https://orcid.org/0000-0002-7704-4568
Figures
Abstract
This study was performed to investigate the usefulness of functional near-infrared spectroscopy (fNIRS) by conducting a comparative analysis of hemodynamic activation detected by fNIRS and positron emission tomography (PET) and magnetic resonance imaging (MRI) in patients with mild cognitive impairment (MCI) and Alzheimer’s disease (AD). Participants were divided into four groups: the subjective memory impairment (SMI), amnestic MCI (aMCI), non-amnestic MCI (naMCI), and AD groups. We recorded the hemodynamic response during the semantic verbal fluency task (SVFT) using a commercial wireless continuous-wave NIRS system. The correlation between the parameters of the neuroimaging assessments among the groups was analyzed. Region of interest-based comparisons showed that the four groups had significantly different hemodynamic responses during SVFT in the bilateral dorsolateral prefrontal cortex (DLPFC). The linear mixed effect model result indicates that the mean ΔHbO2 from the bilateral DLPFC regions showed a significant positive correlation to the overall FDG-PET after controlling for age and group differences in the fNIRS signals. Amyloid PET signals tended to better differentiate the AD group from other groups, and fNIRS signals tended to better differentiate the SMI group from other groups. In addition, a comparison between the group pairs revealed a mirrored pattern between the hippocampal volume and hemodynamic response in the DLPFC. The hemodynamic response detected by fNIRS showed a significant correlation with metabolic and anatomical changes associated with disease progression. Therefore, fNIRS may be considered as a screening tool to predict the hemodynamic and metabolic statuses of the brain in patients with MCI and AD.
Citation: Yoon JA, Kong IJ, Choi I, Cha J, Baek JY, Choi J, et al. (2023) Correlation between cerebral hemodynamic functional near-infrared spectroscopy and positron emission tomography for assessing mild cognitive impairment and Alzheimer’s disease: An exploratory study. PLoS ONE 18(8): e0285013. https://doi.org/10.1371/journal.pone.0285013
Editor: Stephen D. Ginsberg, Nathan S Kline Institute, UNITED STATES
Received: September 15, 2022; Accepted: April 13, 2023; Published: August 10, 2023
Copyright: © 2023 Yoon et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Relevant data are within the paper and its Supporting Information files.
Funding: The authors received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Alzheimer’s dementia is the most common type of neurodegenerative disease, accounting for approximately 70% of all cases of dementia [1]. It is estimated that 15–20% of adults aged 65 years or more have mild cognitive impairment (MCI) [2], which progresses to Alzheimer’s dementia in approximately 30% of the cases. Currently, it is reported that initiating drug therapy at the stage of MCI is the most effective for slowing its progression to Alzheimer’s dementia [3]. The two-hit hypothesis, a hypothesis based on the combination of the vascular and amyloid cascade, postulates that amyloid-beta (Aβ) deposition, which occurs after vascular dysfunction, causes neurodegeneration and cognitive decline during the progression from MCI to Alzheimer’s dementia [4]. Accordingly, cerebral hypoperfusion is an important biomarker for determining the presence of the disease, as well as the risk for progression [3,5]. Positron emission tomography (PET) and single-photon emission computed tomography imaging are useful for assessing the metabolic activity accompanying cerebral blood perfusion and predicting the progression of MCI to Alzheimer’s disease (AD) with high sensitivity and specificity [6–8]. However, imaging takes a long time and is expensive. Thus, it is often the case that testing is not performed at the right time, and patients are not diagnosed and treated early. Multi-channel functional near-infrared spectroscopy (fNIRS) is a promising alternative for the early diagnosis of AD and MCI. Several studies have examined hemodynamic responses during MCI and AD using fNIRS and showed decreased levels of activation in specific brain regions of patients with AD relative to the control group [9–12]. In addition, fNIRS has also been used to evaluate hemodynamic impairments in AD during the resting state, which are unrelated to functional activity [13,14].
Therefore, by confirming the pattern of hemodynamic responses in AD and MCI, useful biomarkers assessable by fNIRS for the early diagnosis of MCI may be discovered. PET assesses changes in metabolic activity accompanied by changes in cerebral blood flow [15]. Establishing the relationships between metabolic activity, structural changes, and amyloid plaque load detected with PET and magnetic resonance imaging (MRI) and hemodynamic responses detected with fNIRS would help in evaluating the reliability of fNIRS. However, no studies have been conducted to investigate these relationships. The present study aimed to investigate the usefulness of fNIRS by conducting a comparative analysis of hemodynamic activation detected by fNIRS and PET in patients with MCI and AD.
Materials and methods
Study participants
The study participants were patients aged 60 years or more who visited the psychiatry clinic of the study hospital due to cognitive impairment between June 2015 and July 2019. A final diagnosis was confirmed by a psychiatrist using a multidisciplinary approach involving medical examination, neuropsychological and neuroimaging assessments, and neurocognitive tests such as the Seoul Neuropsychological Screening Battery (SNSB) 2nd edition, brain MRI, and F-18 (flutemetamol) amyloid PET. Based on the results of the evaluation, the participants were allocated to the subjective memory impairment (SMI), amnestic MCI (aMCI), non-amnestic MCI (naMCI), and AD groups. Patients with AD met the National Institute of Neurological and Communication Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association criteria for probable AD [16]. Patients with MCI met the following Petersen’s criteria [17]: (a) subjective memory complaints by patients or caregivers, (b) normal activities of daily living (ADL) based on clinical findings and the ADL scale (S-IADL) [18], (c) objective memory impairment based on the word list delayed recall test (Seoul Verbal Learning Test, delayed recall) as evidenced by a Z-score of 1.5 below the mean of the age-, sex- and education-matched normal individuals, and (d) the absence of dementia. Patients with (a) above-moderate dementia severity (Clinical Dementia Rating ≥ 2); (b) an axis I diagnosis of delirium, schizophrenia, bipolar disorder, and major depressive disorder; (c) clinically active cerebrovascular disease (e.g., stroke within 6 months, multiple lacunae, and severe white matter hyperintensities with Fazekas scale score = 3) or other conditions causally related to cognitive impairment (e.g., severe organ failure, metabolic or hematologic disorders and clinically significant abnormal laboratory findings); and (d) other neurodegenerative disorders (dementia with Lewy bodies, frontotemporal dementia, and Parkinson’s disease) were excluded. This study was approved by the Institutional Review Board of our hospital (IRB No. 2204-001-113). All methods were performed in accordance with the relevant guidelines and regulations. The patients were enrolled in the study after they had provided written informed consent.
fNIR data acquisition and processing
A total of 107 participants were divided into four groups: SMI (n = 24), aMCI (n = 30), naMCI (n = 29), and AD (n = 24). We recorded the hemodynamic response during the protocol using a commercial wireless continuous-wave NIRS system (OBELAB Inc., Seoul, Republic of Korea) [19,20].
Participants wore the NIRS system, the neuroimaging device for fNIRS, and performed a verbal fluency task. The task was administered over three sessions, with a 30-s break between sessions (Fig 1). Semantic verbal fluency task (SVFT) [21] involves generating as many words as possible within a certain time frame and providing hints about the semantic category of the words generated. The amount of information related to categorization and the number of words that can be retrieved from memory within 1 min are determined. In this study, the task involved generating as many words related to the keywords as possible.
SVFT, semantic verbal fluency task.
Hemodynamic response of the prefrontal cortex was recorded using a high-density NIRS device which was composed of 24 sources (laser diodes) emitting two wavelengths (780/850 nm) and 32 photo-detectors, at a sampling rate of 8.138 Hz [22]. The total number of channels was 48, and the detected light signals for each wavelength were filtered by a band-pass filter (0.005–0.1 Hz) to minimize environmental noise-related light and physiological noise due to body movement (Fig 2). The poor-quality channels (signal-to-noise ratio<30 dB) were rejected before extraction of the hemodynamics data to prevent misinterpretation. The relative hemodynamic changes of each channel during each trial of the tasks were assessed separately using the modified Beer–Lambert law [23]. The results of multiple trials were block-averaged individually before being grand-averaged for each group. As the BOLD-oxygenated hemoglobin (oxy-Hb/HbO2) correlation showed higher correlation than BOLD-deoxy-Hb correlation in previous study [24], the oxy-Hb/HbO2 values (ΔHbO2) during the task period were measured as representing the activation of the prefrontal cortex. To correct the motion artifacts due to the differential pathlength factor of the participants, correlation-based signal improvement (CBSI) correction was performed before analysis after measurement [25].
MRI data acquisition and image analysis
Each participant underwent a structural T1-weighted MRI and a diffusion tensor imaging scan at the clinical evaluation. All images were acquired at Pusan National University Hospital Imaging Center on a Siemens 3-T Trio TIM scanner (Erlangen, Germany).
For each participant’s cortical thickness or volume estimation, a 3-D magnetization-prepared rapid gradient echo sequence was acquired using the following parameters: repetition time = 1800 ms, echo time = 2.07 ms, flip angle = 12˚, acquisition matrix = 256 × 256, field of view = 250 × 250 mm2, slice thickness = 1 mm, and the total number of slices = 256.
All image acquisitions had the same slice orientation parallel to the anterior commissure and posterior commissure line. Motion was restricted with expandable foam cushions. Scans with movement or any other image (reconstruction) artefacts were excluded.
The FreeSurfer version 5.1 software package (http://surfer.nmr.mgh.harvard.edu/) was used to analyze the cortical thickness or volume on 3-D T1-weighted images [26]. Each brain region was identified using the Desikan–Killiany Atlas [27]. We visually examined all images to ensure segmentation accuracy.
Florbetaben PET acquisition and imaging analysis
All participants underwent amyloid PET scans with a Biograph 40 PET/CT scanner (Siemens, Knoxville, TN, USA). Amyloid PET data were acquired between 90 and 110 mins after the injection of 185 MBq of 18F-florbetaben. Amyloid PET images were reconstructed using the iterative ordered-subset expectation maximization algorithm [28].
Quantitative analysis was conducted by normalizing the MRI images to a T1-weighted MRI template and co-registering PET images with the MRI images in all cases. Each brain region was identified using the Automated Anatomical Labeling atlas [29]. We calculated the level of Aβ deposition in each brain region and expressed it as standard uptake value ratios (SUVRs), by using the whole cerebellum as a reference region. Amyloid status (Aβ positive or negative) was determined using a previously described method based on a visual assessment by a nuclear physician blinded to clinical data [30].
18F-FDG PET acquisition and imaging analysis
All participants underwent PET scans using a Biograph 40 PET/CT scanner (Siemens, Knoxville, TN, USA). PET data were acquired 60 mins after the injection of 4.8 MBq/kg of 18F-FDG. PET images were reconstructed using the iterative ordered-subset expectation maximization algorithm. Quantitative analysis was conducted by normalizing the MRI images to a T1-weighted MRI template and co-registering PET images with the MRI images in all cases. We calculated the brain glucose metabolism of each brain region and expressed it as SUVRs, using the global brain area as a reference.
Statistical analysis
Region of interest.
Before performing any statistical tests, the mean ΔHbO2 of each channel during SVFT was averaged to form eight Brodmann regions to yield more stable signals from each participant and to reduce the number of comparisons in subsequent tests. The resulting eight regions of interest (ROIs) include the right and left dorsolateral prefrontal cortex (DLPFC), right and left frontopolar cortex, right and left orbitofrontal cortex (OFC), and right and left ventrolateral prefrontal cortex (VLPFC). Since the majority of the channels in bilateral VLPFC were rejected during the preprocessing step, formal statistical tests were performed on regions other than bilateral VLPFC.
Due to the exploratory nature of the study, we started with all features within the dataset, including 6 fNIRS Brodmann regions, 8 aPET measures, 8 fluorodeoxyglucose (FDG)-PET measures, 88 volume measures, and 68 thickness measures, totaling 180. From this dataset, two sets of fNIRS regions were selected as fNIRS ROIs: (1) right and left OFC, where participants (collapsing across groups) showed significant activation during SVFT, and (2) right and left DLPFC, where the main effect of group was significant, demonstrating a greater activation in the SMI group than other groups. The analysis of variance (ANOVA) results are elaborated in the results section.
Feature selection.
Two formal tests were used to select the features to be submitted for further analyses. First, as a formal test to investigate the relationship between the mean ΔHbO2 from the fNIRS and other neurological measures, linear mixed effect (LME) models predicting the mean ΔHbO2 with each measure were fitted, including subjects’ age as a covariate and the random intercept of the group factor to account for differences in the fNIRS signal across the group that can potentially overshadow the relationship between the two measures. Thus, the LME model for the formal test is shown below.
fNIRS (mean ΔHbO2) ∼ B0+B1* Neurological Measurement + +B2*Age + (1|Group). Second, multivariate analysis of covariance (MANCOVA) tests were performed to see if the group factor (diagnosis) had any differential effect on fNIRS signals and neurological measures, implying that fNIRS can potentially provide additional information above and beyond what conventional neurological measures could provide. MANCOVA was selected to reduce the number of tests required and to scope in detail in a top-down method, and a method of selecting and reporting areas and indicators that showed significant effects in this model was selected. In reporting the test results, when both the right and left parts of the same fNIRS ROI were statistically significant, the mean of the two ROIs was averaged, and the result from the bilateral ROIs was reported. Otherwise, original results from either right or left ROI were illustrated. The threshold for all statistical significance was set at p < .05.
Results
Table 1 shows the number of subjects in each group who had data from each imaging study. The sample size for each test (either LME or MANOVA) varied because the number of subjects who had fNIRS and other measurements differed across the pairs of methods to be compared. Additionally, the outliers beyond the 1.5 interquartile range within each group for each comparison method were also removed per analysis.
There were no significant differences in age, sex ratio, or education level between the groups (all p values < .05) (Table 2). The results of the initial neuropsychological test were significantly different between the groups (Table 3).
Bilateral OFC ROIs showed significant activation during the SVFT test when the subjects were collapsed across groups demonstrating that the orbitofrontal cortex is the functional ROI that showed a task-related activation [t(105) = 8.77, p<0.001) (Fig 3). Furthermore, one-way between-subject ANCOVA was conducted to examine the effect of the diagnosed group on the mean ΔHbO2 of each region after controlling for subject age. The main effect of the group was significant in the left DLPFC [F(3, 101) = 2.79, p = 045], right DLPFC, [F(3, 99) = 2.85, p = 041], and bilateral DLPFC as a whole [F(3, 98) = 3.01, p = 034] (Fig 4). Dunnett’s post-hoc test on groups revealed that the SMI group showed greater bilateral DLPFC activation during SVFT than the naMCI (p = .032), AD (p = .027), and aMCI groups (marginally greater; p = .061) (Fig 5).
The ventral part of the PFC involving the bilateral orbitofrontal cortex showed significant activation. fNIRS: Functional near-infrared spectroscopy, SVFT, semantic verbal fluency task, PFC: Prefrontal cortex.
The dorsal part of the PFC involving the bilateral dorsolateral cortex shows a significant group effect. fNIRS: Functional near-infrared spectroscopy, SVFT, semantic verbal fluency task, PFC: Prefrontal cortex.
SVFT, semantic verbal fluency task.
fNIRS and FDG-PET
The LME model result indicates that the mean ΔHbO2 from the bilateral DLPFC regions showed a significant positive correlation to the overall FDG-PET controlling for age and group difference in the fNIRS signals. Table 4 summarises the model output. This relationship was consistent (in terms of statistical significance and the size of association) across regions where the fNIRS and FDG-PET signals were collected. Importantly, one-way ANCOVA testing the group diagnostic effect on the global FDG-PET, with age as a covariate, revealed that the FDG-PET signal did not differ across groups, nor was there any group-wise numerical trend. Thus, the small correlation between the fNIRS signal and the FDG-PET suggests a potential correspondence between the two that is independent of the diagnostic information contained within the fNIRS signal.
fNIRS and amyloid PET
When the identical model was fitted to the amyloid PET data, no significant relationship between the fNIRS signal and amyloid deposition was found. The null result was replicated when the simple Pearson correlation between the amyloid PET and fNIRS signals was calculated. Thus, the results suggest that the functional signal acquired from the fNIRS does not capture the amount of amyloid deposition with or without controlling for the subject’s age and group differences in fNIRS activation levels.
Next, a MANCOVA and subsequent ANOVAs were conducted to investigate the group diagnostic effect on the signals from the two fNIRS ROIs and the amyloid PET signal in the temporal lobe, where the group difference was the most obvious. The MANCOVA across amyloid PET from the temporal lobe and the bilateral DLPFC fNIRS ROIs revealed that the group effect was significant in the MANCOVA [F(6, 168) = 6.63, p < .001], and the subsequent ANOVAs were significant [amyloid PET: F(3, 85) = 10.56, p < .001; DLPFC fNIRS: F(3, 85) = 3.59, p < .017]. Furthermore, the MANCOVA on the same aPET ROI and the bilateral OFC fNIRS revealed a significant group effect [F(6, 174) = 6.64, p < .001], leading to a significant effect on amyloid PET [F(3, 88) = 11.46, p < .001] and a marginally significant effect on fNIRS [F(3, 88) = 2.63, p = .055], Overall, Fig 6 shows that the amyloid PET signal tends to better differentiate the AD group from others, and the fNIRS signal tends to differentiate the SMI group from the other groups.
PET: Positron emission tomography, fNIRS: Functional near-infrared spectroscopy, ROI: Region of interest.
fNIRS and anatomical measures
When the LME model was fitted to the anatomical measures, no significant correlation between the fNIRS ROIs and gray volume or cortical thickness was found above and beyond what the group difference could explain. Thus, we moved on to the MANCOVA to see whether the group diagnostic effect differed across anatomical changes and fNIRS, where several interesting significant group effects were found for hippocampal gray volume and DLPFC fNIRS ROIs. Specifically, the MANCOVA on the right hippocampal gray volume and right DLPFC fNIRS ROI were significant [F(6, 180) = 4.27, p < .001], leading to a significant effect on the hippocampal gray volume [F(3, 91) = 5.01, p < .01] and a marginally significant effect on fNIRS [F(3, 91) = 3.48, p = .019] (Fig 7). The comparison between the group pairs in each ANOVA revealed a mirrored pattern between the hippocampal volume and hemodynamic response in the DLPFC. That is, similar to what was shown in the amyloid PET results, the hippocampal volume better differentiates the AD subjects from other groups whereas the fNIRS did so for the SMI group compared to the others. This pattern was consistent across hippocampal and DLPFC regions, but the pairwise group difference varied slightly, suggesting that groups underwent differential functional and anatomical changes.
MANOVA: Multivariate analysis of covariance, DLPFC: Dorsolateral prefrontal cortex, fNIRS: Functional near-infrared spectroscopy.
fNIRS and frontal cortical thickness.
As a supplementary check-up, we also administered the same LME and the MANCOVA test on the thickness data of the same frontal regions that the fNIRS was mainly targeting. No significant correlations were found from the LME model, and MANOVA results show that the hemodynamic difference across groups found from the bilateral DLPFC was not based on the anatomical changes in the same regions, demonstrating the divergence between the functional and anatomical measures as a function of the group diagnosis (Fig 8).
LME: Linear mixed effect (LME), MANOVA: Multivariate analysis of covariance.
Discussion
The key finding of our study is that first, the inter-group differences in the hemodynamic response during SVFT showed decreased cortical activation in the AD and MCI groups, compared to the SMI group which is consistent with the results of previous studies [21,31,32]. Thus, the validity of fNIRS in evaluating the hemodynamic responses during a task was demonstrated in the AD and MCI groups in this study. Also, the hemodynamic response of ΔHbO2 from the bilateral DLPFC regions showed a significant positive correlation to the overall FDG-PET. Thus, overall, this result suggests that the metabolic activity accompanying cerebral blood perfusion measured via FDG-PET is weakly but consistently related to the functional hemodynamics measured via fNIRS. Otherwise, no significant correlation between the fNIRS ROIs and gray volume or cortical thickness fNIRS was identified as hemodynamic responses of the amyloid-PET data and anatomical indicators on brain MRI. Based on the results of this study, it can be predicted that fNIRS will provide useful data for early identification of the metabolic changes that distinguish between normal participants and MCI or AD patients. However, regarding the progression of AD, it is considered at a stage where no conclusions can be made about the usefulness of amyloid imaging with PET and measuring cortical thickness, which has been reported as suitable biomarkers, and functional hemodynamic data through fNIRS [33].
A few previous studies have reported that the correlation between neural activity and blood oxygenation level-dependent responses measured with fNIRS and functional MRI (fMRI) is positive [34–36]. Given that fNIRS only assesses the cortex, it is necessary to determine whether the hemodynamic changes it detects and the metabolic changes and anatomical measures have a correlation. In addition, amyloid PET detects amyloidosis in vivo, and it has a high negative predictive value for AD [37]. FDG-PET allows the detection of topographical information about neurodegeneration by analyzing patterns of hypometabolism and is useful for disease staging and prediction of disease progression [38]. It is believed that confirming its relationship with high task-related hemodynamic activation detected by fNIRS is important in these patients.
A previous study involving healthy participants that investigated the relationship between brain metabolism assessed using FDG-PET and hemodynamic response (which is related to brain activity), assessed using fMRI, showed that the correlations between the regional cerebral metabolic rate of glucose and fMRI metrics were significant [39]. Thus, it is believed that it is important to examine the correlation between hemodynamic response in a particular brain region and the metabolic status of the brain. In our study, the metabolic activity measured via FDG-PET was weak but consistently related to the functional hemodynamics measured via fNIRS. Otherwise, the group diagnostic effect on the global FDG-PET did not differ across groups, nor was there any group-wise numerical trend. Thus, the small correlation between the fNIRS signal and FDG-PET suggests a potential correspondence between the two that is independent of the diagnostic information contained within the fNIRS signal. In addition, the overall amyloid PET signal tends to better differentiate the AD group from other groups, and the fNIRS signal tends to differentiate the SMI group from other groups; this has clinical significance because the potential of aMCI advancing to AD could be determined indirectly based on PET findings in the brain regions.
Gray matter atrophy is a cardinal sign of neurodegeneration [40]. In a previous study, there was a correlation between hippocampal volume and the severity of cognitive impairment, suggesting that the measurement of the hippocampal volume may predict disease progression [41]. In this study, the comparison between the group pairs revealed a mirrored pattern between the hippocampal volume and hemodynamic response in the DLPFC. It is similar to the amyloid PET results, which showed that the hippocampal volume better differentiated the AD group from other groups, whereas the fNIRS differentiated the SMI group from other groups. This pattern was consistent across the hippocampal and DLPFC regions, with only a slight variation in the pairwise group difference, suggesting that the groups underwent differential functional and anatomical changes. The anatomical changes occur only after the participants are diagnosed with AD; however, for the same participants, the functional changes observed via fNIRS occur way earlier as they reveal the cognitive impairment (MCI ∼ AD), implying that fNIRS can be a useful tool for detecting the early risk of cognitive impairment, leading to the preventive medication prescription mentioned in the introduction section.
A limitation of our study is that fNIRS and PET were not performed simultaneously. However, it is thought that the interval between the tests was not long enough for clinical severity to change. Another limitation was that we did not vary the task during fNIRS. SVFT is a tool that is easy to use. It effectively assesses frontotemporal and frontal cortex activation [42], and it is one of the representative tasks used during fNIRS. Thus, we believe it was an appropriate choice for this study.
Conclusions
The hemodynamic response detected by fNIRS showed a significant correlation with the metabolic changes observed on PET. Therefore, fNIRS may be useful as a screening tool to predict the hemodynamic and metabolic statuses of the brain in patients with MCI and AD.
Acknowledgments
This study was supported by the Biomedical Research Institute Grant (20220035001), Pusan National University Hospital.
References
- 1. As Association. 2019 Alzheimer’s disease facts and figures. Alzheimers Demen. 2019;15(3): 321–387.
- 2. Roberts R, Knopman DS. Classification and epidemiology of MCI. Clin Geriatr Med. 2013;29(4): 753–772. pmid:24094295
- 3. Petersen RC, Lopez O, Armstrong MJ, et al. Practice guideline update summary: Mild cognitive impairment: Report of the Guideline Development, Dissemination, and Implementation Subcommittee of the American Academy of Neurology. Neurology. 2018;90(3): 126–135. pmid:29282327
- 4. Nelson AR, Sweeney MD, Sagare AP, et al. Neurovascular dysfunction and neurodegeneration in dementia and Alzheimer’s disease. Biochim Biophys Acta Mol Basis Dis. 2016;1862(5): 887–900. pmid:26705676
- 5. de la Torre JC. Cerebral hemodynamics and vascular risk factors: Setting the stage for Alzheimer’s disease. J Alzheimers Dis. 2012;32(3): 553–567. pmid:22842871
- 6. Boles Ponto LL, Magnotta VA, Moser DJ, et al. Global cerebral blood flow in relation to cognitive performance and reserve in subjects with mild memory deficits. Mol Imaging Biol. 2006;8(6): 363–372. pmid:17048070
- 7. Henderson TA. The diagnosis and evaluation of dementia and mild cognitive impairment with emphasis on SPECT perfusion neuroimaging. CNS Spectr. 2012;17(4): 176–206. pmid:22929226
- 8. Marcus C, Mena E, Subramaniam RM. Brain PET in the diagnosis of Alzheimer’s disease. Clin Nucl Med. 2014;39(10): e413. pmid:25199063
- 9. Herrmann MJ, Langer JB, Jacob C, et al. Reduced prefrontal oxygenation in Alzheimer disease during verbal fluency tasks. Am J Geriatr Psychiatry. 2008;16(2): 125–135. pmid:17998307
- 10. Vermeij A, Kessels RP, Heskamp L, et al. Prefrontal activation may predict working-memory training gain in normal aging and mild cognitive impairment. Brain Imaging and Behav. 2017;11(1): 141–154. pmid:26843001
- 11. Metzger FG, Schopp B, Haeussinger FB, et al. Brain activation in frontotemporal and Alzheimer’s dementia: A functional near-infrared spectroscopy study. Alzheimer’s Res. Ther. 2016;8(1): 1–12. pmid:27931245
- 12. Richter MM, Herrmann MJ, Ehlis A-C, et al. Brain activation in elderly people with and without dementia: Influences of gender and medication. World J Biol Psychiatry. 2007;8(1): 23–29. pmid:17366346
- 13. Chiarelli A M, Perpetuini D, Croce P, et al. Evidence of neurovascular un-coupling in mild Alzheimer’s disease through multimodal EEG-fNIRS and multivariate analysis of resting-state data. Biomedicines. 2021; 9(4): 337. pmid:33810484
- 14. Li X, Zhu Z, Zhao W, Sun Y, et al. Decreased resting-state brain signal complexity in patients with mild cognitive impairment and Alzheimer’s disease: a multi-scale entropy analysis. Biomed Opt Express. 2018; 9(4): 1916–1929. pmid:29675329
- 15. Kadir A, Almkvist O, Forsberg A, et al. Dynamic changes in PET amyloid and FDG imaging at different stages of Alzheimer’s disease. Neurobiol Aging. 2012;33(1): 198. e1-14. pmid:20688420
- 16. McKhann G. Report of the NINCDS-ADRDA work group under the auspices of department of health and human service task force on Alzheimer’s disease. Neurology. 1984;34: 939–944.
- 17. Petersen RC. Mild cognitive impairment as a diagnostic entity. J Intern Med. 2004;256(3): 183–194. pmid:15324362
- 18. Ku H-M, Kim J-H, Kwon E-J, et al. A study on the reliability and validity of Seoul-Instrumental Activities of Daily Living (S-IADL). J Korean Neuropsychiatr Assoc. 2004: 189–199.
- 19. Kim J-M, Choi J-K, Choi M, et al. Assessment of cerebral autoregulation using continuous-wave near-infrared spectroscopy during squat-stand maneuvers in subjects with symptoms of orthostatic intolerance. Sci Rep. 2018;8(1): 1–11.
- 20. Shin J, Kwon J, Choi J, et al. Performance enhancement of a brain-computer interface using high-density multi-distance NIRS. Sci Rep. 2017;7(1): 1–10.
- 21. Arai H, Takano M, Miyakawa K, et al. A quantitative near-infrared spectroscopy study: A decrease in cerebral hemoglobin oxygenation in Alzheimer’s disease and mild cognitive impairment. Brain Cogn. 2006;61(2): 189–94. pmid:16466836
- 22. Choi J-K, Kim J-M, Hwang G, et al. Time-divided spread-spectrum code-based 400 fW-detectable multichannel fNIRS IC for portable functional brain imaging. IEEE J Solid-State Circuits. 2016;51(2): 484–495.
- 23. Delpy DT, Cope M, van der Zee P, et al. Estimation of optical pathlength through tissue from direct time of flight measurement. Phys Med Biol. 1988;33(12): 1433. pmid:3237772
- 24. Cui X, Bray S, Bryant D, et al. A quantitative comparison of NIRS and fMRI across multiple cognitive tasks. NeuroImage. 2011;54(4): 2808–2821. pmid:21047559
- 25. Cui X, Bray S, Reiss AL.Functional near infrared spectroscopy (NIRS) signal improvement based on negative correlation between oxygenated and deoxygenated hemoglobin dynamics. Neuroimage. 2010; 49(4): 3039–3046. pmid:19945536
- 26. Dale AM, Fischl B, Sereno MI. Cortical surface-based analysis: I. Segmentation and surface reconstruction. Neuroimage. 1999;9(2): 179–194.
- 27. Desikan RS, Ségonne F, Fischl B, et al. An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. Neuroimage. 2006;31(3): 968–980. [published online first: 2006/03/15]. pmid:16530430
- 28. Shepp LA, Vardi Y. Maximum likelihood reconstruction for emission tomography. IEEE Trans Med Imaging. 1982;1(2): 113–122. pmid:18238264
- 29. Tzourio-Mazoyer N, Landeau B, Papathanassiou D, et al. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage. 2002;15(1): 273–289. pmid:11771995
- 30. Seibyl J, Catafau AM, Barthel H, et al. Impact of training method on the robustness of the visual assessment of 18F-florbetaben PET scans: results from a phase-3 study. J Nucl Med. 2016;57(6): 900–906. pmid:26823561
- 31. Li Y, Yang H, Lei B, et al. Novel effective connectivity inference using ultra-group constrained orthogonal forward regression and elastic multilayer perceptron classifier for MCI identification. IEEE Trans Med Imaging. 2018;38(5): 1227–1239. pmid:30475714
- 32. Yap KH, Ung WC, Ebenezer EG, et al. Visualizing hyperactivation in neurodegeneration based on prefrontal oxygenation: a comparative study of mild Alzheimer’s disease, mild cognitive impairment, and healthy controls. Front Aging Neurosci. 2017;9: 287. pmid:28919856
- 33. Ortner M, Drost R, Heddderich D, et al. Amyloid PET, FDG-PET or MRI?-the power of different imaging biomarkers to detect progression of early Alzheimer’s disease. BMC neurol. 2019;19:1–6.
- 34. Steinbrink J, Villringer A, Kempf F, et al. Illuminating the BOLD signal: Combined fMRI–fNIRS studies. Magn Reson Imaging. 2006;24(4): 495–505. pmid:16677956
- 35. Scarapicchia V, Brown C, Mayo C, et al. Functional magnetic resonance imaging and functional near-infrared spectroscopy: Insights from combined recording studies. Front Hum Neurosci. 2017;11: 419. pmid:28867998
- 36. Cui X, Bray S, Bryant DM, et al. A quantitative comparison of NIRS and fMRI across multiple cognitive tasks. Neuroimage. 2011;54(4): 2808–2821. pmid:21047559
- 37. Grothe MJ, Barthel H, Sepulcre J, et al. In vivo staging of regional amyloid deposition. Neurology. 2017;89(20): 2031–2038. pmid:29046362
- 38. Laforce R Jr, Soucy J-P, Sellami L, et al. Molecular imaging in dementia: Past, present, and future. Alzheimers Demen. 2018;14(11): 1522–1552. pmid:30028955
- 39. Shan Y, Wang Z, Song S, et al. Integrated positron emission tomography/magnetic resonance imaging for resting-state functional and metabolic imaging in human brain: What is correlated and what is impacted. Front Neurosci. 2022;16. pmid:35310105
- 40. Krajcovicova L, Klobusiakova P, Rektorova I. Gray matter changes in Parkinson’s and Alzheimer’s disease and relation to cognition. Curr Neurol Neurosci Rep. 2019;19(11): 1–9. pmid:31720859
- 41. Vijayakumar A, Vijayakumar A. Comparison of hippocampal volume in dementia subtypes. Int Sch Res Notices. 2013;2013. pmid:24959551
- 42. Tupak SV, Badewien M, Dresler T, et al. Differential prefrontal and frontotemporal oxygenation patterns during phonemic and semantic verbal fluency. Neuropsychologia. 2012;50(7): 1565–1569. pmid:22426205