Discriminant analysis of functional
optical topography for schizophrenia
diagnosis
Ching-Cheng Chuang
Kazuyuki Nakagome
Shenghong Pu
Tsuo-Hung Lan
Chia-Yen Lee
Chia-Wei Sun
Discriminant analysis of functional optical topography for
schizophrenia diagnosis
Ching-Cheng Chuang,a,b*Kazuyuki Nakagome,c*Shenghong Pu,dTsuo-Hung Lan,e,f Chia-Yen Lee,gand Chia-Wei Suna
aNational Chiao Tung University, Department of Photonics, Institute of Electro-Optical Engineering and Biomedical Optical Imaging Lab, Hsinchu
30010, Taiwan
bNational Yang-Ming University, Biophotonics and Molecular Imaging Research Center, Taipei 11221, Taiwan cNational Center of Neurology and Psychiatry, Tokyo 187-8551, Japan
dTottori University Faculty of Medicine, Division of Neuropsychiatry, Department of Brain and Neuroscience, Tottori 101, Japan eNational Yang-Ming University, School of Medicine, Department of Medicine, Taipei 11221, Taiwan
fTaichung Veterans General Hospital, Department of Psychiatry, Taichung 407, Taiwan gNational United University, Department of Electrical Engineering, Miaoli 36003, Taiwan
Abstract. Abnormal prefrontal function plays a central role in the cognition deficits of schizophrenic patients; how-ever, the character of the relationship between discriminant analysis and prefrontal activation remains undeter-mined. Recently, evidence of low prefrontal cortex (PFC) activation in individuals with schizophrenia has also been found during verbal fluency tests (VFT) and other cognitive tests with several neuroimaging methods. The purpose of this study is to assess the hemodynamic changes of the PFC and discriminant analysis between schizo-phrenia patients and healthy controls during VFT task by utilizing functional optical topography. A total of 99 subjects including 53 schizophrenic patients and 46 age- and gender-matched healthy controls were studied. The results showed that the healthy group had larger activation in the right and left PFC than in the middle PFC. Besides, the schizophrenic group showed weaker task performance and lower activation in the whole PFC than the healthy group. The result of the discriminant analysis showed a significant difference with P value <0.001 in six channels (CH 23, 29, 31, 40, 42, 52) between the schizophrenic and healthy groups. Finally, 68.69% and 71.72% of subjects are correctly classified as being schizophrenic or healthy with all 52 chan-nels and six significantly different chanchan-nels, respectively. Our findings suggest that the left PFC can be a feature region for discriminant analysis of schizophrenic diagnosis.© The Authors. Published by SPIE under a Creative Commons Attribution 3.0 Unported License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.[DOI:
10.1117/1.JBO.19.1.011006]
Keywords: schizophrenia; prefrontal cortex dysfunction; functional optical topography; discriminant analysis; near-infrared spectroscopy. Paper 130190SSR received Apr. 2, 2013; revised manuscript received May 28, 2013; accepted for publication May 31, 2013; published online Aug. 29, 2013.
1 Introduction
Schizophrenia is an irreversible neurodevelopmental disorder with severe individual, family, and societal burdens.1Several studies have indicated that schizophrenic patients showed impaired performance in various aspects of social cognition, including theory of mind, emotion processing, and agency judgments.2–5The neuroimaging techniques such as functional magnetic resonance imaging (fMRI) and positron emission tomography (PET) have been widely used to image brain func-tions for many neurodegenerative diseases.6–12These techniques have greatly increased our knowledge about the neuropatholog-ical process of schizophrenia based on cognitive and emotional processes. Although these methods offer noninvasive imaging with excellent spatial resolution, the limitation of huge size of instruments and confinement of the participants to restricted positions inside the magnet cannot be applied for patient-oriented diagnosis. Also, PET requires the injection of radio-active materials. These drawbacks make the imaging modalities
difficult or impossible to apply in clinical diagnosis for neo-nates, children, old people, and claustrophobic patients.
Functional optical topography (fOT), imaging based on func-tional near-infrared spectroscopy measurement, is an another noninvasive neuroimaging technique that provides continuous recording of brain hemodynamic activity with detection of oxy- and deoxy-hemoglobin concentration changes. There are several benefits such as lower cost, nonionizing radiation imag-ing, real-time measurement, long time monitorimag-ing, and easy operation.13 Although spatial resolution is limited, fOT offers more comprehensive information of brain activity than blood-oxygenation-level-dependent signal of fMRI.14 Furthermore, the optical method can provide completely patient-oriented measurement. The fOT is also applicable for psychological test, because its temporal resolution is high enough to detect the changes of short duration such as the brain activation during cognitive task. In previous studies, fOT method has been shown to be sensitive enough for monitoring of physiological blood oxygenation changes during cognitive activation in schizophre-nia.15–17Recently, fOT is being increasingly applied for inves-tigation of prefrontal cortex (PFC) dysfunction in schizophrenia, because the social cognition of schizophrenic patients was asso-ciated with PFC activation. PFC dysfunction in individuals with schizophrenia has been detected and evidenced during verbal fluency test (VFT) and other cognitive tests by utilizing fOT.18–22 However, the previous studies only show the group
*Equally contributed to this paper.
Address all correspondence to: Chia-Wei Sun, National Chiao Tung University, Department of Photonics, Institute of Electro-Optical Engineering and Biomedical Optical Imaging Lab, Hsinchu 30010, Taiwan. Tel: +886 3 5712121x56383; Fax: +886 3 5735601; E-mail:[email protected]
averaged data and correlational analysis between fOT and clini-cal symptoms. There are few studies reporting significantly different regions of PFC and discriminant analysis between schizophrenic and healthy subjects based on fOT measurement. The significantly different regions of PFC could be an indicator to evaluate the PFC function of schizophrenia in patient-oriented diagnosis. Thus, the goal of this study is to characterize/analyze fOT image between schizophrenia patients and healthy controls during VFT.
In this paper, we perform statistical analyses of PFC fOT image with VFT. The imaging results between schizophrenia and healthy controls are significantly different. Furthermore, the left PFC shows significant difference between schizophrenic and healthy subjects under the 0.1% significance level. In discrimi-nant analysis, the 68.69% and 71.72% are correctly classified as being schizophrenic or healthy subjects with all 52 channels and six significantly different channels over the left PFC, respec-tively. Our results imply that the left PFC fOT imaging could be a feature region for schizophrenia-aided diagnosis.
2 Material and Methods
2.1 Participants
The study includes 53 schizophrenic patients [female∕male: 27∕26; mean age: 34.4; standard deviation (SD): 11] and 46
age- and sex-matched healthy controls (female∕male: 25∕21; mean age 35.1; SD 11.2), who were recruited from the Tottori University Hospital in Yonago, Japan. The diagnosis for schizophrenia was based on the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV.23Healthy controls had no history of psychiatric or neurologic disorders. All subjects were native Japanese speak-ers. There was no difference between schizophrenic and control groups with respect to age, gender, and body mass index (BMI). However, the education, estimated premorbid IQ, and number of words generated show significant differences. Schizophrenic symptoms were evaluated with the Positive and Negative Syndrome Scale (PANSS) and Global Assessment of Function-ing.24–27Table1shows the clinical characteristics of the schizo-phrenic and healthy groups. All participants gave written informed consent. The study was in accordance with the latest version of the Declaration of Helsinki, and approved by the Ethics Committee of Tottori University Faculty of Medicine.
2.2 Functional Optical Topography
The fOT measurements were conducted with a 52-channel ETG-4000 Optical Topography System (Hitachi Medical Co., Tokyo, Japan). The near-infrared sources at dual-wavelength 695 and 830 nm were used, and the temporal resolution of detection
Table 1 Clinical characteristics of the schizophrenic patients and healthy controls.
Clinical variables Schizophrenia group (n ¼ 53) Healthy group(n ¼ 46) Group difference P value Mean SD Mean SD Age (years) 34.4 11 35.1 11.2 0.756 Gender (female/male) 27∕26 — 25∕21 — 0.735a Handedness (right/left) 48∕5 — 46∕0 — — Education (years) 13.2 2.28 15.24 2.35 <0.001 Estimated premorbid IQ 97.69 12.36 104 10.25 <0.01 Age at onset (years) 23.59 6.97 — — — Number of words generated 12.3 4.11 14.61 5 <0.001 Duration of illness (years) 10.24 9.03 — — — Positive and Negative Syndrome Scale (PANSS) (positive) 16.49 7.4 — — — PANSS (negative) 17.46 5.43 — — — PANSS (general psychopathology) 33.74 7.31 — — — Global Assessment of Functioning 58.67 12.71 — — — Chlorpromazine equivalent dose (mg∕day) 653.65 387.17 — — — Diazepam equivalent dose (mg∕day) 5.68 6.68 — — — Biperiden equivalent dose (mg∕day) 0.47 0.99 — — — Body mass index (BMI) 23.25 4.88 22.67 4.34 0.55
aChi-square test was used for testing group difference. Otherwise,t-test was used.
is 0.1 s. According to the modified Beer–Lambert law,28,29the concentration changes in oxygenated hemoglobin (oxy-Hb) and deoxygenated hemoglobin (deoxy-Hb) could be obtained. Previous studies indicate that [oxy-Hb] is the most sensitive indicator of changes in regional cerebral blood flow in fOT measurement.30,31In fOT system, 17 emitters and 16 detectors are plugged into a probing holder and arranged as 3× 11 array for 52-channel detection with 30-mm source–detector separation. In order to cover the whole PFC, the array was positioned according to the international 10-20 system.32 Specifically, the detector 13 was placed in Fz; the midline of channels nos. 46 and 49 were placed in the Fp2 and Fp1, and the source 23 and 28 were fitted around T4 and T3, res-pectively (as shown in Fig.1).
2.3 Data Acquisition: VFT Tasks
The subjects were seated in a comfortable chair in a silent room and kept eyes open during the fOT monitoring. The VFT consists of a 60-s long VFT (letter version), a 30-s long pretask baseline, and a 70-s long posttask baseline. The three syllables (I: /to/, /se/ or /o/; II: /na/, /i/ or /ta/; III: /a/, /ki/ or /ha/) were announced in the order which was counterbal-anced across participants. The fOT data were obtained by sub-tracting the baseline [oxy-Hb] while the subjects were simply repeating“a,” “i,” “u,” “e,” and “o” loudly. The VFT requires subject to say the words beginning with a specific syllable as many as possible. The schematic diagram of the task is shown in Fig.2.
Fig. 1 The functional optical topography (fOT) channels localization. The localizations of all 52 channels were positioned according to the international 10-20 system. Red and green circles indicate near-infrared light emitter and detector positions, respectively. By using the international 10-20 system, the detector 13 was positioned on theFzmarker point, while the bottom row of channels was placed on a line betweenT3andT4.
Fig. 2 The sequence consists of 60-s long verbal fluency test (VFT). Pretask baseline consists of 30-s long VFT and posttask baseline consists 70-s long VFT. The three syllables (I: /to/, /se/ or /o/; II: /na/, /i/ or /ta/; III: /a/, /ki/ or /ha/) are announced in the order which was counterbalanced across participants.
2.4 Data Analyses
Figure 3 shows the flowchart of fOT data analyses. In each subject, the mean levels in [oxy-Hb] of all 52 channels are calculated from both baseline and activation task of VFT. In order to find the significantly different regions of PFC, the fOT data are analyzed based on statistical tests. We adopt a chi-square goodness-of-fit test for statistical hypothesis test-ing. The chi-square test-statistic can be expressed as33,34
χ2¼Xn i¼1
ðOi− EiÞ2
Ei ;
(1) whereOiis an observed count andEiis an expected count;χ2 is Pearson’s cumulative test statistic; and n is the number of cells in the table. The statistic has an approximate chi-square distribution when the counts are sufficiently large. In our results, CH 5, 10, 11, 15, 23, 25, 26, 34, 42, 44, 45, 46, and 49 in healthy group and CH 5, 6, 10, 11, 12, 14, 17, 21, 24, 25, 39, 41, 45, 46, and 48 in schizophrenic group are not normally distributed. Due to physiological noise, the majority of the channels were found to be nonnormally dis-tributed, and therefore, we used the nonparametric test on all the data. The two-sample Kolmogorov–Smirnov test (K-S test) is one of the most powerful nonparametric methods with highly sensitive comparison in both location and shape of the empirical cumulative distribution functions of the two samples.35,36
The two-sample K-S test is expressed as Dn;n0 ¼ SUP
x jGnðxÞ − Fn0ðxÞj; (2)
where SUP
x is the supremum of the set of distances; andGnðxÞ
and Fn0ðxÞ are the empirical distribution functions of the first
and the second variables, respectively. The null hypothesis is rejected at significance levelα if
ffiffiffiffiffiffiffiffiffiffiffiffiffi nn0
n þ n0
r
Dn;n0> Kα: (3)
Equation (3) impliesGnðxÞ and Fn0ðxÞ are significantly
dif-ferent. The test statisticKαis the maximum difference between the curves. After K-S testing, the significantly different regions of PFC are defined between healthy and schizophrenic groups. Then, the discriminant analysis is applied by leave-one-out cross-validation with k-means clustering.37 In statistics and data mining, the k-means method is one of the most widely used clustering algorithms with nearest mean, drawing its popularity from its speed in practice.37,38Leave-one-out cross-validation involves single observation from the original sample as the validation data, and the remaining observations as the training data. The validation step is looped until all observations are used once as the validation data. In this study, one subject is selected from all 99 subjects in sequence. Thus, the other 98 subjects are treated as training data withk-means clustering. The 99 times clustering of test data is then used for accuracy estimation of discriminant analysis.
3 Results
Figure4shows the [oxy-Hb] mapping of group-level statistics on averaging of healthy controls and schizophrenic patients. The [oxy-Hb] concentration of schizophrenic patients is significantly smaller than healthy controls during VFT.
In case of healthy controls, the [oxy-Hb] response is stronger on both sides than in middle PFC. For quantitative analysis, the mean of [oxy-Hb] on right, middle, and left PFC during VFT are demonstrated in Fig.5. There exists the significant differ-ence between healthy and schizophrenic groups (pairedt-test, *P < 0.005, **P < 0.001).
Figure6shows the [oxy-Hb] mapping via 52 channels com-parisons based on two-sample K-S test. The CH 23, 29, 31, 40, 42, and 52 are significantly different withP < 0.001. We can observe that the channel on the left regions of PFC is sig-nificantly different. This result implies that the left PFC may provide significant feature region to distinguish schizophrenia from healthy subjects in clinical diagnosis. The discriminant analysis reveals the significant difference with K-S test by utilizing leave-one-out cross-validation withk-means clustering. The four possible outcomes from a binary classifier of 52 and six
Fig. 3 The flowchart of data analyses with fOT. The two-sample K-S test of nonparametric methods was used to find the significantly different channels because some channels in healthy group and schizophrenic group are not normally distributed by using the chi-square goodness-of-fit test. Finally, the leave-one-out cross-validation withk-means cluster-ing was used for discriminant analysis from healthy to schizophrenia group.
Fig. 4 Group-level [oxy-Hb] mapping with statistics on average during the VFT task. The prefrontal cortex (PFC) was divided into three regions (right, middle, and left PFC) to analyze. The [oxy-Hb] concentration of healthy group is significantly higher in right and left PFC. The PFC acti-vation of schizophrenic patients is significantly smaller than healthy group (healthy group:N ¼ 46; schizophrenic group: N ¼ 53). Chuang et al.: Discriminant analysis of functional optical topography for schizophrenia diagnosis
dimensions are shown in Table 2. The accuracy of correctly classified with 52 dimensions was 68.69% with a sensitivity of 85% and a specificity of 50% (i.e., true positive¼ 45 of the 53 schizophrenic patients and true negative¼ 23 of the 46 healthy controls). In the case of six significant difference dimensions, the accuracy of correctly classified was 71.72% with a sensitivity of 77% and a specificity of 65% (i.e., true positive ¼ 41 of the 53 schizophrenic patients and true negative ¼ 30 of the 46 healthy controls).
4 Discussion and Conclusions
The [oxy-Hb] responses on both sides of PFC are stronger than middle PFC from healthy controls, which implies deeper light penetration in the interhemispheric fissure that causes the fewer backscattered photons to be received. Furthermore, the [oxy-Hb] mapping of schizophrenic case is significantly weaker than normal on PFC and no significant difference is found among right, left, and middle forehead areas. It indicates the task-dependent profile of functional abnormalities can be observed from the brain metabolism on schizophrenic PFC. Also, the fOT measurement may offer a good potential to detect the structural change of brain. To our knowledge, we are the first to report the significant difference regions of PFC by utilizing discriminant analysis between schizophrenic and healthy sub-jects based on fOT monitoring. Although Azechi et al. used prefrontal activation with frontal lobe tasks for discriminant analyzing between schizophrenic and healthy subjects,39they did not provide the significant difference regions of PFC for dis-criminating schizophrenia. In this study, we pointed out the sig-nificant difference regions of PFC between schizophrenia and healthy controls. Figure2 shows the group-level of [oxy-Hb] in right, left, and middle PFC that is significantly different between schizophrenia and healthy controls. According to two-sample K-S test, the channels of significant difference are analyzed on left PFC. Abnormal structural and/or functional hemisphere asymmetries in schizophrenia disorders have been assessed based on fOT.18,40In an fMRI study, reduced activation of the left frontal regions during word production was found in schizophrenia patients compared with normal.41The other study shows that schizophrenia produces a failure of attentional modu-lation that leads to a breakdown in the selective enhancement or inhibition of semantic/lexical representations, whose biological substrata are widely distributed across left (dominant) temporal and frontal lobes.42These findings can be interpreted as a sign of reduced specific lateralized PFC reactivity, possibly based on a left hemisphere functional deficit between schizophrenia and normal. Our result of the significant difference regions on left PFC is in agreement with previous studies. Therefore, the left PFC shows a great potential of feature region for discrimi-nant analysis of schizophrenia. Table2demonstrates the accu-racy of classification by applying six significant channels, which is better than 52 channels. Therefore, the significant channels in fOT can be used for schizophrenia diagnosis with discriminant analysis. However, the limitation of this study must be taken into account. Discriminant analysis with fOT can reveal significant differences between the probability distributions of signals in healthy and diseased groups, but cannot provide an individual diagnosis. Nevertheless, optical imaging with fOT can still be used as an auxiliary tool in schizophrenia diagnosis.
Fig. 6 The result of two-sample K-S test. The CH 23, 29, 31, 40, 42, and 52 are significantly different with P < 0.001 between healthy and schizophrenia groups. This result further indicates that the activation of left PFC during VFT is significantly different between healthy and schizophrenia group.
Table 2 The four possible outcomes from a binary classifier of 52 and 6 dimensions.
Accuracy of classification
k-means classification
(all 52 dimensions) dimensions by Kolmogorov–Smirnov test)k-means classification (6 significant Schizophrenic Healthy Schizophrenic Healthy Original truth Schizophrenic (N ¼ 53) 45 8 41 12
Healthy (N ¼ 46) 23 23 16 30 Note: 52 dimensions: accuracy¼ ð23 þ 45Þ∕ð46 þ 53Þ 100 ¼ 68.69%; 6 dimensions: accuracy ¼ ð30 þ 41Þ∕ð46 þ 53Þ 100 ¼ 71.72%. Fig. 5 The results of group-level statistics on averaging. The activation
of right and left PFC between healthy and schizophrenia groups are significantly different withP < 0.001 and with P < 0.005 on middle PFC (pairedt-test, *P < 0.005, **P < 0.001).
In conclusion, the [oxy-Hb] PFC mapping of schizophrenia cases is weaker than healthy ones during VFT. The low [oxy-Hb] change of cognitive activation is concomitant with PFC dysfunction of schizophrenia. The significantly different channels of left PFC provide high-accuracy discrimination for schizophrenia. fOT offers both structural and functional information simultaneously with patient-oriented diagnosis. Furthermore, it could be used for the neural circuitry abnormal-ity detection and estimation of pharmacological treatment and/ or cognitive rehabilitation.
Acknowledgments
This work was supported in part by the Taiwan National Science Council under Grant NSC 101-2628-E-009-026-MY3, NSC 102-2622-E-009-007-CC3, NSC 101-3011-P-010-003, NSC 101-2627-E-010-001 and NSC 102-2627-E-010-001.
References
1. T. Walsh et al.,“Rare structural variants disrupt multiple genes in neuro-developmental pathways in schizophrenia,”Science320(25), 539–543 (2008).
2. R. I. M. Dunbar,“The social brain hypothesis,”Evol. Anthropol.6(5), 178–190 (1998).
3. G. Bush, P. Luu, and M. I. Posner,“Cognitive and emotional influences in anterior cingulate cortex,”Trends Cogn. Sci.4(6), 215–222 (2000). 4. T. Jellema et al.,“Neural representation for the perception of the
inten-tionality of actions,”Brain Cognition44(2), 280–302 (2000). 5. J. K. Burns,“An evolutionary theory of schizophrenia: cortical
connec-tivity, metarepresentation, and the social brain,” Behav. Brain Sci. 27(6), 831–885 (2004).
6. D. S. Manoach et al., “Schizophrenic subjects activate dorsolateral prefrontal cortex during a working memory task, as measured by fMRI,”Biol. Psychiat.45(9), 1128–1137 (1999).
7. R. E. Gur et al.,“An fMRI study of facial emotion processing in patients with schizophrenia,”Am. J. Psychiat.159(12), 1992–1999 (2002).
8. Y. Zhou et al., “Functional disintegration in paranoid schizophrenia using resting-state fMRI,”Schizophr. Res.97(1–3), 194–205 (2007). 9. Y. Zhoua et al.,“Functional dysconnectivity of the dorsolateral
pre-frontal cortex in first-episode schizophrenia using resting-state fMRI,”
Neurosci. Lett.417(3), 297–302 (2007).
10. D. Mamo et al.,“Differential effects of aripiprazole on D2, 5-HT2, and 5-HT1A receptor occupancy in patients with schizophrenia: a triple tracer PET study,”Am. J. Psychiat.164(9), 1411–1417 (2007). 11. R. Mizrahi et al.,“Adverse subjective experience with antipsychotics
and its relationship to striatal and extrastriatal D2 receptors: a PET study in schizophrenia,”Am. J. Psychiat.164(4), 630–637 (2007). 12. O. D. Howes et al.,“Elevated striatal dopamine function linked to
prodro-mal signs of schizophrenia,”Arch. Gen. Psychiat.66(1), 13–20 (2009). 13. G. Boas,“Noninvasive imaging of the brain,” Opt. Photon. News 15(1),
52–55 (2004).
14. S. C. Bunce et al.,“Functional near-infrared spectroscopy,”IEEE Eng.
Med. Biol.25(4), 54–62 (2006).
15. Y. Fukuji et al.,“Changes of cerebral blood volume in frontal region during mirror drawing task measured by two channel NIRS in schizo-phrenics,”Int. Congr. Ser.1232, 667–671 (2002).
16. B. S. Folley and S. Park,“Verbal creativity and schizotypal personality in relation to prefrontal hemispheric laterality: a behavioral and near-infrared optical imaging study,”Schizophr. Res.80(2), 271–282 (2005). 17. S. Sumitani et al., “Activation of the prefrontal cortex during the wisconsin card sorting test as measured by multichannel near-infrared spectroscopy,”Neuropsychobiology53(2), 70–76 (2006).
18. Y. Kubotaa et al.,“Prefrontal activation during verbal fluency tests in schizophrenia-a near-infrared spectroscopy (NIRS) study,”Schizophr. Res.77(1), 65–73 (2005).
19. A.-C. Ehlis et al., “Cortical activation during two verbal fluency tasks in schizophrenic patients and healthy controls as assessed by
multi-channel near-infrared spectroscopy,” Psychiat Res Neuroimag.
156(1), 1–13 (2007).
20. R. Takizawa et al.,“Reduced frontopolar activation during verbal flu-ency task in schizophrenia: a multi-channel near-infrared spectroscopy study,”Schizophr. Res.99(1–3), 250–262 (2008).
21. H. Hori et al.,“Schizotypal traits in healthy women predict prefrontal activation patterns during a verbal fluency task: a near-infrared spectros-copy study,”Neuropsychobiology57(1–2), 61–69 (2008).
22. K. Ikezawa et al.,“Impaired regional hemodynamic response in schizo-phrenia during multiple prefrontal activation tasks: a two-channel near-infrared spectroscopy study,”Schizophr. Res.108(1–3), 93–103 (2009). 23. M. First et al.,“Structured Clinical Interview for DSM-IV-TR Axis I Disorders, Research Version, Patient Edition (SCID-I/P),” Biometrics Research, NYSPI, New York (2002).
24. S. R. Kay, A. Fiszbein, and L. A. Opler,“The positive and negative syndrome scale (PANSS) for schizophrenia,” Schizophr. Bull.13(2), 261–275 (1987).
25. S. Leuchta et al.,“What does the PANSS mean?,” Schizophr. Res.
79(2–3), 231–238 (2005).
26. S. H. Jones et al.,“A brief mental health outcome scale reliability and validity of the Global Assessment of Functioning (GAF),”Br. J.
Psychiat.166(5), 654–659 (1995).
27. M. Startup, M. C. Jackson, and S. Bendix,“The concurrent validity of the Global Assessment of Functioning (GAF),” Br. J. Clin. 41(4), 417–422 (2002).
28. Y. Yamashita et al.,“Noninvasive near-infrared topography of human brain activity using intensity modulation spectroscopy,” Opt. Eng.
35(4), 1046–1049 (1996).
29. L. Kocsis, P. Herman, and A. Eke,“The modified Beer–Lambert law revisited,”Phys. Med. Biol.51(5), N91–N98 (2006).
30. Y. Hoshi,“Functional near-infrared spectroscopy: current status and future prospects,”J. Biomed. Opt.12(6), 062106 (2007).
31. C.-M. Lu et al., “Use of fNIRS to assess resting state functional connectivity,”J. Neurosci. Methods186(2), 242–249 (2010).
32. M. Okamoto et al.,“Three-dimensional probabilistic anatomical cranio-cerebral correlation via the international 10–20 system oriented for transcranial functional brain mapping,” NeuroImage 21(1), 99–111
(2004).
33. R. L. Plackett,“Karl Pearson and the chi-squared test,”Int. Stat. Rev.
51(1), 59–72 (1983).
34. D. M. Glover, W. J. Jenkins, and S. C. Doney, Lecture 3: Least Squares and Regression Techniques, Goodness of Fit and Tests, Non-linear Least Squares Techniques, Woods Hole Oceanographic Institute, Massachusetts (2008).
35. R. H. C. Lopes, I. Reid, and P. R. Hobson,“The two-dimensional Kolmogorov-Smirnov test,” in Proc. Sci., XI International Workshop on Advanced Computing and Analysis Techniques in Physics Research, Amsterdam, The Netherlands, pp. 1–12 (2007).
36. G. W. Corder and D. I. Foreman, Nonparametric Statistics for Non-Statisticians: A Step-by-Step Approach, Wiley, Hoboken, New Jersey (2009).
37. D. Arthur, B. Manthey, and H. Roeglin, “K-means has polynomial smoothed complexity,” in Proceedings of the 50th Annual IEEE Symposium on Foundations of Computer Science (FOCS), pp. 405– 414 (2009).
38. H. Zha et al.,“Spectral relaxation for K-means clustering,” Adv. Neural Inform. Process. Syst. 14(NIPS’01), 1057–1064 (2001).
39. M. Azechi et al.,“Discriminant analysis in schizophrenia and healthy subjects using prefrontal activation during frontal lobe tasks: a near-infrared spectroscopy,”Schizophr. Res.117(1), 52–60 (2010). 40. M. J. Herrmann et al.,“Cerebral oxygenation changes in the prefrontal
cortex: effects of age and gender,”Neurobiol. Aging27(6), 888–894
(2006).
41. D. A. Yurgelun-Todd et al.,“Functional magnetic resonance imaging of schizophrenic patients and comparison subjects during word produc-tion,” Am. J. Psychiat. 153(2), 200–205 (1996).
42. P. G. Nestor et al.,“Semantic disturbance in schizophrenia and its relationship to the cognitive neuroscience of attention,”Biol. Psychol.
57(1–3), 23–46 (2001).
![Figure 3 shows the flowchart of fOT data analyses. In each subject, the mean levels in [oxy-Hb] of all 52 channels are calculated from both baseline and activation task of VFT](https://thumb-ap.123doks.com/thumbv2/9libinfo/7584234.126690/5.944.96.454.87.375/figure-flowchart-analyses-subject-channels-calculated-baseline-activation.webp)
