SEARCH

SEARCH BY CITATION

Keywords:

  • near-infrared spectroscopy;
  • prefrontal cortex;
  • reproducibility;
  • test–retest reliability;
  • verbal fluency task

Abstract

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. ACKNOWLEDGMENTS
  7. REFERENCES

Aim:  To determine whether intrasubject reproducibility could be observed in the frontal cortex and to assess the mental-health status of subjects in each session.

Methods:  We measured changes in oxygenated hemoglobin concentration ([oxy-Hb]) during a letter version of the verbal fluency task using near-infrared spectroscopy imaging in twenty healthy adults over two sessions approximately two months apart. Additionally, the mental-health status of the subjects in each session was evaluated according to the State–Trait Anxiety Inventory, the Zung Self-rating Depression Scale, the Profile of Mood States, and the revised edition of the Neuroticism–Extroversion–Openness Personality Inventory. The association between those scores and [oxy-Hb] changes during the verbal fluency task in each session was investigated.

Results:  Performance on the verbal fluency task was about equal across the two sessions, and frontal activation during the task was observed globally in approximately the same region. In the test–retest reliability, acceptable values were shown in both the Intraclass Correlation Coefficients of the mean [oxy-Hb] changes and the correlation coefficients of the whole waveforms for each subject in the two sessions. Mental-health status as measured by several questionnaires was within the healthy range, and no correlation with the frontal activation was seen, except in several channels.

Conclusion:  The current results suggest that the measurement experience exerted very little influence, except for in a very small region. In addition, the intrasubject reproducibility of frontal activation measured by multi-channel near-infrared spectroscopy was well demonstrated in mentally healthy subjects at intervals of two months.

NEAR-INFRARED SPECTROSCOPY (NIRS) is a non-invasive technique that measures changes in the relative hemoglobin concentration using near-infrared light.1 Near-infrared light at 700–1000 nm has a high transparency in a living body. Using this property, NIRS imaging can measure cortical function as assessed by the changes in cerebral oxygenated hemoglobin ([oxy-Hb]) and deoxygenated hemoglobin ([deoxy-Hb]) concentration.2–4

NIRS imaging has various advantages: near-infrared light is non-invasive and safe, NIRS has a high-time resolution of 0.1 s, and subjects can be examined under natural conditions in a sitting posture. Given these advantages, NIRS imaging has the potential to be applied to bedside monitoring of brain function in clinical settings. In fact, several studies have shown activation differences between healthy controls and patients with schizophrenia,5–7 mood disorder,5,8,9 post-traumatic stress disorder,10 panic disorder,11–13 and eating disorder.14 These investigations have revealed frontal dysfunction in psychiatric patients, and some studies have even shown differences in activation patterns by disorder. These studies are almost all of cross-sectional design. However, it is important for clinical applications to confirm inter-session variability in cerebral activation within individuals.

Several studies have reported multiple measurements of the cerebral cortex of healthy volunteers using NIRS. Sato et al.15 demonstrated sufficient within-subject reproducibility of the NIRS signal amplitude, the location of activation centers, and the time-course of activation for each subject in the sensorimotor cortex during a finger-tapping task at mean intervals of 6 months. Plichta et al.16 have also shown similar results in sensorimotor activation assessed by event-related design at intervals of 3 weeks.

Watanabe et al.17 examined the reproducibility of frontal activation during the word fluency task in five subjects at mean intervals of 6–7 months. This research concluded that the reliability of NIRS for these tasks was acceptable for use in clinical studies. Kono et al.18 also demonstrated the considerable replicability of [oxy-Hb] changes in the prefrontal cortex during a category version of the verbal fluency task over four repeated sessions at weekly intervals. Moreover, Schecklmann et al.19 confirmed the short- (3 weeks) and long-term (1 year) reliability of frontal function with a large sample. These previous studies have indicated acceptable reliability. In usual treatment, however, changes in the patient's condition are generally evaluated on a monthly basis. The test–retest reliability at intervals of several months has not been examined.

In this study, we measured the [oxy-Hb] changes during a letter version of the verbal fluency task using NIRS imaging in healthy adults over two sessions approximately two months apart to determine whether temporal changes could be observed in the frontal cortex. Additionally, past studies have not shown the mental-health status of subjects, although Schecklmann et al.19 proposed that the reduction of [oxy-Hb] changes in the second session could partly result from lower stress, arousal, and physiological activation. Thus, in the present study, the mental-health status of subjects in each session was described, and the association between the [oxy-Hb] changes during the verbal fluency task and the mental-health status in each session was investigated.

METHODS

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. ACKNOWLEDGMENTS
  7. REFERENCES

Subjects

Twenty healthy subjects (12 women and 8 men) participated. The mean age was 22.7 ± 1.95 years old. Handedness was evaluated with the Edinburgh Inventory,20 and all subjects were confirmed to be right-handed (laterality index: mean ± SD = 0.95 ± 0.07). No subjects were taking medication or had a history of psychiatric disorders at the time of participation. All subjects provided written informed consent prior to participation. This study was approved by the Ethical Review Board of the Mie University School of Medicine (approval date: 14 May 2004).

Clinical assessments

The subjects were screened for the presence or absence of the DSM-IV axis I disorder using the Mini International Neuropsychiatric Interview (MINI)21 in the first NIRS measurement. The Japanese version of the Social Readjustment Rating Scale (SRRS)22 and the 12-item General Health Questionnaire (GHQ-12)23 were performed to confirm that the subjects had not experienced striking life events or remarkable fluctuation of general mental-health status across the two sessions. The subjects were clinically evaluated further using the State–Trait Anxiety Inventory (STAI),24 the Zung Self-rating Depression Scale (SDS),25 the Profile of Mood States (POMS),26 and the revised edition of the Neuroticism–Extroversion–Openness Personality Inventory (NEO–PI–R)27 for assessment of their state and trait anxiety, subjective depression, stress regarding life events, mood states, and personality, respectively.

Activation task

A verbal fluency task (VFT) was adopted as the frontal activation task. The VFT used here consisted of a 30-s pre-task baseline period, a 60-s task period consisting of three 20-s blocks, and a 70-s post-task baseline period.5,7,8 During the task period, the subjects were instructed to generate as many words as possible beginning with a particular syllable. The initial syllables were changed every 20 s during the 60-s task period. The combinations of three initial syllables were changed across each of the sessions. The number of words correctly generated by the subjects was assumed as each subject's performance score on the task. In the pre-task and post-task baseline periods, the subjects were asked to repeat a train of syllables (‘/a/, /i/, /u/, /e/, /o/’). The subjects were directed to take a seat with their eyes open, and to avoid making any body movements.

NIRS measurements

For the NIRS measurement, a 52-channel NIRS system (Hitachi ETG-4000) was used. Near-infrared laser diodes with two wavelengths (695 nm, 830 nm) were used as the light sources. The 3 × 11 probes were attached to each subject's frontal area. The distance between each emitter and detector probe was set at 3.0 cm. The lowest probe was positioned along the Fp1-Fp2 line in accordance with the international 10/20 system.

The time resolution was set at 0.1 s. Changes in the relative [oxy-Hb] and [deoxy-Hb] concentration were measured. The obtained data were analyzed using the ‘Integral mode’: the pre-task baseline was determined as the mean over a 10-s period just before the task period, and the post-task baseline was determined as the mean over the last 10 s of a 70-s post-task period; in other words, the two baselines thus obtained were used, and linear fitting was performed using the data between the two baselines. The moving average method was adapted for the analyzed data to remove any short-term motion artifacts (moving average window: 5 s). The analysis focused on [oxy-Hb] changes, because in a previous study, [oxy-Hb] was demonstrated to be the more sensitive indicator of changes in cerebral blood flow in NIRS measurements.28

Experimental procedure

To assess the temporal changes, two sessions of NIRS measurements were performed at intervals of approximately 2 months (mean ± SD = 61.8 ± 7.06 days) for each subject. STAI-State was performed just before each NIRS measurement. Other assessments such as GHQ-12, STAI-Trait, SDS, SRRS, and POMS were performed around each NIRS measurement. The assessment of NEO–PI–R was performed between the first and second session.

Statistical analysis

Grand averaged waveforms of [oxy-Hb] and [deoxy-Hb] were obtained after elimination of the artifact channels.4 The average changes in [oxy-Hb] in each channel were calculated for the pre-task baseline (10 s), task (60 s), and post-task baseline (10 s) segments for each subject. The obtained data were analyzed by two-way repeated-measures anova with [oxy-Hb] changes serving as dependent variables, and ‘task segments’ (pre-task baseline, task, post-task baseline) and ‘sessions’ (first, second) serving as independent variables in all 52 channels. A false-discovery rate (FDR) procedure (q = 0.05) was used to control for alpha errors due to multiple comparisons.29

Next, in order to investigate the activation pattern changes between the two sessions, the [oxy-Hb] changes in the first session were compared to those of the second session along the time course of the task; for this series, the paired t-test was preformed, considering the grand-averaged waveforms every 0.1 s in each channel. For the correction of multiple comparisons, statistical significance was defined as the differences between two sessions reaching the 5% significance level across 100 consecutive sequential points (10 s) during the task periods in the paired t-test. To evaluate the similarity of [oxy-Hb] waveform shapes between the two sessions in each subject, Pearson's correlation coefficient was used between the whole waveforms in the two sessions for each subject.

In addition, to examine the test–retest reliability for single subjects and for the group, the Intraclass Correlation Coefficient (ICC, one-way random effect model) of the mean value during the task segments for each channel and all available channels were calculated, respectively.

Furthermore, we investigated the relationship between the [oxy-Hb] changes during the VFT and the mental-health status at the NIRS measurement (STAI, 2 factors; SDS, 1 factor; POMS, 6 factors; NEO–PI–R, 5 factors). First, the [oxy-Hb] changes during the task were divided into three parts (‘task1’, ‘task2’, and ‘task3’ for the first, second, and third 20-s periods, respectively) to examine the relationship between the frontal function and mental-health status in detail, because a significant difference was observed in a part of the waveform between the two sessions by paired t-tests (Table 1). The obtained [oxy-Hb] data were averaged for each subject over these three time parts, and Spearman's correlation coefficients were calculated (P < 0.0012 = P < 0.05/3 task parts/14 factors). If the problem of multiple comparisons led to an increase in the alpha error, we did not include an element of measurement channels into alpha correction, treating a single channel as a functional ‘cluster size’ based on Obata et al.30 All statistical analyses were carried out using spss version 15.0 for Windows.

Table 1.  Summary of the t-test results for oxygenated hemoglobin concentration changes
ChannelHemisphereStart–end (sec)Consecutive number
  • 1st < 2nd.

  • No significant differences.

  • time (sec): Consecutive number of seconds in the channels with statistical significance.

  • The origin (0 s) of the time axis is set to meet the start point of the verbal fluency task.

1st > 2nd
Task segments   
 11R14.4–81.066.7
  8L14.9–28.713.9
  1R15.4–54.739.4
  2R16.0–54.538.6
  5R16.3–37.321.1
 12R17.1–81.264.2
 22R20.5–31.711.3
 43R52.7–63.711.1
Post-task Segments   
 47R55.7–97.241.6
 48L56.3–74.418.2
 49L82.8–96.413.7
 38L84.5–94.410.0
 3784.7–95.510.9
 1685.2–96.411.3
 26R85.9–103.217.4

RESULTS

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. ACKNOWLEDGMENTS
  7. REFERENCES

Clinical assessments and task performance

Table 2 shows that task performance (i.e. the number of words generated during VFT) did not differ significantly between the two sessions, while it was slightly but not significantly increased in the second session compared to the first session. The number of life events, assessed by SRRS during the month prior to the measurement, also revealed no significant differences between the two sessions (1st: 1.10 ± 1.29, 2nd: 1.10 ± 1.07, t(19) = 0.00, P = 1.00). In addition, the general mental-health status between the first and second session, as assessed using GHQ-12 was shown to fall within the healthy range, and there were no significant differences between the two sessions in any of the assessment scores.

Table 2.  Differences of task performance, GHQ12, SRRS, SDS, STAI, POMS, and NEO–PI–R scores between first session and second session
 1st session2nd sessionP-value
  1. GHQ12, 12-item General Health Questionnaire; NEO–PI–R, Neuroticism–Extroversion–Openness Personality Inventory–Revised; POMS, Profile of Mood States; SDS, Self-rating Depression Scale; SRRS, Social Readjustment Rating Score; STAI, State–Trait Anxiety Inventory.

Task performance15.60 ± 4.9917.10 ± 4.940.075
GHQ123.06 ± 2.322.38 ± 2.920.251
SRRS24.58 ± 28.3524.05 ± 27.260.939
SDS37.33 ± 7.6237.39 ± 7.960.961
STAI State40.90 ± 7.8839.35 ± 8.460.381
STAI Trait45.80 ± 10.0244.75 ± 11.070.365
POMS Tension–anxiety11.85 ± 7.0312.35 ± 7.950.704
 Depression–dejection12.25 ± 10.6512.50 ± 11.250.899
 Anger–hostility12.15 ± 8.9211.35 ± 8.990.592
 Vigor13.30 ± 5.7313.15 ± 7.320.894
 Fatigue10.60 ± 5.5810.20 ± 7.140.719
 Confusion9.30 ± 5.208.50 ± 4.790.316
NEO–PI–R Neuroticism101.42 ± 26.06
 Extraversion109.68 ± 16.06
 Openness118.21 ± 18.02
 Agreeableness117.68 ± 16.05
 Conscientiousness111.00 ± 22.59

[Oxy-Hb] changes between two sessions

Figure 1 shows the grand mean waveforms of the [oxy-Hb] and [deoxy-Hb] changes in the first session and second session for each channel. For both sessions, these graphs indicate more [oxy-Hb] activation and [deoxy-Hb] deactivation in the task segments than in the pre-task and post-task segments.

image

Figure 1. The upper figures show the grand-averaged waveforms of oxygenated hemoglobin concentration ([oxy-Hb]) (red line) and deoxygenated hemoglobin ([deoxy-Hb]) (blue line) changes during the verbal fluency task in the first (left) and second (right) session. The lower images show topographical mappings of the grand mean [oxy-Hb] changes during the verbal fluency task at 20 and 50 s from the start of the task, and at 30 s after the completion of the task in each group. The red, green, and blue areas in the topographs indicate an increase, no change, and a decrease in [oxy-Hb], respectively.

Download figure to PowerPoint

The results of the repeated measures analysis for the [oxy-Hb] changes revealed the main effects of ‘task segments’ in 43 of 52 channels (CH1–3, 6, 8–14, 16, 18–22, 24–29, 31, 34–52, F = 10.50–71.35). These findings are indicative of increases in [oxy-Hb] along the task segments. The main effect of ‘sessions’ was found to be significant in seven channels (CH1, 2, 8, 10–12, 43, F = 4.94–10.16), and the interaction between ‘task segments’ and ‘sessions’ was significant in these seven channels (CH1, 2, 8, 10–12, 43, F = 4.69–9.26). However, no significant differences in these seven channels were observed after the FDR correction.

The results of the paired t-tests for the inter-session comparison of [oxy-Hb] changes during performance of the VFT are also shown in Table 1. The [oxy-Hb] increases during the task in the second session were significantly smaller in eight channels, mainly of the right superior region, as compared with those of the first session. The [oxy-Hb] increases of the post-task periods in the second session were significantly smaller in seven channels, mainly of the anterior prefrontal region, as compared with those of the first session. To evaluate the similarity of [oxy-Hb] waveform shapes between the two sessions in each subject, Figure 2 shows the whole waveforms for each subject in the two sessions. The results of Pearson's correlation coefficient between the time courses of [oxy-Hb] changes in the two sessions for each subject showed substantial similarity for most subjects with a high correlation coefficient (mean ± SD across the subjects: 0.77 ± 0.21).

image

Figure 2. The whole waveforms of oxygenated hemoglobin concentration ([oxy-Hb]) changes during the task for each subject in the first session (black line) and second session (gray line). In each graph, the vertical axis shows [oxy-Hb] changes (mMmm) and the horizontal axis shows the time course of the task. The verbal fluency task period is shown between the two vertical parallel bars. Sex, age, and Pearson's correlation coefficient of time courses between two sessions for each subject are shown in the upper right of each graph.

Download figure to PowerPoint

Test–retest reliability

In the group and repeated reliability, the ICC of the mean value of [oxy-Hb] changes during the task segments are indicated in Table 3. The ICC showed that acceptable reliability was observed for the mean of all available channels, while there was great variability of both single and average measure ICC among channels at a single-channel level.

Table 3.  Intraclass correlation coefficients (ICC) of the mean value of oxygenated hemoglobin concentration changes during the task segments in the two sessions
Single channels: mean of 52 channels (range)All available channels
Single measure ICCAverage measure ICCSingle measure ICCAverage measure ICC
0.43 (−0.14–0.77)0.59 (−0.32–0.87)0.570.73

Correlation with clinical assessments

The significant correlation coefficients for the [oxy-Hb] changes in each channel and each questionnaire score are summarized in Table 4. In the first session, the [oxy-Hb] changes of task1 in CH37 were significantly correlated with the Openness scores in the NEO–PI–R. In the second session, the [oxy-Hb] changes of task1 in CH1 were significantly correlated with the Extraversion scores in the NEO–PI–R, and the [oxy-Hb] changes of task2 in CH18 were significantly correlated with the Tension–Anxiety scores in the POMS. No significant correlations were observed between the other two questionnaires (SDS and STAI) and the [oxy-Hb] changes in any channels.

Table 4.  Summary of the significant correlation coefficients for [oxy-Hb] changes in each channel and each mental-health status
Task partsChannelQuestionnaireSessionSpearman's rhoP-value
  1. Task parts, task 1: 0–20 s, task 2: 20–40 s, task 3: 40–60 s.

  2. NEO, Neuroticism–Extroversion–Openness; POMS, Profile of Mood States.

task137NEO Openness1st0.834<0.0001
2nd0.1370.613
task11NEO Extraversion1st0.2870.264
2nd0.7540.0011
task218POMS Tension–Anxiety1st0.3820.118
2nd−0.6990.0011

DISCUSSION

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. ACKNOWLEDGMENTS
  7. REFERENCES

The present study examined [oxy-Hb] changes during VFT using NIRS imaging in healthy adults over two repeated sessions, and the association between the [oxy-Hb] changes during the VFT and mental-health status in each session. The task performance of the VFT was about equal in the two sessions, and the frontal activation during the task was observed globally in approximately the same channels. The results in the present study replicated those of previous studies examining the reliability of multi-time measurements.15–19 These findings suggest that there may be little change in frontal activation during the VFT in healthy participants.

As for the results of the time-course analysis, no significant differences between the two sessions were obtained in several channels using the mean [oxy-Hb] change values during the entire period. However, the [oxy-Hb] changes during the task in the first session were larger than in the second session in the right superior frontal region. This finding could be explained by the fact that novelties are processed in the right frontal cortex, while cognitive practices are processed in the left prefrontal cortex.31 The [oxy-Hb] changes in the post-task periods in the first session were partly larger than in the second session in the anterior prefrontal region. It is difficult to explain the mechanism of the difference in the post-task periods. Two previous studies have indicated an [oxy-Hb] re-increase in the post-task period in patients with schizophrenia.5,7 There might be a possibility that the frontal activation pattern in the post-task period as well as during the task represents some kinds of individual states such as lower stress, arousal, and physiological activation as pointed out by Schecklmann et al.19

In test–retest reliability, acceptable values were shown in both the ICC of the mean [oxy-Hb] changes and the correlation coefficients of the whole waveforms for each subject in the two sessions. The ICC analyses indicate high reliability at the group and region of interest (ROI) but not at a single channel or single subject, as was repeatedly confirmed in previous studies.16,18,19 The time-course similarity of the activation waveforms across the session within each subject, as shown in Figure 2, revealed that the waveform pattern was similar in the two sessions. The temporal characteristics of frontal activation might distinguish between subjects, rather than reflect sensitivity to mental-health status.

In fact, the mental-health statuses as measured by the questionnaires did not correlate with frontal activation in most channels. This is partly because the questionnaire scores in the present subjects showed only a small variation within a normal range. These results suggest that small fluctuations in the mental health status of a healthy person might not affect the frontal activation. In a small region of the frontal cortex, however, the [oxy-Hb] changes were significantly correlated with subscores of the personality and mood scale. The present association between ‘Openness’ and ‘Extraversion’ in the NEO–PI–R (or NEO Five Factor Inventory [NEO-FFI]) and the function of the frontal cortex roughly confirms several previous results,32,33 and the present findings speculate that the frontal function might be related to personality. However, the significant associations between the scores in the questionnaires and [oxy-Hb] changes should be observed in both sessions if personality is associated with frontal activation. A correlation between these scores and the [oxy-Hb] changes was found in only one session in the present study, and thus our findings are tentative. We did not perform a multivariate analysis to examine the relationship among these factors in the present study, because each score in the questionnaires showed small variation within the healthy range, and was not significantly correlated with frontal function, as described above.

There are several limitations to this study. First, we performed NIRS investigations just twice. Second, for the sake of simplicity, we examined only young adults: a previous study has shown age-dependent differences in cerebral activation using NIRS.34 Third, we did not examine the variation of the [deoxy-Hb] changes excluded from the analysis. Finally, the problem of multiple comparisons might have led to an increase in the alpha error in the analysis of the time-course changes and the association between frontal activation and mental-health status.

In conclusion, the current results suggest that the measurement experience had very little influence, except for in a very small region of the brain, and that the intrasubject reproducibility of frontal activation measured by multi-channel NIRS was acceptably demonstrated in mentally healthy subjects at intervals of two months. Further studies addressing multiple repeated measurements at longer intervals, and investigating healthy subjects who exhibit high scores in the questionnaire in terms of their mental-health status are needed to clarify the variation of frontal activation in healthy adults.

ACKNOWLEDGMENTS

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. ACKNOWLEDGMENTS
  7. REFERENCES

This work was supported by KAKENHI (a Grant-in-Aid for Scientific Research) in the Priority Areas ‘Applied Genomics’ from the Ministry of Education, Culture, Sports, Science and Technology of Japan awarded to Y. O. and H. T. (#17019029), and by a KAKENHI (a Grant-in-Aid for Scientific Research) (C) from the Japanese Society for the Promotion of Science awarded to H. T. (#19591349).

We gratefully acknowledge Mr Tetsuhei Takami of the Mie University School of Medicine for his collaboration.

REFERENCES

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. ACKNOWLEDGMENTS
  7. REFERENCES