Mié Matsui, PhD, Department of Neuropsychiatry & Neuropsychology, Graduate School of Medicine, University of Toyama, 2630 Sugitani, Toyama 930–0194, Japan. Email: firstname.lastname@example.org
Abstract Near-infrared spectroscopy (NIRS) is an optical method to determine oxygenated and deoxygenated hemoglobin concentration changes in the human cerebral cortex. The purpose of this study was to examine the hemodynamic response of the prefrontal area during words memory learning using NIRS. A total of 23 healthy subjects participated in the present study. Hemodynamic response in the prefrontal cortex was measured using a NIRS system. The number of words recalled and stimulus category repetition (SCR) were analyzed by the words memory learning task. During the words memory learning task, oxygenated hemoglobin concentrations increased and deoxygenated hemoglobin concentrations decreased. This typical pattern was maintained during each memory stage, but the degree of change of [oxyHb] during encoding from the first condition to the second condition was significantly larger than that during retrieval. This suggests that memory organization is facilitated during encoding of the first condition, and that the retrieval period through two conditions still involves more activation in the prefrontal area than the encoding period. An increase of [oxyHb] was not recognized and activation was inhibited when the strategy was applied. Subjects produced more SCR in the second condition than in the first condition in spite of strategy instruction. This result suggests healthy people can find out implicit category by themselves following learning even without instruction. There were no significant relationships between the behavior indices and the changes in hemoglobin. Further studies are needed to clarify usefulness of NIRS in patients with psychiatric disorder.
Near-infrared spectroscopy (NIRS) is a method for measuring concentration changes in local hemoglobin in which near-infrared radiation is applied to the living body. NIRS is non-invasive, portable and has excellent time resolution, compared to brain function imaging methods such as positron emission tomography (PET), single photon emission computed tomography (SPECT), and functional magnetic resonance imaging (fMRI). Because the absorption curve between oxyHb and deoxyHb is different depending on the wavelength of the near-infrared radiation, it is possible to measure concentration changes of oxyHb and deoxyHb using near-infrared radiation. Total hemoglobin (totalHb) shows local blood flow volume, and change of hemoglobin reflects neural activity.1–3
It has been suggested that there exists a relationship between NIRS measurements and higher brain function after Jöbsis4 reported on measurement of the human brain for the first time. Hoshi et al.2 reported an increase in [oxy- and totalHb] that occurs during a calculation task using NIRS, and found that such a change in the hemoglobin concentration is remarkable when the task is difficult. Also, Fallgatter et al.5 observed activation of bilateral prefrontal lobe during the Wisconsin Card Sorting Test, and Hermann et al.6 reported activation of prefrontal function during verbal fluency tasks. Recently, NIRS has been applied to patients with various brain disorders, including Alzheimer’s disease,7 poststroke aphasia,8 brain tumors9 and psychiatric disorders.10 For instance, it was found to cause an inefficient increase in hemoglobin after the task was completed in patients with schizophrenia.10 However, there has been little work involving the use of NIRS in the field of memory research, and the hemodynamics during memory tasking are not well-understood.
The purpose of this study was to examine the cerebral hemodynamics of the prefrontal area during a word memory task using NIRS in healthy adults. The words memory test was the Japanese verbal learning task for which activation of the prefrontal lobe was previously examined by SPECT.11 The authors’ first goal was to examine what degree of oxygen metabolization occurs in the prefrontal area during the word memory task. The authors predicted that an increase in [oxyHb] and a decrease in [deoxyHb] were detected during the task compared with the rest state. Second, the authors examined the activation of the prefrontal area during the task in terms of temporal changes. The authors investigated the effects of cerebral hemodynamics by examining the difference between encoding and retrieval as memory processes, and the facilitation of learning by repetitive presentation. Third, the authors investigated whether a change in cerebral hemodynamics takes place in the prefrontal area after giving memorization strategy instructions in advance. Finally, the relationship between memory performance and cerebral hemodynamics was examined. The authors predicted that the hemodynamic change would become more remarkable in the case of better performance.
A total of 23 subjects consisting of nine men and 14 women aged 20–26 years with a mean of 22.9 years and standard deviation (SD) of 2.8 years participated in the present study. Subjects were excluded if they had a history of psychiatric illness, head trauma, neurological illness, serious medical or surgical illness, or substance abuse. The subjects had estimated IQ of 106.0 (SD = 6.4) as assessed by the Japanese Adult Reading Test. All subjects were right-handed.
Written informed consent was obtained from each subject for participation in this investigation, which was approved by the Ethical Committee of University of Toyama.
A two-channel portable OM-220 NIRS system (Shimadzu, Kyoto, Japan) was used at a sampling time of 1.0 s and measured the intensity of light detected at two wavelengths, 780 and 830 nm and calculated the change in optical density. It then calculated the changes in [oxyHb] and [deoxyHb] according to the Beer–Lambert Law.12 Two sets of emission and detection probes were bilaterally attached to the forehead of each subject. Specifically, the detection probes were set at Fp1 and Fp2 corresponding to the international 10–2013 system of electrode placement with the emission probes being lateral at 4 cm on both sides at approximately F7 and F8, respectively. Okamoto et al.14 examined cranio-cerebral correlation using MRI via the guidance of the international 10–20 system for electrode placement. According to the findings, anatomical locations of Fp1, Fp2, F3 or F4 were mainly on left superior prefrontal gyrus (Brodmann’s area 10), right superior frontal gyrus (Brodmann’s area 10), left superior/inferior frontal gyrus (Brodmann’s area 10/47) or right superior/inferior frontal gyrus (Brodmann’s area 10/47), respectively. The cerebral tissue examined by NIRS is thought to be a banana-shaped region between the two optodes (a light emitter and a detector),15 which includes the dorsolateral prefrontal cortex.
The Japanese version of the verbal learning task16 is composed of a 16-word list. Four exemplars from each of four categories are constructed so that related items never consecutively appeared in the list. Therefore, the unblocked list is a memory task for implicitly categorized words and measures the degree of semantic organization. For this NIRS study, the authors used the unblocked lists composed of two equivalent versions. The list words were selected from common Japanese words, so that familiarity with each word was approximately the same. Three trials involving each list were repeated consecutively. The words were presented by means of a tape recording at a rate of one word within 1 s with 1 s intervals and the subjects were required to recall the words by writing them after each 16-word set was presented. The total number of words recalled in each list from the three trials was then calculated. Categorical clustering was also evaluated as stimulus category repetition (SCR) when recalling each list. The SCR was defined as the total immediately followed by an item in the same category in the recall. The total SCR scores were calculated from the performance of the three trials for each list.
Each subject was seated in a chair in a relaxed position resting on a fixed stage throughout the measurements to avoid movement artifacts. The authors employed the verbal learning task to activate the prefrontal cortex, since memory organization was previously used in several activation studies on the prefrontal cortex.11,17 In this study, the subjects were asked to recall the words in writing. The time-course of the verbal learning task with one block of NIRS measurements was as follows: (i) resting time of 60 s; (ii) writing 1 – control task which involved writing the Japanese syllabary ‘a, i, u, e, o . . .’ on white paper using a pencil for 30 s; (iii) encoding 1 – listening 16 words presented for 35 s; (iv) retrieval 1 – writing as many words as could be remembered on white paper using a pencil for 60 s; (v) encoding 2 – repeating encoding 1 for 35 s; (vi) retrieval 2 – repeating retrieval 1 for 60 s; (vii) encoding 3 – repeating encoding 1 for 35 s; (viii) retrieval 3 – repeating retrieval 1 for 60 s; (ix) writing 2 – repeating writing 1 for 30 s; and (x) resting time 2 for 60 s. These were the first conditions that were used. During the encoding phase, the subjects were instructed to try to memorize the words that they heard. They were not informed of the categories in the lists before the examination for the first condition. After the first condition, a brief line orientation test was substituted for the non-verbal task. Then, the second condition was started following the same time-course as for the first condition. The verbal leaning tasks consisted of two versions and were presented in random order to the subjects for the first or second condition. Before the second condition, group with strategy (n = 12) were taught 16 words that consisted of four categories and were instructed to memorize them using grouping the words. Group without strategy (n = 11) was not instructed with such a memorizing strategy before the second condition and were asked to repeat the same way as the first condition.
To examine the effect of learning, the number of recalls and SCR were analyzed by two-way repeated measures anova, with condition first or second and trial 1, 2 or 3 as within-subject factors in group without strategy. To examine the effect of strategy, the number of recalls and SCR for the second condition were analyzed by two-way repeated measures anova, with group with or without strategy instruction as the between-subject factor and trial 1, 2 or 3 as the within-subject factor.
Near-infrared spectroscopy data
To examine whether there were statistically significant changes in the NIRS parameters (oxy-Hb and deoxy-Hb), the authors first individually compared the data during baseline 30 s before the writing task and that of each task period using the Wilcoxon’s rank order test. For each subject, the average changes in [oxyHb] and [deoxyHb] were calculated for baseline and encoding period as well as for the writing and retrieval periods. A memory process (encoding, retrieval) × hemoglobin (oxy-, deoxy-) anova per hemisphere was conducted on the changes for the first condition. To examine differences of [oxyHb] and [deoxyHb] between the strategy instruction and no instruction before the second condition, memory process (encoding, retrieval) × condition (1st, 2nd) × hemoglobin (oxy-, deoxy-) anova were conducted on the changes in group with strategy or group without strategy.
Correlation between behavioral and near-infrared spectroscopy data
To ensure that the NIRS results reflected the important aspects of task performance, the authors calculated Pearson’s product-moment correlation coefficient between the memory performance and NIRS data.
The results of performance in the Japanese version of the verbal learning task are shown in Table 1. The results of anova for the number of recalls showed a significant main effect of the trial (F(2,40) = 33.091, P < 0.001) in group without strategy, but no significant main effect of condition (F(1,20) = 1.509, n.s.) and interaction (F(2,40) = 1.640, n.s.). With regard to SCR, there were significant main effects of condition (F(1,20) = 4.635, P < 0.05) and trial (F(2,40) = 9.211, P < 0.05) in group without strategy, but no significant interaction of condition by trial (F(2,40) = 1.385, n.s.). Regarding the effects of strategy, there was no significant main effect of strategy instruction for the number of recalls (F(1,21) = 0.079, n.s.) and SCR (F(1,21) = 1.555, n.s.).
Table 1. Number of recalls and stimulus category repetition
SCR, stimulus category repetition.
No instruction (n = 11)
Instruction (n = 12)
No instruction (n = 11)
Instruction (n = 12)
Near-infrared spectroscopy results
The typical time courses of [oxyHb] and [deoxyHb] during the tasks are shown in Fig. 1. [OxyHb] significantly increased and [deoxyHb] significantly decreased through the task for first condition (Right [oxyHb], baseline–encoding, T = 24, P < 0.001; writing–retrieval, T = 48, P < 0.005; Right [deoxyHb], baseline–encoding, T = 5, P < 0.001; writing–retrieval, T = 0, P < 0.001; Left [oxyHb], baseline–encoding, T = 30, P < 0.001; writing–retrieval, T = 46, P < 0.005; Left [deoxyHb], baseline–encoding, T = 3, P < 0.001; writing–retrieval, T = 19, P < 0.001). For second condition, the increase in [oxyHb] was not significant and [deoxyHb] significantly decreased (Right [deoxyHb], baseline–encoding, T = 56, P < 0.05; Left [deoxyHb], baseline–encoding, T = 21, P < 0.001; writing–retrieval, T = 61, P < 0.05).
Figure 2 shows the change in the concentrations of oxyHb and deoxyHb for the first condition. There was a main effect of hemoglobin in either hemisphere (right: F(1,44) = 48.747, P < 0.001; left: F(1,44) = 52.473, P < 0.001). However, there was no significant main effect of memory process (right: F(1,44) = 0.020, n.s.; left: F(1,44) = 0.001, n.s.) and no interaction of memory process with hemoglobin (right: F(1,44) = 2.383, n.s.; left: F(1,44) = 3.052, n.s.) in either hemisphere.
The typical time courses of [oxyHb] and [deoxyHb] in the group with strategy instruction and the group without it are shown in Figs 3 and 4, respectively. The change in [oxyHb] and [deoxyHb] due to strategy instruction is shown in Fig. 5. This analysis was conducted on the left hemisphere since there were similar results between hemispheres according to above analysis and this task is related to language. In group with strategy instruction, the results of anova showed a significant main effect of hemoglobin (F(1,22) = 7.672, P < 0.011), but no significant main effects of memory process (F(1,22) = 0.604, n.s.) and condition (F(1,22) = 0.903, n.s.). There was a significant interaction of condition by hemoglobin (F(1,22) = 4.346, P < 0.05), while there were no other significant interactions. In group without strategy instruction, the results of anova showed significant main effects of hemoglobin (F(1,20) = 23.244, P < 0.001) and condition (F(1,20) = 8.332, P < 0.01), but no significant main effect of memory process (F(1,20) = 0.279, n.s.). There were significant interactions of condition by hemoglobin (F(1,20) = 14.363, P < 0.01) and condition by memory process (F(1,20) = 6.926, P < 0.05), while there were no other significant interactions.
Correlation between behavioral and near-infrared spectroscopy data
Neither the number of recall or SCR was significantly correlated with any change in [oxyHb] or [deoxyHb].
In this study, the authors examined the cerebral hemodynamics of the prefrontal area during a word memory task, and activation in the prefrontal area during this task was found to be similar to that of a previous neuropsychological task.18 This implies that word memorizing performance is involved in prefrontal function. Since this word memory task is sensitive to memory organization, it is likely that NIRS measures the function of memory organization in the prefrontal area. The present results overall do not contradict those of previous studies as a whole.11,17
Although activation of the prefrontal lobe occurred during the task for the first condition, the change was not clear for the second condition. That is why it might not be necessary to be involved any further in the prefrontal lobe during the task due to the practice effect for the second condition. Ohtani et al.19 has also reported that the level of increase in oxy Hb was reduced when the tasks were repeated. Moreover, it might have been easy to carry out the task for healthy adults. This result is in accord with the difference in the hemoglobin change that is dependent on the difficulty of the task. A typical pattern of increased [oxyHb] and decreased [deoxyHb] was maintained during each memory stage, but the degree of decreased change of [oxyHb] during encoding from the first condition to the second condition was significantly larger than that during retrieval. It follows from this that the encoding period during first presentation of stimulus involves more activation in the prefrontal area than does the encoding period during second presentation, and that memory organization is facilitated during encoding of the first condition. In addition, the retrieval period through two conditions still involves more activation in the prefrontal area than the encoding period.
The authors found that there were significant changes in hemoglobin due to strategy instruction. An increase in [oxyHb] was not recognized and activation was inhibited when the strategy was applied. The strategy is a clue for planning of executive function assuming that the prefrontal lobe is involved in it, and it was predicted that subjects can perform the task more easily if they are given the strategy in advance. However, the present results showed there was no significant difference between group with strategy instruction and group without it for the number of recalls and SCR. In contrast, both groups produced more SCR in the second condition than in the first condition, although there was no significant difference of the number of recalls between conditions on either group. This result suggests healthy people can find out implicit category by themselves following learning even without instruction. This and the practice effect might yield the present result. Another possible explanation for this no response is that hemodynamic responses observed in the first condition might be non-specific changes mainly attributed to systemic circulation changes. If so, two groups with or without strategy instruction might have shown the same pattern depending on conditions. However, the present results showed [oxyHb] in the second condition between two groups had a different pattern. Some previous studies reported that the more subjects increased their attention to the tasks, the more the BOLD signal or oxyHb concentration in the prefrontal cortex decreased.20,21 The feature of these changes in healthy people will be useful for examining the effects of intervention during the cognitive rehabilitation of patients with mental disease in the future.
There were no significant relationships between the behavior indices and changes in hemoglobin. The results of lack of this association during words memory might indicate that the responsible area does not cover prefrontal area related to memory organization. The anterior part of the prefrontal cortex, including Brodmann area 10 and 47, seems to be the main area of contribution to the NIRS measurement in the present study. Previous studies11,17 have indicated that the left inferior prefrontal cortex (Brodmann area 45, 46, 47) has been found to show increased activation during semantic encoding. In the present method, only two bilateral recoding sites on the forehead were used, and the recoding would be limited to a part of the prefrontal cortex. Further study should cover inferior frontal cortex using multichannel NIRS and investigate activation of memory organization. Another point is that the subjects might efficiently perform the task without activation of the prefrontal area since they became used to doing it, and this point should be taken into account for selection of the task in the future.
Periodic changes that do not depend on the task may influence measurements of NIRS, and such voluntary changes may be influenced by breathing and the heart beat,22 and it is important to take into account the influence of such a physiological change when measurements are interpreted. In the authors’ analysis, they used the calculation methods to remove a physiological change following Yonezawa et al.23 Namely, linear baseline correction with use of the mean value during the resting period before the entire tasks and the last resting period after the entire tasks was applied, to remove any longitudinal signal drift.
In contrast, it is useful in clinical practice, owing to its advantages of non-invasiveness, high time resolution, portability, the possibility of performing examinations in a natural setting and its low running cost. The authors can repeatedly use NIRS, because it is non-invasive, therefore, allowing it to be used in drug therapy or cognitive rehabilitation for psychiatric or neurological patients in the future.
This study examined activation in prefrontal area during performance of a word memory task using NIRS, and found that [oxyHb] increases and [deoxyHb] decreases. This typical pattern was maintained during each memory stage, but the degree of change of [oxyHb] during encoding from the first condition to the second condition was significantly larger than that during retrieval. This suggests that memory organization is facilitated during encoding of the first condition, and that the retrieval period through two conditions still involves more activation in the prefrontal area than the encoding period. This was significant for changes in the hemoglobin due to the strategy instruction. An increase of oxyHb was not recognized and activation was inhibited when the strategy was applied. The strategy is a clue for the planning of executive function assuming that the prefrontal lobe is involved in it, and subjects produced more SCR in the second condition than in the first condition in spite of strategy instruction. This result suggests healthy people can find out implicit category by themselves following learning even without instruction. There were no significant relationships between the behavior indices and the changes in hemoglobin. Further studies are needed to clarify usefulness of NIRS in patients with psychiatric disorder.
This study was supported by a Grant-in-Aid for Scientific Research (C) (2), 16530445 from the Japan Society for the Promotion of Science.