinen MS, Kuperberg GR. Spati- 1991). 2003;13(2):170–7. otemporal signatures of lexical-semantic prediction. Cereb Cortex. 60. Ford JM, et al. N400 evidence of abnormal responses to speech in Alzhei- 2016;26(4):1377–87. mer’s disease. Electroencephalogr Clin Neurophysiol. 1996;99(3):235–46. 63. Hillebrand A, Barnes GR. Beamformer analysis of MEG data. Int Rev Neuro- 61. Kiang M, Patriciu I, Roy C, Christensen BK, Zipursky RB. Test–retest reli- biol. 2005;68:149–71. ability and stability of N400 effects in a word-pair semantic priming paradigm. Clin Neurophysiol. 2013;124(4):667–74. Ready to submit your research ? Choose BMC and benefit from: fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Translational Medicine Springer Journals

Multimodal characterization of the semantic N400 response within a rapid evaluation brain vital sign framework

Free
11 pages

Loading next page...
 
/lp/springer_journal/multimodal-characterization-of-the-semantic-n400-response-within-a-DEcZo0WybT
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Biomedicine; Biomedicine, general; Medicine/Public Health, general
eISSN
1479-5876
D.O.I.
10.1186/s12967-018-1527-2
Publisher site
See Article on Publisher Site

Abstract

Background: For nearly four decades, the N400 has been an important brainwave marker of semantic processing. It can be recorded non-invasively from the scalp using electrical and/or magnetic sensors, but largely within the restricted domain of research laboratories specialized to run specific N400 experiments. However, there is increas- ing evidence of significant clinical utility for the N400 in neurological evaluation, particularly at the individual level. To enable clinical applications, we recently reported a rapid evaluation framework known as “brain vital signs” that successfully incorporated the N400 response as one of the core components for cognitive function evaluation. The current study characterized the rapidly evoked N400 response to demonstrate that it shares consistent features with traditional N400 responses acquired in research laboratory settings—thereby enabling its translation into brain vital signs applications. Methods: Data were collected from 17 healthy individuals using magnetoencephalography (MEG) and electroen- cephalography (EEG), with analysis of sensor-level effects as well as evaluation of brain sources. Individual-level N400 responses were classified using machine learning to determine the percentage of participants in whom the response was successfully detected. Results: The N400 response was observed in both M/EEG modalities showing significant differences to incongruent versus congruent condition in the expected time range (p < 0.05). Also as expected, N400-related brain activity was observed in the temporal and inferior frontal cortical regions, with typical left-hemispheric asymmetry. Classification robustly confirmed the N400 effect at the individual level with high accuracy (89%), sensitivity (0.88) and specificity (0.90). Conclusion: The brain vital sign N400 characteristics were highly consistent with features of the previously reported N400 responses acquired using traditional laboratory-based experiments. These results provide important evidence supporting clinical translation of the rapidly acquired N400 response as a potential tool for assessments of higher cognitive functions. Keywords: N400, ERP, MEG, Semantic language, Clinical application *Correspondence: rdarcy@sfu.ca Faculty of Applied Science, Simon Fraser University, Burnaby, BC, Canada Full list of author information is available at the end of the article © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/ publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 2 of 11 likely responsible for N400 [24], and these results are also Background supported by findings from lesion studies [25]. Measurements of brainwave activity through event- Further to its functional relevance as an indicator related potentials (ERPs) are becoming increasingly use- of neural processing in healthy individuals, the N400 ful in providing objective, physiology-based measures of response has also shown significant potential as a diag - brain function [1]. ERPs are derived from electroenceph- nostic and prognostic tool in clinical populations [4, 17, alography (EEG), and can provide information about cor- 26–33]. Studies in brain-injured patients with disor- tical electrical activity corresponding to different aspects ders of consciousness showed that the N400 response of neural processing [2, 3]. In particular, higher order was correlated with functional recovery [4]. Moreo- cognitive functions like semantic processing indexed by ver, changes in N400 response also predicted cognitive the N400 ERP are among the most promising responses decline in patients as they progressed from mild cogni- for emerging clinical applications [4–7]. The N400 tive impairment (MCI) to dementia [5, 31]. Yet despite response was first described when Kutas and Hillyard these promising findings, the use of the N400 ERP presented participants with visual sentences that either beyond the research setting has been hindered by two had a semantically related (i.e. congruent) or semantically main challenges: First, given that ERPs are produced unrelated (i.e. incongruent) ending [8]. It was observed as by averaging the neural response signals across a large a negative deflection of the incongruent relative to con - number of trials, traditional N400 studies require pro- gruent condition waveforms which peaked at approxi- longed testing paradigms [1, 34]. These paradigms are mately 400  ms latency following stimulus presentation, particularly problematic in clinical populations due to and the authors suggested that this differential was a neu - fluctuations in vigilance levels and lack of capability or ral marker of semantic language processing. motivation [30, 35]. In addition, rather than measuring In the 38  years since its initial report, the N400 only a single brain response in clinical populations (e.g. response has been studied extensively using a variety of sensation, attention, or language), there are now calls for stimulus paradigms in various healthy and clinical pop- concurrent evaluations of a spectrum of brain responses ulations [9–14]. While the initial N400 work utilized which provide a more complete profile of brain function sentence-based stimuli, subsequent studies showed that [34]. This is particularly crucial in longitudinal monitor - prime-target word pairs also successfully elicited this ing of brain function changes in clinical populations [36]. response [15, 16]. Additionally, non-language-based Under these circumstances, the traditional ERP test- stimuli such as mental arithmetic and action sequences ing paradigms may require hours to evaluate, which is have also been shown to produce the N400 response [17], impractical within most clinical settings. and the strength of this response has been found to be To assess the N400 response within a short testing time correlated with various stimulus properties [18]. Oth- while providing information about other brain function ers have demonstrated overlapping features in the tem- indicators, our group has been undertaking systematic poral and spatial characteristics of the N400 response development of rapid evaluation techniques in recent when elicited using language- as well as non-language- years. We previously demonstrated the successful evalu- based stimuli [17], with the spectral content in particu- ation of the N400 response in 100 healthy individuals lar demonstrating potential in distinguishing between using a point-of-care enabled device [34], then employed different neural processes [19]. In fact, one of the key this device to track the progress of rehabilitation therapy spectral features of the N400 response has been shown in a brain-injured patient [6]. More recently, we demon- to be a reduction in beta band oscillations when process- strated a rapid evaluation platform known as the ‘brain ing incongruent relative to congruent stimuli in semantic vital sign’ framework [37], which enables the rapid language paradigms [20]. assessment of several brain function indicators including The cortical generators of the N400 response have been the N400 (semantic language), N100 (sensory processing) investigated using numerous noninvasive imaging modal- [38] and P300 (attention orienting) [39]. The brain vital ities, such as functional magnetic resonance imaging sign framework employs a portable, low-density EEG sys- (fMRI), electroencephalography (EEG), as well as magne- tem, with automated, user-friendly software for easy clin- toencephalography (MEG). Results have revealed wide- ical applications. The testing paradigm utilizes a short, spread cortical activations across the left temporal lobe, 5-min auditory stimulus sequence in which tone and along with smaller areas of activity in the right temporal word stimuli are interlaced to maximize the number of as well as bilateral inferior frontal and parietal regions trials and signal-to-noise ratio. Results in healthy adults [11, 21–23]. Specifically, areas of the bilateral temporal showed that, not only were the target responses success- cortices (Brodmann Areas [BA] 20/21/22) and left infe- fully elicited at the individual level, but the platform also rior frontal gyrus (BA 45/47) have been shown to be key captured expected age-related changes in attention and cortical regions within the distributed language network Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 3 of 11 cognition that were undetected using conventional clini- 5 sec cal screening measures [37]. Although the rapid evaluation brain vital sign frame- work showed initial promise as a potential avenue for clinical application of the N400 ERP, the component characteristics of this rapidly elicited N400 (rN400) 5 min response have not yet been described. Given the short, complex stimulus paradigm, it is crucial to characterize Color represents type of stimulus Tone Word this response with respect to its spatiotemporal, spectral, Fig. 1 Illustration of auditory stimulus sequence of the brain vital and neuroanatomical features, and compare them with sign framework. Blocks of five tones and two words repeated 60 times for a total scan time of about 5 min. Words represent known N400 characteristics reported in studies using prime-target pairs, containing both semantic congruent (pink– more conventional approaches over the last few decades. orange) and incongruent (pink–blue) pairs. Tones (standard = green The current study utilized MEG with simultaneous EEG and deviant = black) elicit sensory (N100) and attention (P300) to investigate the temporal, spatial, spectral, and neuro- measures anatomical characteristics of the rN400 response elicited within the brain vital sign framework. We hypothesized that the rN400 response will exhibit features consistent with known characteristics of the N400 response, includ- semantically linked (incongruent condition, 50%, e.g. doc- ing: (1) increased ERP negativity and MEG signal power tor-egg) to generate the differential processing measures. for the incongruent relative to congruent condition dur- Words in both groups were balanced for characteristics ing the 300–500 ms post-stimulus interval; (2) decreased such as word frequency and length, and the words in the beta- band power for the incongruent relative to congru- semantically linked group had a minimum Cloze prob- ent condition during the same interval; and (3) increased ability of 0.8 [40]. The stimuli were recorded in a male activation of temporal and frontal cortices (BA 20, 21, 22, voice and root-mean-square normalized using Audacity 45 and 47) for processing of incongruent relative to con- software. The stimulus sequence contained 30 trials each gruent stimuli. of the congruent and incongruent conditions. Methods Participant details MEG and EEG data acquisition Seventeen (17) right-handed healthy participants with no A 151-channel CTF MEG (MEG International Services history of neurological problems or psychoactive medi- Limited, Canada) was used with concurrent 3-channel cation were recruited (22.6 ± 2.4  years, 10 males). Par- EEG, both recorded in a magnetically shielded room with ticipants were undergraduate or graduate students, had the participants in the supine position. Data were sam- normal hearing, normal or corrected-to-normal vision, pled at 1200 Hz using axial gradiometers (5-cm baseline) and were fluent in English. The study was approved by with synthetic 3rd order gradients employed for noise ethics boards at Fraser Health Authority and Simon cancellation. Continuous head position monitoring was Fraser University, and all participants provided written undertaken by three head position indicator coils located informed consent. at fiducial points (HPI, positioned at nasion, left and right pre-auricular points). EEG recordings utilized Ag/AgCl Auditory stimuli scalp electrodes placed at Fz, Cz and Pz locations, with As introduced elsewhere [37], the rapid assessment impedances kept below 5  kOhms. Four additional elec- framework utilizes a compressed auditory stimulus trodes were placed on the head corresponding to refer- sequence with interlaced tones and words to elicit brain ence (left mastoid), ground (forehead), horizontal (outer responses across four different functional domains— canthus of left eye) and vertical (supra-orbital ridge of auditory sensation (N100 ERP), attention (P300 ERP), left eye) electro-occulogram (EOG). To facilitate the and semantic language (N400 ERP)—in approximately alignment of MEG scanner and head coordinate systems, 5  min (Fig.  1). The sequence comprised 60 blocks, with the shape of the participants’ head and the 3-dimensional each block containing five tones and two words repre - position of HPI coils and EEG/EOG electrodes were senting a prime-target pair. Semantic language process- recorded using a Polhemus electromagnetic digitization ing responses were derived from conditionally averaging system prior to data collection (Polhemus Incorporated, the trials corresponding to the target word in the pair. USA). Auditory stimulation was presented binaurally Semantically linked words (congruent condition, 50%, using insert earphones, and participants were instructed e.g. doctor-nurse) were contrasted with words not to maintain visual fixation on a crosshair displayed on Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 4 of 11 the overhead screen (white cross on black background) and incongruent conditions across participants in each throughout the session. frequency [43]. This entailed the calculation of T-statistic for each time point and frequency between the congru- Data preprocessing ent and incongruent conditions in the 800  ms following Raw data for both MEG and EEG were first visually stimulus presentation. Thereafter, 1000 permutations inspected, and artifactual channels removed from further were undertaken and new T-statistic calculated for every analysis. Data were then down-sampled to 300 Hz, notch permutation leading to a null distribution against which filtered to remove frequencies corresponding to power the significance of the true T-statistic was assessed (with line (60 Hz) with its harmonics as well as HPI coils, and p < 0.05 considered to be significant). low-pass filtered to 100  Hz. Data from 2 of the 17 par - ticipants were excluded from subsequent analysis due to Neuroanatomical effects poor quality. Source level analysis was performed using SPM8 (Wel- come Trust Centre for Neuroimaging, UK) with the for- MEG analysis ward and inverse modeling steps elaborated in previously Following band-pass filtering (0.5–45  Hz), independent published work [46]. Source analysis for localizing neural component analysis (ICA) was performed with runica generators of the semantic language process was under- algorithm in EEGLAB [41] in order to remove artifact taken using minimum norm estimates (MNE) to main- from ocular, cardiac, and muscular sources. tain consistency with prior N400 studies in MEG [24, 44]. Group constraints were employed during inversion [47], Temporal effects and source reconstruction was based on trial-averaged Since head position within the MEG helmet can vary data within the entire frequency range (0.5–45  Hz) and across participants, global field power (GFP) was uti - active epoch (0–900  ms relative to stimulus presenta- lized to provide a measure of the overall activity across tion). Source-level contrast images were derived using all channels [42]. Individual-level GFP was computed for data in the 0.5–45  Hz frequency range and previously the congruent and incongruent conditions using trial- identified window of 300–500  ms. Statistical modeling averaged event-related fields. A bootstrapping approach employed a general linear model (GLM) with T-contrasts was utilized to determine time intervals of significant dif - [48]. ference between conditions, in which the GFP signals at each time point were permuted between the congruent EEG analysis and incongruent conditions across all subjects [43]. Using To facilitate future translation into point-of-care enabled this approach, the interval of significance was identified platforms, concurrently collected EEG data were also to be 300–500 ms and used as the window of interest in analyzed to extract ERPs. Contamination from ocular subsequent analyses, consistent with prior literature [44, sources was removed from the EEG signal using an adap- 45]. The mean GFP value in this time interval was then tive filtering approach [49]. For this process, the recorded calculated for each condition (congruent and incongru- EOG signals were used as reference inputs and processed ent) and participant, and compared using paired t test at using finite impulse response filters (m = 3), followed by the group level. recursive least squares-based removal from the EEG sig- nal (λ = 0.9999). Subsequent to artifact removal, standard Spectral effects analysis steps including filtering (1–10 Hz), segmentation Sensor level time–frequency analysis was undertaken by (− 200 to 900 ms) and conditional averaging were under- convolution of the data with Morlet wavelets (6 cycles) taken to generate ERPs [1, 2]. The mean value of the ERP using the continuous wavelet transform function in waveform at the Cz electrode site in the 300–500 ms time MATLAB (The Mathworks Inc., USA). The coefficients interval was calculated for each condition and partici- corresponding to 0.5–45  Hz frequency in the − 200 pant, and compared using paired t test at the group level. to 900  ms time window relative to stimulus onset were extracted, and log power was computed as the square of Individual‑level analysis the absolute value of the coefficients. To better under - To evaluate reliability of the rN400 ERP at the indi- stand the event-related spectral changes, the mean log vidual level, a machine learning-based approach was power in the baseline period (− 100 to 0  ms) was sub- undertaken using a two-category support vector tracted from the log power in the post-stimulus period machine (SVM) classifier following previously pub- for every trial within the frequency band. Significance lished methods [37, 50]. Briefly, an SVM classifier with was assessed using a bootstrapping approach by permut- a radial kernel was trained to distinguish between the ing the trial-averaged wavelet power in the congruent congruent and incongruent condition waveforms using Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 5 of 11 single-run, trial-averaged data from all three electrode Results sites. During each session, 90% of the available data Temporal and spectral effects in MEG were randomly selected to train the classifier, while Sensor-level GFP demonstrated differential processing the remaining 10% were used for testing classifica- of the target word depending upon whether they were tion accuracy. This procedure was repeated 10 times semantically related (congruent condition) or semanti- under tenfold cross-validation, such that the classifier cally unrelated (incongruent condition) to the first word. was trained and tested on all available data. Results In particular, in the 300–500  ms post-stimulus interval, were averaged across all sessions, and measures were there was increased power for the incongruent relative to derived from the confusion matrix corresponding congruent condition (p < 0.05, Fig. 2a, b). In addition, the to accuracy, sensitivity, and specificity. To further processing of incongruent words resulted in a significant assess the reliability of the analysis, results were veri- reduction in beta band power relative to the processing fied using non-parametric permutation statistics [34, of congruent words (p < 0.05, Fig.  2c). This decrease was 51]. In short, this involved randomly redistributing observed in the 335–440 ms time interval, overlapping in the congruent and incongruent class labels among all time with the N400 response. Although there appeared datasets and performing the same classification proce- to be some differences also present in other frequency dures. This process was repeated 1000 times, and the bands, none of them were statistically significant. resulting accuracies were used to create a null distribu- tion against which the true classification accuracy was Temporal effects in EEG compared. Probabilities less than 0.05 were deemed to ERP waveforms exhibited greater negativity in the incon- be significant for SVM classification outcome. gruent relative to congruent condition occurring within Fig. 2 Sensor-level MEG results showing differential processing in incongruent compared to congruent condition. a Grand-averaged GFP demonstrating increased power for incongruent relative to congruent condition. Shaded region denotes window of interest (300–500 ms). b Mean GFP averaged across the time window specified in part A, calculated for each subject and presented as mean ± SEM across subjects. *p < 0.05. c Time–frequency wavelet spectral power averaged over all MEG channels. Colour bar represents log power values Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 6 of 11 the 300–500  ms interval, which was maximal at the Pz electrode (p < 0.05, Fig. 3a–c). The trained SVM classifier successfully distinguished between the congruent and incongruent conditions with 88.89% accuracy, 88% sen- sitivity, and 90% specificity. All classification results were verified to be statistically significant through permuta - tion analysis (p < 0.05). Neuroanatomical effects in MEG Differential processing of incongruent words was source- localized to the inferior frontal, inferior parietal, and tem- poral regions (incongruent > congruent contrast, p < 0.005, k = 20). Key areas included left inferior, middle and supe- rior temporal gyri (BA 20, 21 and 22) and regions encom- Fig. 4 Source localization results. Top: Incongruent word processing activates a left-lateralized distributed region of cortex passing both the anterior and posterior portions of the including temporal, inferior frontal and inferior parietal areas left inferior frontal gyrus (BA 45, 47). Additionally, areas (incongruent > congruent contrast, p < 0.005unc.). Bottom: No of the right temporal and inferior frontal gyri were also suprathreshold clusters were identified for the reverse contrast activated. In comparison, no suprathreshold clusters were (congruent > incongruent). Color bar represents T-statistic values observed for the reverse contrast of congruent > incongru- ent (Fig. 4 bottom panel). Fig. 3 ERP results demonstrating differential processing of semantic congruence and incongruence. a–c Grand-averaged ERP waveforms at the Fz, Cz, and Pz electrode sites, respectively. Shaded regions denote windows of interest (300–500 ms). d Mean ERP amplitudes averaged over the windows of interest, calculated for each subject and presented as Mean ± SEM across subjects. *p < 0.05 Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 7 of 11 Discussion Results showed that sensor-level GFP exhibited increased Main findings activity in the incongruent relative to congruent condi- This study employed MEG with concurrent EEG to inves - tion, peaking at similar latencies relative to rN400 ERP tigate the temporal, spectral, and neuroanatomical char- (Fig.  2a, b). It is important to note that polarity differ - acteristics of the rapidly elicited N400 response (rN400) ences between the two modalities may be accounted for generated through the brain vital sign framework. Using given that GFP is a power measure and is thus always a compressed auditory stimulus sequence comprising non-negative, whereas ERP can be either positive or both tones and prime-target word pairs, we demon- negative. strated that the resulting rN400 response exhibited fea- While the present study targeted the semantic process- tures consistent with characteristics previously reported ing effect indexed by the N400 and accordingly focused for the N400 response in semantic language paradigms on the 300–500 ms window of interest to be concordant [17, 18]. In particular, we found that: (1) the sensor-level with previous literature [18, 44], other temporal differ - temporal characteristics showed rN400 ERP in the incon- ences between the two conditions were also present at gruent relative to congruent condition, peaking approxi- earlier latencies within the ERP/ERF traces. These effects mately 300–500 ms after stimulus presentation and with may be related to processes in support of semantic lan- concomitant changes in GFP (Hypothesis 1); (2) a signifi - guage comprehension such as phonological matching cant decrease in beta-band spectral power was observed [52], letter-string processing [45] or detection of mis- during the same interval in the incongruent relative to match based on predicted input [53]. These earlier effects congruent condition (Hypothesis 2); and (3) source local- may be further explored in future studies. ization analysis showed that rN400 processes activated cortical regions spanning the temporal, inferior frontal, Hypothesis 2: spectral effects and parietal regions known to be associated with the Time–frequency results demonstrated a significant N400 response (Hypothesis 3). These main findings are decrease in beta band power in the incongruent condi- summarized in Table 1. tion relative to the congruent (Fig.  2c). These spectral changes occurred over the same time interval as the Hypothesis 1: temporal effects rN400 response, and provide further confirmatory evi - The sensor-level temporal effects showed a robust rN400 dence of the processing differences between the two ERP for processing the incongruent relative to the con- conditions. A previous MEG study reported similar beta- gruent words (Fig.  3), consistent with previous findings band power reductions, and source-localized this effect based on sentences and semantic prime-target word pairs to the left inferior frontal gyrus and temporal regions, within auditory and visual modalities [17]. The response with the authors postulating that the observed N400 in the present study was observed to be maximal at the effects may have represented a dynamic communica - parietal (Pz) electrode location, also consistent with prior tion link between these regions [20]. Additionally, beta works suggesting a centro-parietal scalp distribution for band power suppression has also previously been associ- the N400 ERP [9]. Importantly, these findings were also ated with increased level of cortical processing across a supported by our concurrent results using MEG which diverse range of experimental paradigms, such as motor measures the magnetic counterpart of the rN400 ERP. movement [54], working memory [55] and information Table 1 Comparison of  the  features of  interest between  the  N400 response elicited using traditional approaches and the rN400 response elicited under the rapid assessment brain vital sign framework Modality Feature of interest Traditional approach N400 Rapid framework (rN400) EEG Peak amplitude (cong. vs. incong.) ERP: |V | > |V | ERP: |V | > |V | incong cong incong cong Peak latency (ms) ~ 400 ms 420 ms Scalp topographyCentro-parietal maxima Max at parietal (Pz) c,d MEG Amplitude difference (cong. vs. incong.) ∆ ∆ 300–500 ms 300–500 ms Spectral effects ⇓ beta-band power ⇓ beta-band power c,d,f Cortical activation ⇑ IFG, TL, IPL ⇑ IFG, TL, IPL Effects are based on comparison of the incongruent condition with the congruent condition data. EEG-based features include peak amplitude (V ), peak latency (ms), and scalp topography. MEG-based features include amplitude difference during the 300–500 ms window (∆ ), spectral effects, and cortical activations. 300–500ms Cong. congruent condition, incong. incongruent condition, IFG inferior frontal gyrus, TL temporal lobe (superior, middle and inferior temporal gyri), IPL inferior parietal lobule. Only statistically significant features are shown a b c d e f Kutas and Federmeier [17], Lau et al. [9], Halgren et al. [44], Maess et al. [24], Wang et al. [20], Helenius et al. [23] Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 8 of 11 retrieval [56]. In light of these findings, the reduction in functional integrity as well as for tracking rehabilitation beta band power observed in the current study may be progress. Beukema and colleagues reported the impor- interpreted as a potential reflection of increased process - tance of including N400 in assessments of DOC patients ing for the incongruent relative to congruent conditions [7], while Steppacher et  al. demonstrated the N400 as a within the relevant brain regions. It should also be noted crucial tool for assessing information processing abilities that, although reduced power is visually observed for the that are predictive of eventual recovery in DOC patients theta frequency band in the current study, this effect was [4]. Similarly, the N400 response has also been utilized not statistically significant. to track rehabilitation progress in traumatic brain injury [6] and for assessments of stroke patients [28]. Moreo- Hypothesis 3: neuroanatomical effects ver, the N400 response has been found to be abnormal in Our results showed left-lateralized activations in the Alzheimer’s disease [60], and was identified as a promis - temporal cortices (BA 20, 21, 22) as well as inferior fron- ing marker in differentially identifying MCI patients who tal gyri (BA 44, 45) (Fig. 4 upper panel). This is in agree - may transition to dementia [5]. These demonstrations in ment with prior works using fMRI and EEG, confirming clinical populations, combined with the excellent reliabil- the left temporal lobe as the largest source of the N400 ity and stability of N400 effects [61] provide an impetus effect, with a smaller contribution from the right tempo - for clinical integration of this promising response. The ral areas [21]. In addition, other EEG based works have present study makes N400 assessments clinically acces- identified contributions from the left perisylvian cortex sible by balancing the need for rapid assessments in [11], and bilateral inferior frontal gyri [22]. MEG based clinical settings with the inherent desire for high qual- source localization has largely confirmed these findings, ity data while retaining the key known features of the and suggested contributions from cortical areas includ- N400 response. Our results demonstrated that the rap- ing the left superior and middle temporal gyri as well as idly elicited N400 response through the brain vital sign the inferior parietal and frontal areas [23, 44]. The con - framework exhibit many of the similar characteristics verging neuroimaging results and theoretical models [9, compared to traditional N400 paradigms [9, 17, 62]. 57, 58] have led to increasing consensus that semantic Additionally, the robust identification of the N400 language processing is supported by a left lateralized net- effect at the individual level using automated expert- work of brain regions [9, 24, 44]. Our results are consist- independent machine learning approaches provides ent with these previous findings, as more left-lateralized additional support for clinical application of this rapid activations were observed in both the temporal and infe- assessment technique. The 89% hit rate in the present rior frontal regions. In addition to the left hemisphere study is quite comparable to previous reports—with activity, the right hemisphere activations observed in the prior machine learning based analysis reporting results current study were also in line with other studies using in the 86–92% range [34, 37] and other analytical tech- auditory stimuli [59]. niques also reporting observable N400 effects in similar The lack of suprathreshold clusters in the congru - proportions of healthy participants [7, 30]. ent > incongruent contrast (Fig.  4 lower panel) is also consistent with previous literature. MEG studies of N400 Caveats have shown largely overlapping areas of activation in Despite the promising findings in this study, two main both congruent and incongruent conditions, with greater limitations should also be noted. As this is the first study extent of activations in the incongruent condition due to characterizing the rapidly elicited rN400 response within increased demands associated with incongruent stimulus the brain vital sign framework, the focus was on exam- [24]. Similarly, fMRI results showed increased hemody- ining its spatiotemporal and neuroanatomical effects and namic activity for the incongruent condition compared comparing them with known features of the traditional to congruent [21]. Together, these hemodynamic and N400 response. However, given the myriad of language- electromagnetic results support our findings regarding and non-language-based experimental paradigms in lack of suprathreshold clusters in the congruent > incon- which the N400 response has previously been described, gruent contrast. it is not feasible to compare the rN400 response to every other traditional paradigm in one study. Rather, the cur- Clinical implications rent study focused on comparisons with language-based Beyond the extensive laboratory based evaluations of paradigms, and utilized response features and character- N400, clinical applications are increasingly utilizing istics that have been identified as commonalities across the N400 response in a variety of patient populations. different studies in order to account for variable modali - The N400 is being particularly studied in disorders of ties and experimental parameters (e.g. experimental consciousness (DOC) as a potential marker of residual condition, stimulus duration and type, inter-stimulus Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 9 of 11 BC V3V 1Z2, Canada. ImageTech Lab, Surrey Memorial Hospital, 13750 96 Av, interval) [9, 17, 62]. Nonetheless, future studies may Surrey, BC V3V 1Z2, Canada. be conducted to examine more detailed comparisons between the brain vital sign rN400 response and tradi- Acknowledgements The authors would like to acknowledge the volunteers for their participation tional N400 responses. Additionally, as the first study of in the study. We also thank Matt Courtemanche for assistance with data col- rN400 response, the current study utilized a distributed lection as well as Mary-Carmen Graham and Qun Gao for administrative and source modeling approach for source localization to be technical support. consistent with previous MEG studies of N400 [24, 44]. Competing interests However, given the inherent limitations of this approach Some authors are associated with HealthTech Connex Inc. This may qualify in biasing sources towards the cortical surface, future them to financially benefit from the commercialization of the NeuroCatch platform capable of measuring brain vital signs. studies are needed to confirm these results using alter - nate source localization techniques such as spatial filter - Availability of data and materials ing using beamformer [63]. The data that support the findings of this study are available from the cor - responding author upon reasonable request. Consent for publication Conclusion Not applicable. In this study, we investigated the spatiotemporal and Ethics approval and consent to participate neuroanatomical features of the N400 response as elic- The study was approved by ethics boards at Fraser Health Authority and ited by the rapid assessment brain vital signs framework. Simon Fraser University, and all participants provided written informed consent. Using both MEG and EEG, our results showed that the rapidly elicited N400 response exhibits characteristics Funding consistent with those reported in traditional semantic This work was partly supported by a grant from Mathematics of Information Technology And Complex Systems (MITACS, grant #IT03240). This study was language-based N400 paradigms. These characteristics also supported by a grant from the Surrey Memorial Hospital Foundation include temporal features showing maximal response and a grant from Natural Sciences and Engineering Research Council (NSERC, within 300–500  ms latency; topographic scalp distribu- grant # RGPIN-2015-04018), awarded to TC. SGH was supported by the Multi-Year Funding scholarship from Simon Fraser University and CL holds a tion demonstrating maximal response at the posterior Canadian Institutes of Health Research (CIHR) CGS Doctoral scholarship (grant Pz electrode; spectral effects showing reduction in beta # GSD-140381). band power; and source localization to left-lateralized temporal and inferior frontal areas. With the increas- Publisher’s Note ing use of the N400 response in patient assessments for Springer Nature remains neutral with regard to jurisdictional claims in pub- lished maps and institutional affiliations. neurological conditions such as dementia and traumatic brain injury, the convergent M/EEG results of the cur- Received: 3 March 2018 Accepted: 26 May 2018 rent study provide further support for the possibility of translating the N400 response from research to clinical settings through a rapid assessment framework for evalu- ating cognitive functions. References 1. Luck SJ. An introduction to the event-related potential technique. Cam- Abbreviations bridge: MIT Press; 2014. ERP: event related potential; MEG: magnetoencephalography; BA: brodmann 2. Gawryluk JR, D’Arcy RCN. Electroencephalography: basic concepts and area; rN400: rapidly elicited N400; GFP: global field power; GLM: general linear brain applications. In: Handbook of physics in medicine and biology. New model; IFG: inferior frontal gyrus; TL: temporal lobe; EEG: electroencephalogra- York: CRC Press; 2010. pp. 24-1–14. phy; fMRI: functional magnetic resonance imaging; MCI: mild cognitive impair- 3. Duncan CC, et al. Event-related potentials in clinical research: guidelines ment; ICA: independent component analysis; MNE: minimum norm estimate; for eliciting, recording, and quantifying mismatch negativity, P300, and SVM: support vector machine; IPL: inferior parietal lobule; DOC: disorders of N400. Clin Neurophysiol. 2009;120(11):1883–908. consciousness. 4. Steppacher I, Eickhoff S, Jordanov T, Kaps M, Witzke W, Kissler J. N400 predicts recovery from disorders of consciousness. Ann Neurol. Author contributions 2013;73(5):594–602. Conceptualization and study design: SGH, CL, and RD. Literature search: SGH, 5. Olichney JM, Yang JC, Taylor J, Kutas M. Cognitive event-related poten- CL, and RD. Data collection: SGH, CL, SF, and TC. Analysis planning: SGH and tials: biomarkers of synaptic dysfunction across the stages of Alzheimer’s CL. Data analysis: SGH and CL. Result presentation: SG, CL, and RD. Analysis disease. J Alzheimers Dis. 2011;26(Suppl 3):215–28. outcome verification: SG, CL, XS, and RD. Result interpretation: All authors. 6. Fleck-Prediger C et al. Clinical applications of the Halifax conscious- Manuscript preparation: SG, CL, and RD. Critical editing and approval of sub- ness scanner: tracking recovery in a severely brain injured patient. mission: All authors. All authors read and approved the final manuscript. International Brain Injury Association, International NeuroTrauma Let- ter, 2015. http://www.inter natio nalbr ain.org/artic les/int-issue -37/ Author details 7. Beukema S, et al. A hierarchy of event-related potential markers of Faculty of Applied Science, Simon Fraser University, Burnaby, BC, Canada. auditory processing in disorders of consciousness. NeuroImage Clin. Health Science and Innovation, Surrey Memorial Hospital, Fraser Health 2016;12:359–71. Authority, Surrey, BC, Canada. HealthTech Connex Inc, Surrey, BC, Canada. 8. Kutas M, Hillyard SA. Reading senseless sentences: brain potentials reflect Surrey NeuroTech Lab, Surrey Memorial Hospital, 13750 96 Avenue, Surrey, semantic incongruity. Science. 1980;207(4427):203–5. Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 10 of 11 9. Lau EF, Phillips C, Poeppel D. A cortical network for semantics:(de) con- 33. Wilson KR, O’Rourke H, Wozniak LA, Kostopoulos E, Marchand Y, Newman structing the N400. Nat Rev Neurosci. 2008;9(12):920–33. AJ. Changes in N400 topography following intensive speech language 10. Connolly JF, Phillips NA, Stewart SH, Brake WG. Event-related potential therapy for individuals with aphasia. Brain Lang. 2012;123(2):94–103. sensitivity to acoustic and semantic properties of terminal words in 34. Sculthorpe-Petley L, et al. A rapid event-related potential (ERP) sentences. Brain Lang. 1992;43(1):1–18. method for point-of-care evaluation of brain function: develop- 11. D’Arcy RCN, Connolly JF, Service E, Hawco CS, Houlihan ME. Separating ment of the Halifax Consciousness Scanner. J Neurosci Methods. phonological and semantic processing in auditory sentence processing: 2015;245:64–72. a high-resolution event-related brain potential study. Hum Brain Mapp. 35. Wang JT, Young GB, Connolly JF. Prognostic value of evoked responses 2004;22(1):40–51. and event-related brain potentials. Can J Neurol Sci. 2004;31(4):438–50. 12. D’Arcy RCN, Service E, Connolly JF, Hawco CS. The influence of increased 36. Sergent C, et al. Multidimensional cognitive evaluation of patients with working memory load on semantic neural systems: a high-resolution disorders of consciousness using EEG: a proof of concept study. NeuroIm- event-related brain potential study. Cognit Brain Res. 2005;22(2):177–91. age Clin. 2016;13:455–69. 13. Holcomb PJ, Neville HJ. Auditory and visual semantic priming in lexical 37. Ghosh Hajra S, et al. Developing brain vital signs: initial framework decision: a comparison using event-related brain potentials. Lang Cognit for monitoring brain function changes over time. Front Neurosci. Process. 1990;5(4):281–312. 2016;10:211. 14. Neville HJ, Mills DL, Lawson DS. Fractionating language: different neural 38. Davis PA. Eec ff ts of acoustic stimuli on the waking human brain. J Neuro - subsystems with different sensitive periods. Cereb Cortex (New York, NY, physiol. 1939;2(6):494–9. 1991). 1992;2(3):244–58. 39. Sutton S, Tueting P, Zubin J, John ER. Information delivery and the sen- 15. Bentin S, McCarthy G, Wood CC. Event-related potentials, lexical deci- sory evoked potential. Science. 1967;155(3768):1436–9. sion and semantic priming. Electroencephalogr Clin Neurophysiol. 40. Taylor WL. ‘Cloze procedure’: a new tool for measuring readability. J Bull. 1985;60(4):343–55. 1953;30(4):415–33. 16. Rugg MD. The effects of semantic priming and word repetition on event- 41. Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of related potentials. Psychophysiology. 1985;22(6):642–7. single-trial EEG dynamics including independent component analysis. J 17. Kutas M, Federmeier KD. Thirty years and counting: finding meaning in Neurosci Methods. 2004;134(1):9–21. the N400 component of the event-related brain potential (ERP). Annu 42. Skrandies W. Global field power and topographic similarity. Brain Topogr. Rev Psychol. 2011;62:621–47. 1990;3(1):137–41. 18. Kutas M, Federmeier KD. Electrophysiology reveals semantic memory use 43. Maris E, Oostenveld R. Nonparametric statistical testing of EEG-and MEG- in language comprehension. Trends Cognit Sci. 2000;4(12):463–70. data. J Neurosci Methods. 2007;164(1):177–90. 19. Roehm D, Bornkessel-Schlesewsky I, Schlesewsky M. The internal struc- 44. Halgren E, et al. N400-like magnetoencephalography responses modu- ture of the N400: frequency characteristics of a language related ERP lated by semantic context, word frequency, and lexical class in sentences. component. In: Chaos and complexity letters, vol. 2; 2007. pp. 365–95. Neuroimage. 2002;17(3):1101–16. 20. Wang L, et al. Beta oscillations relate to the N400 m during language 45. Pylkkänen L, Marantz A. Tracking the time course of word recognition comprehension. Hum Brain Mapp. 2012;33(12):2898–912. with MEG. Trends Cognit Sci. 2003;7(5):187–9. 21. Van Petten C, Luka BJ. Neural localization of semantic context 46. Liu CC, Ghosh Hajra S, Cheung TPL, Song X, D’Arcy RCN. Spontaneous effects in electromagnetic and hemodynamic studies. Brain Lang. blinks activate the precuneus: characterizing blink-related oscillations 2006;97(3):279–93. using magnetoencephalography. Front Hum Neurosci. 2017;11:489. 22. Haan H, Streb J, Bien S, Rösler F. Individual cortical current density recon- 47. Litvak V, Friston K. Electromagnetic source reconstruction for group stud- structions of the semantic N400 effect: using a generalized minimum ies. Neuroimage. 2008;42(4):1490–8. norm model with different constraints (L1 and L2 norm). Hum Brain 48. Friston KJ, Holmes AP, Worsley KJ, Poline J, Frith CD, Frackowiak RSJ. Sta- Mapp. 2000;11(3):178–92. tistical parametric maps in functional imaging: a general linear approach. 23. Helenius P, Salmelin R, Service E, Connolly JF, Leinonen S, Lyytinen H. Hum Brain Mapp. 1994;2(4):189–210. Cortical activation during spoken-word segmentation in nonreading- 49. He P, Wilson G, Russell C. Removal of ocular artifacts from electro-enceph- impaired and dyslexic adults. J. Neurosci. 2002;22(7):2936–44. alogram by adaptive filtering. Med Biol Eng Comput. 2004;42(3):407–12. 24. Maess B, Herrmann CS, Hahne A, Nakamura A, Friederici AD. Local- 50. Parvar H, Sculthorpe-Petley L, Satel J, Boshra R, D’Arcy RCN, Trappenberg izing the distributed language network responsible for the N400 TP. Detection of event-related potentials in individual subjects using sup- measured by MEG during auditory sentence processing. Brain Res. port vector machines. Brain Inform. 2014;2(1):1–12. 2006;1096(1):163–72. 51. P. Golland, B. Fischl. Permutation tests for classification: towards statistical 25. Dronkers NF, Wilkins DP, Van Valin RD, Redfern BB, Jaeger JJ. Lesion analy- significance in image-based studies. In: Biennial international conference sis of the brain areas involved in language comprehension. Cognition. on information processing in medical imaging. Berlin: Springer; 2003 pp. 2004;92(1):145–77. 330–341 26. Balconi MFAU, Arangio RFAU, Guarnerio C. Disorders of consciousness 52. Connolly JF, Phillips NA. Event-related potential components reflect and N400 ERP measures in response to a semantic task. J Neuropsychia- phonological and semantic processing of the terminal word of spoken try Clin Neurosci. 2013;25:237. sentences. J Cognit Neurosci. 1994;6(3):256–66. 27. Connolly JF, D’Arcy RCN. Innovations in neuropsychological assess- 53. Dikker S, Rabagliati H, Farmer TA, Pylkkanen L. Early occipital sensi- ment using event-related brain potentials. Int J Psychophysiol. tivity to syntactic category is based on form typicality. Psychol Sci. 2000;37(1):31–47. 2010;21(5):629–34. 28. D’Arcy RCN, et al. Electrophysiological assessment of language function 54. Alegre M, Gurtubay IG, Labarga A, Iriarte J, Malanda A, Artieda J. Alpha following stroke. Clin Neurophysiol. 2003;114(4):662–72. and beta oscillatory changes during stimulus-induced movement para- 29. Marchand Y, D’Arcy RCN, Connolly JF. Linking neurophysiological and digms: effect of stimulus predictability. NeuroReport. 2003;14(3):381–5. neuropsychological measures for aphasia assessment. Clin Neurophysiol. 55. Tallon-Baudry C. Oscillatory synchrony and human visual cognition. J 2002;113(11):1715–22. Physiol. 2003;97(2):355–63. 30. Neumann N, Kotchoubey B. Assessment of cognitive functions in severely 56. Pesonen M, Björnberg CH, Hämäläinen H, Krause CM. Brain oscillatory paralysed and severely brain-damaged patients: neuropsychological and 1–30 Hz EEG ERD/ERS responses during the different stages of an audi- electrophysiological methods. Brain Res Protoc. 2004;14(1):25–36. tory memory search task. Neurosci Lett. 2006;399(1):45–50. 31. Olichney JM, et al. Patients with MCI and N400 or P600 abnormalities are 57. Friederici AD. Towards a neural basis of auditory sentence processing. at very high risk for conversion to dementia. Neurology. 2008;70(19 Pt Trends Cognit Sci. 2002;6(2):78–84. 2):1763–70. 58. Hickok G, Poeppel D. Neural basis of speech perception. Hum Audit Syst 32. Wang K, Cheung EFC, Gong Q, Chan RCK. Semantic processing dis- Fundam Org Clin Disord. 2015;129:149–60. turbance in patients with schizophrenia: a meta-analysis of the N400 59. Friederici AD, Ruschemeyer SA, Hahne A, Fiebach CJ. The role of left component. PLoS ONE. 2011;6(10):e25435. inferior frontal and superior temporal cortex in sentence comprehension: Ghosh Hajra et al. J Transl Med (2018) 16:151 Page 11 of 11 localizing syntactic and semantic processes. Cereb Cortex (New York, NY, 62. Lau EF, Weber K, Gramfort A, Hämäläinen MS, Kuperberg GR. Spati- 1991). 2003;13(2):170–7. otemporal signatures of lexical-semantic prediction. Cereb Cortex. 60. Ford JM, et al. N400 evidence of abnormal responses to speech in Alzhei- 2016;26(4):1377–87. mer’s disease. Electroencephalogr Clin Neurophysiol. 1996;99(3):235–46. 63. Hillebrand A, Barnes GR. Beamformer analysis of MEG data. Int Rev Neuro- 61. Kiang M, Patriciu I, Roy C, Christensen BK, Zipursky RB. Test–retest reli- biol. 2005;68:149–71. ability and stability of N400 effects in a word-pair semantic priming paradigm. Clin Neurophysiol. 2013;124(4):667–74. Ready to submit your research ? Choose BMC and benefit from: fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions

Journal

Journal of Translational MedicineSpringer Journals

Published: Jun 4, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off