White matter abnormalities associated with auditory hallucinations in schizophrenia: A combined study of voxel-based analyses of diffusion tensor imaging and structural magnetic resonance imaging

Available online at www.sciencedirect.com

Psychiatry Research: Neuroimaging 156 (2007) 93 – 104 www.elsevier.com/locate/psychresns

White matter abnormalities associated with auditory hallucinations in schizophrenia: A combined study of voxel-based analyses of diffusion tensor imaging and structural magnetic resonance imaging Jeong-Ho Seoka,b , Hae-Jeong Parkc , Ji-Won Chuna , Seung-Koo Leec , Hyun Sang Choa,d , Jun Soo Kwone , Jae-Jin Kima,c,d,⁎ a

Institute of Behavioral Science in Medicine, Yonsei University College of Medicine, Severance Mental Health Hospital, 696-6 Tanbul-dong Gwangju, Gyeonggi, 464-100 South Korea b Department of Psychiatry, Hallym University College of Medicine, Anyang Gyeonggi, South Korea c Department of Diagnostic Radiology, Yonsei University College of Medicine, Seoul, South Korea d Department of Psychiatry, Yonsei University College of Medicine, Seoul, South Korea e Department of Psychiatry, Seoul National University College of Medicine, Seoul, South Korea Received 15 December 2005; received in revised form 18 January 2007; accepted 3 February 2007

Abstract White matter (WM) abnormalities in schizophrenia may offer important clues to a better understanding of the disconnectivity associated with the disorder. The aim of this study was to elucidate a WM basis of auditory hallucinations in schizophrenia through the simultaneous investigation of WM tract integrity and WM density. Diffusion tensor images (DTIs) and structural T1 magnetic resonance images (MRIs) were taken from 15 hallucinating schizophrenic patients, 15 non-hallucinating schizophrenic patients and 22 normal controls. Voxel-based analyses and post-hoc region of interest analyses were obtained to compare the three groups on fractional anisotropy (FA) derived from DTI as well as WM density derived from structural MRIs. In both the hallucinating and non-hallucinating groups, FA of the WM regions was significantly decreased in the left superior longitudinal fasciculus (SLF), whereas WM density was significantly increased in the left inferior longitudinal fasciculus (ILF). The mean FA value of the left frontal part of the SLF was positively correlated with the severity score of auditory hallucinations in the hallucinating patient group. Our findings show that WM changes were mainly observed in the frontal and temporal areas, suggesting that disconnectivity in the left fronto–temporal area may contribute to the pathophysiology of schizophrenia. In addition, pathologic WM changes in this region may be an important step in the development of auditory hallucinations in schizophrenia. © 2007 Elsevier Ireland Ltd. All rights reserved. Keywords: Schizophrenia; Auditory hallucinations; Left superior longitudinal fasciculus; DTI; Voxel-based morphometry

⁎ Corresponding author. Institute of Behavioral Science in Medicine, Yonsei University College of Medicine, Severance Mental Health Hospital, 696-6 Tanbul-dong Gwangju, Gyeonggi, 464-100 South Korea. Tel.: +82 31 760 9402; fax: +82 31 761 7582. E-mail address: [email protected] (J.-J. Kim). 0925-4927/$ - see front matter © 2007 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.pscychresns.2007.02.002

94

J.-H. Seok et al. / Psychiatry Research: Neuroimaging 156 (2007) 93–104

1. Introduction The pathophysiology of schizophrenia has been extensively explored by many investigators since the advent of advanced, non-invasive, in vivo neuroimaging methods. Using computerized tomography, Johnstone et al. (1976) first reported the dilatation of the lateral ventricles in patients with chronic schizophrenia. Since then, an increasing number of neuroimaging researchers have confirmed structural brain abnormalities in schizophrenia which are disparately located but functionally related (for review, see Shenton et al., 2001). However, we do not know exactly how these regional alterations affect one another. Ever since Kraepelin (1919/1971) noted the importance of a fronto–temporal connection in schizophrenia, a functional disconnection has been an important concept in the pathophysiology of schizophrenia. Andreasen et al. (1998) suggested a dysfunction in the cortical–subcortical– cerebellar circuitry in schizophrenia, including several nodes such as the prefrontal cortex, the striatum, the thalamus, and the cerebellum. White matter (WM) abnormalities in schizophrenia may contain clues important to the understanding of these functional disconnections in schizophrenia. There have been studies on WM abnormalities associated with specific symptoms of schizophrenia using voxel-based analyses of diffusion tensor images (DTI) and structural magnetic resonance images (MRIs). In particular, auditory hallucinations in schizophrenia may be associated with the abnormal activation of brain regions which have been known to be involved in auditory information processing (Lawrie and Abukmeil, 1995; Dierks et al., 1999; Ropohl et al., 2004). Moreover, structural abnormalities in the fronto–temporal region may play an important role in the production of auditory hallucinations in schizophrenia (Rajarethinam et al., 2000; Gaser et al., 2004). Hubl et al. (2004) also reported WM changes associated with auditory hallucinations of schizophrenic patients in the bilateral arcuate fasciculus and the anterior part of the corpus callosum. Recently, interest in WM abnormalities in schizophrenia has been increased and these abnormalities can be investigated more closely with the advanced neuroimaging or neuropathological methods (see review, Kubicki et al., 2007). For example, in vivo MRI includes two special techniques: DTI, which can provide important clues to the structure and geometric organization of WM, and voxel-based morphometry (VBM) of structural MR images, which has been useful in characterizing subtle changes in brain structure.

DTI is a non-invasive MRI technique that provides information about the molecular diffusion of water within the tissue and thus allows us to characterize intrinsic features of tissue microstructure and microdynamics (Basser, 1995). The diffusion anisotropy in WM is thought to have originated from the specific organization of bundles of myelinated axonal fibers running in parallel, although the exact mechanism of diffusion anisotropy is still not completely understood. Among the different indices for quantifying diffusion anisotropy, fractional anisotropy (FA) has been the most widely used index to evaluate the integrity of fiber tracts and myelination of neural axons in the WM (Basser and Pierpaoli, 1996; Le Bihan et al., 2001). In the analysis of schizophrenia using DTI, Buchsbaum et al. (1998) were the first to report a significantly lower FA value in the prefrontal WM and diminished fronto–striatal connectivity in patients with schizophrenia compared with controls. Subsequently, many investigators have identified disseminated or focal WM abnormalities in schizophrenia using DTI (for review, see Kubicki et al., 2007). The application of DTI in the investigation of WM abnormalities in schizophrenia has been recently reviewed (Kubicki et al., 2005). VBM of structural MRI is an objective technique that can be used to conduct a voxel-wise comparison of the probability of the presence of cerebral gray or white matter between subject groups, also described as the density or concentration of gray or white matter. White matter density observed in a VBM study may reflect the relative local amount of white matter tissue among the subjects (Honea et al., 2005). Spatially normalized structural brain images of different groups can be compared on a voxel-by-voxel basis. This method has a lower validity but a better internal consistency than MR volumetry with manual tracing (Wright et al., 1995). VBM of structural MR images was advanced after the optimized VBM method was introduced in 2001, and recently many investigators have used the optimized method for VBM of structural brain images (Good et al., 2001). VBM studies investigating WM abnormalities of schizophrenia have reported decreased WM densities in several regions, including the fronto–temporal area, the corpus callosum, the internal capsule, and the anterior commissure (Sigmundsson et al., 2001; Suzuki et al., 2002; Spalletta et al., 2003; Hulshoff Pol et al., 2004), whereas increased WM densities in the bilateral parietal area have been noted in schizophrenic patients (Suzuki et al., 2002). Gaser et al. (2004) reported correlations between the severity of auditory hallucination and volume loss in left Heschl's and supramarginal gyrus and right frontal gyri.

J.-H. Seok et al. / Psychiatry Research: Neuroimaging 156 (2007) 93–104

We have two research questions in this study. The first question investigates which WM abnormalities are associated with auditory hallucinations in patients with schizophrenia, and the second question asks whether WM abnormalities in schizophrenia may overlap in the voxel-based analysis of DTIs and structural MRIs. Since DTI and structural MRI techniques both have merit in detecting changes in WM tract integrity and volume, respectively, a combined examination of these two techniques can provide useful information. Theoretically, findings obtained from the two techniques may be different because they measure different dimensions of WM abnormalities. 2. Methods 2.1. Subjects A total of 15 hallucinating patients with schizophrenia, 15 non-hallucinating patients with schizophrenia and 22 age-matched, normal, healthy volunteers participated in this study. Patients were recruited from the outpatient clinics at the Severance Mental Health Hospital. All patients fulfilled the Diagnostic and Statistical Manual (DSM-IV) criteria for schizophrenia as diagnosed by two psychiatrists using the Structured Clinical Interview for DSM-IV (SCID) patient version (First et al., 1996). For each patient, the overall severity of schizophrenic symptoms was assessed using the Positive and Negative Syndrome Scale (PANSS) (Kay et al., 1987), and the severity of auditory hallucinations was evaluated using the Auditory Hallucination Subscale (AHS) of the Psychotic Symptom Rating Scales (Haddock et al., 1999). The normal, healthy volunteers were recruited from the local community and were screened for any current or lifetime history of a DSM-IV Axis I disorder using the SCID non-patient version. All subjects were right-handed, and none of the subjects reported a history of substance abuse before enrolling in this study. Subjects with a lifetime history of significant medical or neurological illnesses on medical history and physical examination were excluded. The study was approved by the Institutional Review Board for research involving human subjects. Written informed consent was obtained from all subjects before beginning the study. The hallucinating group consisted of patients with intractably enduring auditory hallucinations despite treatment with conventional dosages of atypical and/or typical antipsychotic medications. Total scores on the AHS had a mean of 22.7 (S.D. = 5.0). The nonhallucinating group included patients who had never

95

experienced auditory hallucinations during their illness. All patients received one or two antipsychotic medications including risperidone, olanzapine, clozapine, amisulpride and haloperidol. The mean chlorpromazine-equivalent doses of the hallucinating group (mean = 761.7, S.D. = 455.0) were higher than those of the non-hallucinating group (mean = 475.9, S.D. = 328.2), but the differences were not statistically significant (t = 1.973, P = 0.058). The proportions of men and women in the hallucinating group (8 male/7 female), non-hallucinating group (7 male/ 8 female) and the healthy control group (11 male/ 11 female) were not statistically significant (χ2 = 0.133, P = 0.936). The mean ages of these three groups were 29.1 years (S.D. = 4.3, range= 22–35), 30.0 years (S.D. = 3.9, range= 24–36) and 30.3 years (S.D. = 3.1, range = 25–35), respectively. The mean years of education were 12.1 years (S.D. = 1.3), 13.4 years (S.D. = 2.2) and 13.0 years (S.D. = 1.6), respectively. The ages and the education years were not significantly different among the three groups (age: F = 0.487, df= 2, 49, P = 0.617; education years: F = 2.380, df= 2, 49, P = 0.103). At the time of MR scanning, the mean duration of illness was not significantly different between the two patient groups (mean = 8.6 years, S.D. = 4.2 in the hallucinating group and mean = 6.3 years, S.D.= 3.5 in the non-hallucinating group, t = 1.590, P = 0.123). 2.2. MRI acquisition MR images were acquired at Severance Hospital using a Philips 1.5 T scanner (Philips Intera, Philips Medical System, Best, The Netherlands) with a SENSE head coil. Head motion was minimized with restraining foam pads offered by the manufacturer. To minimize artifacts produced by head movement, structural T1s and DTIs were sequentially acquired in a single scan time without position change. A high-resolution T1-weighted MRI volume data set was acquired axially with acquisition parameters of 256 × 256 acquisition matrix, 240-mm field of view, 0.9375 × 0.9375 × 1.5 mm3 voxels, TE 4.6 ms, TR 25 ms, and a flip angle of 30°. Diffusion-encoded images, parallel to the anterior commissure–posterior commissure line, were obtained using a single-shot echo-planar acquisition with the following parameters: 128 × 128 acquisition matrix; 220mm field of view; 1.72 × 1.72 × 2 mm3 voxels; about 45 axial slices; TE 62 ms; TR 7390 ms; flip angle 90°; slice gap 0 mm; 3 averaging per slice; b-factor of 600 smm− 2; non-cardiac gating. Diffusion-weighted images were acquired from 32 different directions with the baseline image being obtained without diffusion weighting. The

96

Regions

Cluster characteristics

Mean fractional anisotropy

Voxel number Zmax Coordinates x

y

Mean white matter density

Hall. group

Non-hall. group

Control group

z

(n = 15)

(n = 15)

(n = 22)

6

Post-hoc comparison † 0.34 a (0.04) 0.35 a

(0.03) 0.40 b

Left cingulum bundle Rostral part 86

3.64

− 14 39

Caudal part

4.66

− 14 −46 2

0.34 a

(0.07) 0.24 b

(0.03) 0.28 b

Left superior longitudinal fasciculus (SLF) Anterosuperior part 77 4.21

− 16 32

42

0.35 a

(0.07) 0.31 a

(0.04) 0.41 b

Anterior part

65

4.75

− 10 62

20

0.26 a

(0.02) 0.23 b

(0.03) 0.28 c

Middle part

63

3.79

− 22 − 24 36

0.49 a

(0.03) 0.44 b

(0.03) 0.49 a

4.38

− 32 − 50 − 42 0.34 a

(0.05) 0.30 a

(0.03) 0.41 b

118

Left middle 449 cerebellar peduncle

F Hall. group Non-hall. group Control F (2,49)⁎ (2,49)⁎ (n = 15) (n = 15) (n = 22) (0.03) 15.011§ 0.63 (0.02) (0.04) 14.390§ 0.28 (0.04)

0.64 (0.02) 0.27 (0.03)

0.64 (0.01) 0.28 (0.02)

0.817

(0.06) 13.504§ 0.42 (0.03) (0.02) 27.325§ 0.23 (0.02) (0.03) 14.037§ 0.81 (0.02) (0.08) 15.732§ 0.34 (0.02)

0.41 (0.03) 0.23 (0.03) 0.81 (0.02) 0.33 (0.01)

0.43 (0.02) 0.24 (0.02) 0.82 (0.01) 0.34 (0.02)

1.131

⁎ One-way ANOVA tests were performed for the comparison of mean FA and white matter density among the three groups. Data are presented as mean (S.D.). Post-hoc comparisons were performed by Tukey's HSD method, and homogenous subsets were displayed as superscripts of a, b or c. Hall., hallucinating; Non-hall., non-hallucinating.; FA, fractional anisotropy; WM, white matter; SLF, superior longitudinal fasciculus. §: P b 0.001.

†

0.870

1.268 0.671 2.125

J.-H. Seok et al. / Psychiatry Research: Neuroimaging 156 (2007) 93–104

Table 1 Clusters of white matter regions showing significant differences in fractional anisotropy (FA) among the hallucinating group, the non-hallucinating group and the normal control group, and the comparison of FA and WM density in the clusters

J.-H. Seok et al. / Psychiatry Research: Neuroimaging 156 (2007) 93–104

97

Fig. 1. Statistical parametric maps displaying significant differences among the hallucinating group, the non-hallucinating group and the normal control group (main effects of interest) ⁎. (a) Brain regions showing significant difference in the analysis of fractional anisotropy values. (b) Brain regions showing significant difference in the analysis of white matter densities. ⁎ Statistics: one-way ANOVA, uncorrected P b 0.001, voxel threshold k N 50. Coordinates in parentheses indicate the position of sagittal and coronal sections of magnetic resonance images. Abbreviations: SLF, superior longitudinal fasciculus; ILF, inferior longitudinal fasciculus.

spatial distortions induced by eddy currents in diffusionweighted images were corrected using the AIR5 algorithm (Kim et al., 2006) by registering the diffusionweighted images to the non-diffusion-weighted image. 2.3. Image preprocessing Preprocessing of diffusion-weighted images was performed using SPM2. After the DTIs were reconstructed

from diffusion-encoded images, eigenvalues of the diffusion tensor were used to calculate FA values for the entire voxel in the brain. FA maps were coregistered to structural T1 images with an affine transformation matrix derived from registering T2-weighted, b0-images of diffusion-weighted images to T1 images of each subject. FA images were spatially normalized into the Talairach space by applying parameters of normalized coregistered T2-weighted images to an MNI (Montreal

98

J.-H. Seok et al. / Psychiatry Research: Neuroimaging 156 (2007) 93–104

Table 2 Clusters of white matter (WM) regions showing significant differences in mean WM density in the hallucinating group, the non-hallucinating group and the normal control group, and the comparison of WM density and FA in the clusters Regions

Cluster characteristics

Mean white matter density

Mean fractional anisotropy

Voxel Zmax Coordinates number

Hall. group

Hall. group

x Left superior longitudinal fasciculus (sup. frontal WM) Genu of corpus callosum Cingulum bundle Left inferior longitudinal fasciculus (mid. temporal WM) Right uncal WM

y

z

167

3.56

− 26

4

48

729

3.49

4

24

5

4781

5.25

0 − 10

40

190

3.45

− 52 − 21

−6

195

3.42

15

− 7 − 20

Nonhall. group

Control F group (2,49)⁎

Nonhall. group

Control F (2,49)⁎

(n = 15) (n = 15) (n = 22)

(n = 15) (n = 15) (n = 22)

Post-hoc comparison † 0.61 a 0.59 a,b 0.57 b (0.02) (0.04) (0.03) 0.50 a 0.50 a 0.56 b (0.06) (0.05) (0.04) 0.26 a 0.26 a 0.24 b (0.01) (0.02) (0.01) 0.38 a 0.38 a 0.34 b (0.03) (0.03) (0.03) 0.24 a 0.23 a 0.21 b (0.03) (0.01) (0.02)

Post-hoc comparison † 0.28 0.27 0.28 (0.04) (0.03) (0.02) 0.29 0.29 0.31 (0.02) (0.03) (0.02) 0.63 0.64 0.64 (0.02) (0.02) (0.01) 0.34 0.33 0.34 (0.02) (0.01) (0.02) 0.20 a,b 0.18 a 0.21 b (0.04) (0.04) (0.03)

10.173 § 9.152 § 21.642 § 9.554 § 9.017 §

0.870 3.156 0.817 2.125 3.286‡

⁎One-way ANOVA tests were performed for the comparison of mean WM density and FA among the three groups. Data are presented as mean (S.D.). Post-hoc comparisons were performed by Tukey's HSD method, and homogenous subsets were displayed as superscripts of a or b. Hall., hallucinating; Non-hall., non-hallucinating; WM, white matter; FA, fractional anisotropy. §: P b 0.001, ‡ P b 0.05.

†

Neurological Institute, McGill University) T2 template. Normalized FA maps were then smoothed with a 6-mm full-width at half-maximum (FWHM) isotropic Gaussian kernel for SPM analysis. The structural T1 image-processing methods were based on an optimized, voxel-based morphometry technique (Good et al., 2001) and implemented in SPM2 (http://www.fil.ion.ucl.ac.uk/spm/software/). Anatomical T1 images from 52 subjects were spatiallynormalized to the MNI standard T1 template in standard Talairach space (Talairach and Tournoux, 1988). The normalized data were then smoothed with an 8-mm FWHM isotropic Gaussian kernel, and an average T1 template was created. Thereafter, study-specific probabilistic maps of grey matter (GM), white matter (WM) and cerebro-spinal fluid space (CSF) were created by averaging the 52 smoothed GM, WM and CSF segmentations of spatially normalized individual brain images. The final WM images were extracted from segmenting the individual raw T1 images using GM, WM and CSF probability maps derived at the first step, and then were used for the second smoothing step. We did not apply the modulation step for this analysis. Segmented WM images were smoothed with a 14-mm FWHM isotropic Gaussian kernel for SPM analysis. 2.4. Statistical analysis First, a voxel-wise one-way analysis of variance (ANOVA) was performed with the factor “group” (hallu-

cinating group, non-hallucinating group and control group) as a single factor for FA or WM density. We found WM regions with significant differences among the three groups at a statistical threshold of an uncorrected P b 0.001 (Z = 3.09) consisting of a minimum of 50 neighboring voxels. We defined these regions as regions of interest (ROIs). Next, the mean regional FA and WM density values corresponding to these ROIs in FA analyses were calculated using the individual FA and WM density maps. For the ROIs in WM density analysis, the same calculations of mean regional FA and WM density were performed. Post hoc independent t tests were applied between the groups for clinical data and mean ROI values when appropriate. For multiple comparisons between groups, Tukey's HSD correction method was applied. Pearson's correlation coefficients between the ROI values (mean FA value or WM density of each ROI from FA and WM density analysis) and the clinical symptom score (AHS score in the hallucinating group or the PANSS dimensional scores in the two patient groups) were computed for correlation analysis. Because the segmentation around the region of the basal ganglia was inconsistent across subjects and we only focused on WM changes in the present study, this region was excluded from analysis. Clusters were assigned to the underlying WM fiber tracts using three-dimensional anatomical data and a fiber tract-based atlas of human WM (Wakana et al., 2004). Data analysis was done using SPM2 and SPSS 13.0 software. ROI visualization was realized with MRIcro.

J.-H. Seok et al. / Psychiatry Research: Neuroimaging 156 (2007) 93–104

3. Results 3.1. WM changes found in FA maps As shown in Table 1 and Fig. 1(a), two clusters in the cingulum bundle, three clusters in the superior longitudinal fasciculus (SLF) and one cluster in the middle cerebellar peduncle in the left hemisphere were identified which showed significant differences among the three groups. Post hoc analysis revealed that FAs in the rostral part of the cingulum bundle, the anterior part of the SLF and the middle cerebellar peduncle were significantly reduced in both patient groups compared with the control group. The mean FA in the middle part of the SLF was significantly reduced only in the nonhallucinating group compared with to the control group, and the mean FA values of the ROI sets in the left SLF in the hallucinating group were significantly higher than those in the non-hallucinating group. Meanwhile, the FA in the caudal part of the cingulum bundle was significantly increased in the hallucinating group compared with both the control group and the nonhallucinating group. In addition, the mean WM densities in these ROIs were not significantly different among the three groups. 3.2. WM changes found in structural WM density maps There were five regions showing significant differences in mean WM densities among the three groups. As shown in Table 2 and Fig. 1(b), WM density in the ROI of the left SLF was significantly increased only in the hallucinating group relative to the control group. WM density in the genu of the corpus callosum was

99

significantly decreased in both patient groups compared with the control group. However, WM densities in the ROIs in the bilateral cingulum bundle, the left inferior longitudinal fasciculus (ILF) and the right uncal WM were significantly increased in both patient groups. In addition, mean FAs in these ROIs were not significantly different among the three groups except in the right uncal WM ROI. The mean FA in the right uncal WM was significantly decreased in the non-hallucinating group compared with the control group. 3.3. Relationship between WM variables and symptom severity in the patient groups As shown in Fig. 2, the severity score on the AHS was significantly correlated with the mean FA value in the ROI in the anterosuperior part of the left SLF in the hallucinating group (r = 0.679, df = 13, P b 0.01), and the mean FA in this ROI was also correlated with a positive (r = 0.420, df = 28, P b 0.05) and a general psychopathology (r = 0.402, df = 28, P b 0.05) score on the PANSS in the two patient groups. The mean FA value in the ROI in the caudal part of the left cingulum bundle was also correlated with a positive symptom score on the PANSS (r = 0.445, df = 28, P b 0.05) and with the dosage of antipsychotic medication (r = 0.426, df = 28, P b 0.05). When controlling for antipsychotic medication dosage, a partial correlation of a positive symptom score on the PANSS was significant only with the mean FA value of the ROI in the left SLF (r = 0.438, df = 27, P b 0.05) and not with the mean FA value of the ROI in the caudal part of the left cingulum bundle (r = 0.344, df = 27, P = 0.068). A mean WM density of the ROI in right uncal WM was positively correlated with a negative symptom score on

Fig. 2. The cluster on the left superior longitudinal fasciculus showing positive correlations between mean fractional anisotropy (FA) values and severity scores on the Auditory Hallucination (AH) Scale in the hallucinating group. ⁎ White arrow indicates the cluster on the left superior longitudinal fasciculus. The hallucinating patient group showed significant positive correlations between mean FA values and the scores on the Auditory Hallucination Scale in this region (r = 0.679, df = 13, P b 0.01).

100

J.-H. Seok et al. / Psychiatry Research: Neuroimaging 156 (2007) 93–104

the PANSS (r = 0.412, df = 28, P b 0.05). In other ROI sets from FA or WM density analyses, there was no significant correlation between the severity of auditory hallucinations or other symptom scores and the mean FA values or WM densities. 4. Discussion In this study, we investigated WM abnormalities and their relationships to the severity of symptoms in schizophrenia using two different neuroimaging techniques, including DTI and structural MRI. Most regions showing significant differences among the two patient and the healthy control groups were located in the left hemisphere, except for the right uncus. This finding supports the evidence for left hemispheric dysfunction in patients with schizophrenia (Carter et al., 1996) and is consistent with a previous study (Spalletta et al., 2003). Significant WM changes occurred in the frontal and temporal regions consisting of a long association fiber tract such as the SLF and the ILF. A significant reduction of FA in the frontal part of the SLF was observed in the patient groups compared with the control group. Significant increases of WM density in the frontal part of the SLF and the temporal part of the ILF were also observed in the patient groups. These findings may be evidence of the hypofrontality (Weinberger and Berman, 1996) and fronto–temporal disconnectivity of WM (Lawrie et al., 2002; Burns et al., 2003) observed in schizophrenia, and they were somewhat consistent with the previous studies (Buchsbaum et al., 1998; Lim et al., 1999; Ardekani et al., 2003). Some parts of the frontal SLF and the temporal ILF contain the arcuate fasciculus, which connects to the fronto–temporal language association cortex, even though other parts of the SLF and the ILF have extensive connections throughout the entire cortical area from the frontal and temporal poles to the occipital pole, respectively (Wakana et al., 2004). Another focus of this study was whether WM abnormalities were associated with auditory hallucinations in schizophrenia. There have been continuing reports about structural abnormalities (Barta et al., 1990; Rajarethinam et al., 2000; Gaser et al., 2004) as well as functional abnormalities (McGuire et al., 1993; Silbersweig et al., 1995; Dierks et al., 1999) of the brain regions related to auditory information processing in schizophrenic patients with auditory hallucinations. In contrast to GM abnormalities, which have been reported as abnormal activations of the fronto–temporal cortex (including Heschl's gyrus) during auditory hallucinations (Dierks et al., 1999; Lennox et al., 2000; Ropohl et al.,

2004), WM abnormalities associated with auditory hallucinations have yet to be sufficiently investigated. The hallucinating group showed higher mean FA and WM density of ROIs in the left SLF compared with the non-hallucinating group, even though differences in mean FA and WM density of ROIs might not reach a statistically significant threshold. Interestingly, the mean FA value of the ROI in the superior frontal part of the left SLF was positively correlated with the severity score of auditory hallucinations and the positive symptoms score of the PANSS in the hallucinating group. The workload for additional erroneous auditory information processing may contribute to spare WM tract integrity and local tissue volume in this region in the hallucinating group. The frontal part of the SLF was also considered to constitute a frontal portion of the arcuate fasciculus. Taken together, it seems that this region may play an important role in the pathophysiology of auditory hallucinations in schizophrenia. In fact, previous DTI studies have reported that FA of the lateral part of the left SLF was increased in patients with auditory hallucinations (Hubl et al., 2004). Furthermore, this region may be a structural basis of dysfunctional connectivity in schizophrenia (Lawrie et al., 2002) and erroneous processing of auditory information which induces auditory hallucinations (Lawrie and Abukmeil, 1995). In a correlation analysis of this study, positive correlations between the severities of positive schizophrenic symptoms and the mean FA values of the ROIs in the left SLF were noted. Recently, there have been DTI studies suggesting that regional FA values could reflect the severities of schizophrenic symptoms. For example, a significant inverse correlation was observed between negative symptoms and WM anisotropy (Wolkin et al., 2003). We also found significant WM alterations in the cingulum bundle in both hallucinating and non-hallucinating groups. The cingulum bundle is a major limbic system fiber tract connecting the cingulate gyrus with other limbic structures and cortical areas. A significant reduction of FA in the rostral part of the cingulum bundle in both patient groups was consistent with a previous study (Sun et al., 2003). The anterior cingulate cortex is strongly interconnected with the amygdala, the nucleus accumbens, the dorsomedial nucleus of the thalamus, and the dorsolateral prefrontal cortex (Vogt et al., 1979). Cortico-limbic functional disconnectivity in schizophrenia may be at work through this regional abnormality. In addition, this study reported that the mean FA of the ventral part of the caudal cingulum and the mean WM density in the dorsal part of the cingulum were significantly increased in both patient groups.

J.-H. Seok et al. / Psychiatry Research: Neuroimaging 156 (2007) 93–104

Previous DTI studies, which used a predefined ROI approach and mainly focused on the FA of the dorsal part of the cingulum bundle, reported a decrease in the FA and the mean area of the cingulum bundle in patients with schizophrenia (Kubicki et al., 2003; Sun et al., 2003). We used a voxel-wise analysis for the whole WM area without an a priori hypothesis. The cingulum bundle has different connections and functions in the anterior and posterior parts of the cingulate gyrus (Vogt et al., 1992). The posterior cingulate has been known to be involved in spatial orientation and sensory monitoring. Increased FA in the posterior cingulum bundle and its correlation with a positive symptom score and antipsychotic dosage may reflect compensatory hyperfunction in spatial orientation and erroneous selfmonitoring in schizophrenia, even though this assumption needs to be validated. This study suggests that the pathological process of the cingulum bundle in schizophrenia may be differently affected according to the specific part of the cingulum bundle. Decreased FA in the left middle cerebellar peduncle was also observed in the patient groups. The cerebellum has extensive connections with the cerebral cortex, the thalamus and the other subcortical structures, and has been known to play an important role in the coordination of cognitive function as well as motor function. Since Andreasen et al. (1996) suggested that cerebellar dysfunction may underlie the pathophysiology of schizophrenia, interest in cerebellar dysfunction in schizophrenia has increased. The middle cerebellar peduncle is a major pathway to the cerebellum in the cortico–ponto–cerebellar tract, and the cerebrum has influences on the cerebellum via the pons through this tract (Afifi and Bergman, 1998). Disconnectivity in the middle cerebellar peduncle may contribute to cerebellar dysfunction in schizophrenia. Decreased FA in the middle cerebellar peduncle was found in some previous studies (Okugawa et al., 2004, 2005), but not in others (Wang et al., 2003). We found a reduction of WM density only in the genu of the corpus callosum in both patient groups when compared with the control group. A previous study (Hulshoff Pol et al., 2004) reported decreased WM density in the genu of the corpus callosum and that the overall severity of the illness was negatively correlated with WM densities in the bilateral corpus callosum and the anterior commissure. However, we did not observe correlations between WM density of the corpus callosum and symptom factors. Mean WM densities in several WM regions, including the frontal and temporal areas and the dorsal cingulum bundle, were significantly increased in the patient groups. These results are quite

101

different from previous VBM studies of schizophrenia. Most VBM studies investigating WM abnormalities in schizophrenia have reported volume deficits in several WM regions (Honea et al., 2005). Various factors may contribute these findings. The mean age of the subjects in this study was relatively young and their clinical course seemed to be relatively acute and mild as compared with previous WM VBM studies on schizophrenia (Sigmundsson et al., 2001; Ananth et al., 2002; Spalletta et al., 2003; Hulshoff Pol et al., 2004). Pathophysiological changes of WM occurring in the early phase of schizophrenia may be different from those occurring in the later phase. There have been reports of increased interstitial neuronal densities in WM in schizophrenia (Akbarian et al., 1996; Eastwood and Harrison, 2003, 2005). Increased interstitial neurons may contribute to an increase in WM volume density in these regions. Another study also reported increased WM density in the parietal lobe (Suzuki et al., 2002). White matter volume change observed in this study may reflect a WM change in younger patients with schizophrenia. In addition, we found an increased WM density in the two patient groups and a decreased mean FA in the uncal WM only in the non-hallucinating group. This finding may be related to disconnectivity in the limbic network. However, the difference may be attributed to an incomplete segmentation (such as a periventricular region) because this region is surrounded by CSF space between the brain stem and the medial temporal lobe. As discussed in a recent review on a VBM study for structural MRI in schizophrenia (Honea et al., 2005), diverse results may be due to a different choice of variables in the automated process, such as smoothing kernel size and linear versus non-linear transformation, as well as to differences in the patient groups. As expected, ROI sets from FA and WM density analyses did not overlap even though these ROI sets were located in the same fasciculus such as the SLF or the ILF. This result was consistent with a previous study with a similar approach (Burns et al., 2003). This finding supports the hypothesis that the voxel-wise analyses of DTI and structural MRI may reflect qualitatively different dimensions of WM characteristics, which are tract integrity (Basser and Pierpaoli, 1996) and local composition of brain tissue (Wright et al., 1995), respectively. There are several limitations in this study. We used an uncorrected P value for the threshold for significance in statistical parametric mapping. Smoothing kernel size is one of the issues in the field of DTI analysis (Park et al., 2004; Jones et al., 2005). From our experience, we

102

J.-H. Seok et al. / Psychiatry Research: Neuroimaging 156 (2007) 93–104

thought 6 × 6 × 6 mm FWHM appropriate for detecting narrow fiber structures in diffusion tensor images. Various kernel sizes have been used for VBM analysis of structural MR images (Honea et al., 2005). In order to reduce errors in segmentation of WM structures and misalignment during spatial normalization, we used 14 × 14 × 14 mm FWHM for VBM analysis of structural T1 images. The hallucinating patient group also took slightly higher dosages of antipsychotic medications than the non-hallucinating patient group, even though the difference in antipsychotic dosages between the two patient groups was not statistically significant. This factor may have affected the difference between the mean FA value and WM density among the two patient groups and the control group. However, it has not been reported that antipsychotic medication may induce significant WM changes, even though there have been reports that the basal ganglia volume was increased after typical antipsychotic medication and decreased or unchanged after atypical antipsychotic medication (Corson et al., 1999; Lang et al., 2004), as well as a report that gray matter volume was decreased after haloperidol medication and not significantly changed after olanzapine medication (Lieberman et al., 2005). There was no significant correlation between mean FA or WM density of ROIs and the dosage of antipsychotic medication in the patient subjects of this study. In conclusion, we simultaneously investigated the changes of WM tract integrity and density in schizophrenia using voxel-wise analyses of DTI and structural WM images. We found that abnormal WM changes were observed mainly in the left hemispheres of the patients with schizophrenia. WM changes primarily observed in the frontal and temporal areas and disconnectivity in the left fronto–temporal area may contribute to the pathophysiology of schizophrenia. In particular, the mean FA value of the left frontal part of the SLF was positively correlated with the severity score of auditory hallucinations in the hallucinating patient group, and this group showed pathologically strengthened tract integrity of the left SLF compared with the non-hallucinating patient or normal control groups. Therefore, pathologic WM changes in this region may be an important step in the development of auditory hallucinations in schizophrenia. There was no overlapping WM region in either modality showing a significant change in schizophrenia, and thus DTI and structural VBM seem to reflect different dimensions of WM changes in schizophrenia. Further studies are needed to elaborate the WM abnormalities of patients with schizophrenia and their clinical implications in the treatment of this debilitating disorder.

Acknowledgement This study was supported by a grant of the Korea Science and Engineering Foundation, Interdisciplinary Research (Contract grant number: R01-2002-00000362-0). References Afifi, A.K., Bergman, R.A., 1998. Cerebellum. In: Afifi, A.K., Bergman, R.A. (Eds.), Functional Neuroanatomy, Text and Atlas, International Ed. McGraw–Hill, New York, pp. 303–329. Akbarian, S., Kim, J.J., Potkin, S.G., Hetrick, W.P., Bunney Jr., W.E., Jones, E.G., 1996. Maldistribution of interstitial neurons in prefrontal white matter of the brains of schizophrenic patients. Archives of General Psychiatry 53, 425–436. Ananth, H., Popescu, I., Critchley, H.D., Good, C.D., Frackowiak, R.S., Dolan, R.J., 2002. Cortical and subcortical gray matter abnormalities in schizophrenia determined through structural magnetic resonance imaging with optimized volumetric voxel-based morphometry. American Journal of Psychiatry 159, 1497–1505. Andreasen, N.C., O'Leary, D.S., Cizadlo, T., Arndt, S., Rezai, K., Ponto, L.L., Watkins, G.L., Hichwa, R.D., 1996. Schizophrenia and cognitive dysmetria: a positron-emission tomography study of dysfunctional prefrontal–thalamic–cerebellar circuitry. Proceedings of the National Academy of Sciences of the United States of America 93, 9985–9990. Andreasen, N.C., Paradiso, S., O'Leary, D.S., 1998. “Cognitive dysmetria” as an integrative theory of schizophrenia: a dysfunction in cortical–subcortical–cerebellar circuitry? Schizophrenia Bulletin 24, 203–218. Ardekani, B.A., Nierenberg, J., Hoptman, M.J., Javitt, D.C., Lim, K.O., 2003. MRI study of white matter diffusion anisotropy in schizophrenia. Neuroreport 14, 2025–2029. Barta, P.E., Pearlson, G.D., Powers, R.E., Richards, S.S., Tune, L.E., 1990. Auditory hallucinations and smaller superior temporal gyral volume in schizophrenia. American Journal of Psychiatry 147, 1457–1462. Basser, P.J., 1995. Inferring microstructural features and the physiological state of tissues from diffusion-weighted images. NMR in Biomedicine 8, 333–344. Basser, P.J., Pierpaoli, C., 1996. Microstructural and physiological features of tissues elucidated by quantitative-diffusion-tensor MRI. Journal of Magnetic Resonance. Series B 111, 209–219. Buchsbaum, M.S., Tang, C.Y., Peled, S., Gudbjartsson, H., Lu, D., Hazlett, E.A., Downhill, J., Haznedar, M., Fallon, J.H., Atlas, S.W., 1998. MRI white matter diffusion anisotropy and PET metabolic rate in schizophrenia. Neuroreport 9, 425–430. Burns, J., Job, D., Bastin, M.E., Whalley, H., Macgillivray, T., Johnstone, E.C., Lawrie, S.M., 2003. Structural disconnectivity in schizophrenia: a diffusion tensor magnetic resonance imaging study. British Journal of Psychiatry 182, 439–443. Carter, C.S., Robertson, L.C., Nordahl, T.E., Chaderjian, M., OshoraCelaya, L., 1996. Perceptual and attentional asymmetries in schizophrenia: further evidence for a left hemisphere deficit. Psychiatry Research 62, 111–119. Corson, P.W., Nopoulos, P., Miller, D.D., Arndt, S., Andreasen, N.C., 1999. Change in basal ganglia volume over 2 years in patients with schizophrenia: typical versus atypical neuroleptics. American Journal of Psychiatry 156, 1200–1204.

J.-H. Seok et al. / Psychiatry Research: Neuroimaging 156 (2007) 93–104 Dierks, T., Linden, D.E., Jandl, M., Formisano, E., Goebel, R., Lanfermann, H., Singer, W., 1999. Activation of Heschl's gyrus during auditory hallucinations. Neuron 22, 615–621. Eastwood, S.L., Harrison, P.J., 2003. Interstitial white matter neurons express less reelin and are abnormally distributed in schizophrenia: towards an integration of molecular and morphologic aspects of the neurodevelopmental hypothesis. Molecular Psychiatry 8, 821–831. Eastwood, S.L., Harrison, P.J., 2005. Interstitial white matter neuron density in the dorsolateral prefrontal cortex and parahippocampal gyrus in schizophrenia. Schizophrenia Research 79, 181–188. First, M.B., Spitzer, R.L., Gibbon, M., Williams, J.B.W., 1996. Structured Clinical Interview for DSM-IV Axis I Disorders. New York State Psychiatric Institute, Biometric Research, New York. Gaser, C., Nenadic, I., Volz, H.P., Buchel, C., Sauer, H., 2004. Neuroanatomy of “hearing voices”: a frontotemporal brain structural abnormality associated with auditory hallucinations in schizophrenia. Cerebral Cortex 14, 91–96. Good, C.D., Johnsrude, I.S., Ashburner, J., Henson, R.N., Friston, K.J., Frackowiak, R.S., 2001. A voxel-based morphometric study of ageing in 465 normal adult human brains. Neuroimage 14, 21–36. Haddock, G., McCarron, J., Tarrier, N., Faragher, E.B., 1999. Scales to measure dimensions of hallucinations and delusions: the psychotic symptom rating scales (PSYRATS). Psychological Medicine 29, 879–889. Honea, R., Crow, T.J., Passingham, D., Mackay, C.E., 2005. Regional deficits in brain volume in schizophrenia: a meta-analysis of voxelbased morphometry studies. American Journal of Psychiatry 162, 2233–2245. Hubl, D., Koenig, T., Strik, W., Federspiel, A., Kreis, R., Boesch, C., Maier, S.E., Schroth, G., Lovblad, K., Dierks, T., 2004. Pathways that make voices: white matter changes in auditory hallucinations. Archives of General Psychiatry 61, 658–668. Hulshoff Pol, H.E., Schnack, H.G., Mandl, R.C., Cahn, W., Collins, D.L., Evans, A.C., Kahn, R.S., 2004. Focal white matter density changes in schizophrenia: reduced inter-hemispheric connectivity. Neuroimage 21, 27–35. Johnstone, E.C., Crow, T.J., Frith, C.D., Husband, J., Kreel, L., 1976. Cerebral ventricular size and cognitive impairment in chronic schizophrenia. Lancet 2, 924–926. Jones, D.K., Symms, M.R., Cercignani, M., Howard, R.J., 2005. The effect of filter size on VBM analyses of DT-MRI data. Neuroimage 26, 546–554. Kay, S.R., Fiszbein, A., Opler, L.A., 1987. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophrenia Bulletin 13, 261–276. Kim, D.J., Park, H.J., Kang, K.W., Shin, Y.W., Kim, J.J., Moon, W.J., Chung, E.C., Kim, I.Y., Kwon, J.S., Kim, S.I., 2006. How does distortion correction correlate with anisotropic indices? A diffusion tensor imaging study. Magnetic Resonance Imaging 24, 1369–1376. Kraepelin, E., 1919/1971. Dementia Praecox. Churchill Livingstone, New York. Kubicki, M., Westin, C.F., Nestor, P.G., Wible, C.G., Frumin, M., Maier, S.E., Kikinis, R., Jolesz, F.A., McCarley, R.W., Shenton, M.E., 2003. Cingulate fasciculus integrity disruption in schizophrenia: a magnetic resonance diffusion tensor imaging study. Biological Psychiatry 54, 1171–1180. Kubicki, M., Westin, C.F., McCarley, R.W., Shenton, M.E., 2005. The application of DTI to investigate white matter abnormalities in schizophrenia. Annals of the New York Academy of Sciences 1064, 134–148. Kubicki, M., McCarley, R., Westin, C.F., Park, H.J., Maier, S., Kikinis, R., Jolesz, F.A., Shenton, M.E., 2007. A review of diffusion tensor

103

imaging studies in schizophrenia. Journal of Psychiatric Research 41, 15–30. Lang, D.J., Kopala, L.C., Vandorpe, R.A., Rui, Q., Smith, G.N., Goghari, V.M., Lapointe, J.S., Honer, W.G., 2004. Reduced basal ganglia volumes after switching to olanzapine in chronically treated patients with schizophrenia. American Journal of Psychiatry 161, 1829–1836. Lawrie, S.M., Abukmeil, S.S., 1995. Auditory hallucinations and perception of external speech. Lancet 346, 1036. Lawrie, S.M., Buechel, C., Whalley, H.C., Frith, C.D., Friston, K.J., Johnstone, E.C., 2002. Reduced frontotemporal functional connectivity in schizophrenia associated with auditory hallucinations. Biological Psychiatry 51, 1008–1011. Le Bihan, D., Mangin, J.F., Poupon, C., Clark, C.A., Pappata, S., Molko, N., Chabriat, H., 2001. Diffusion tensor imaging: concepts and applications. Journal of Magnetic Resonance Imaging 13, 534–546. Lennox, B.R., Park, S.B., Medley, I., Morris, P.G., Jones, P.B., 2000. The functional anatomy of auditory hallucinations in schizophrenia. Psychiatry Research: Neuroimaging 100, 13–20. Lieberman, J.A., Tollefson, G.D., Charles, C., Zipursky, R., Sharma, T., Kahn, R.S., Keefe, R.S., Green, A.I., Gur, R.E., McEvoy, J., Perkins, D., Hamer, R., Gu, M., Tohen, H., 2005. Antipsychotic drug effects on brain morphology in first-episode psychosis. Archives of General Psychiatry 62, 361–370. Lim, K.O., Hedehus, M., Moseley, M., de Crespigny, A., Sullivan, E.V., Pfefferbaum, A., 1999. Compromised white matter tract integrity in schizophrenia inferred from diffusion tensor imaging. Archives of General Psychiatry 56, 367–374. McGuire, P.K., Shah, G.M., Murray, R.M., 1993. Increased blood flow in Broca's area during auditory hallucinations in schizophrenia. Lancet 342, 703–706. Okugawa, G., Nobuhara, K., Minami, T., Tamagaki, C., Takase, K., Sugimoto, T., Sawada, S., Kinoshita, T., 2004. Subtle disruption of the middle cerebellar peduncles in patients with schizophrenia. Neuropsychobiology 50, 119–123. Okugawa, G., Nobuhara, K., Sugimoto, T., Kinoshita, T., 2005. Diffusion tensor imaging study of the middle cerebellar peduncles in patients with schizophrenia. Cerebellum 4, 123–127. Park, H.J., Westin, C.F., Kubicki, M., Maier, S.E., Niznikiewicz, M., Baer, A., Frumin, M., Kikinis, R., Jolesz, F.A., McCarley, R.W., Shenton, M.E., 2004. White matter hemisphere asymmetries in healthy subjects and in schizophrenia: a diffusion tensor MRI study. Neuroimage 23, 213–223. Rajarethinam, R.P., DeQuardo, J.R., Nalepa, R., Tandon, R., 2000. Superior temporal gyrus in schizophrenia: a volumetric magnetic resonance imaging study. Schizophrenia Research 41, 303–312. Ropohl, A., Sperling, W., Elstner, S., Tomandl, B., Reulbach, U., Kaltenhauser, M., Kornhuber, J., Maihofner, C., 2004. Cortical activity associated with auditory hallucinations. Neuroreport 15, 523–526. Shenton, M.E., Dickey, C.C., Frumin, M., McCarley, R.W., 2001. A review of MRI findings in schizophrenia. Schizophrenia Research 49, 1–52. Sigmundsson, T., Suckling, J., Maier, M., Williams, S., Bullmore, E., Greenwood, K., Fukuda, R., Ron, M., Toone, B., 2001. Structural abnormalities in frontal, temporal, and limbic regions and interconnecting white matter tracts in schizophrenic patients with prominent negative symptoms. American Journal of Psychiatry 158, 234–243. Silbersweig, D.A., Stern, E., Frith, C., Cahill, C., Holmes, A., Grootoonk, S., Seaward, J., McKenna, P., Chua, S.E., Schnorr, L., Jones, T., Frackowiak, R.S.J., 1995. A functional neuroanatomy of hallucinations in schizophrenia. Nature 378, 176–179.

104

J.-H. Seok et al. / Psychiatry Research: Neuroimaging 156 (2007) 93–104

Spalletta, G., Tomaiuolo, F., Marino, V., Bonaviri, G., Trequattrini, A., Caltagirone, C., 2003. Chronic schizophrenia as a brain misconnection syndrome: a white matter voxel-based morphometry study. Schizophrenia Research 64, 15–23. Sun, Z., Wang, F., Cui, L., Breeze, J., Du, X., Wang, X., Cong, Z., Zhang, H., Li, B., Hong, N., Zhang, D., 2003. Abnormal anterior cingulum in patients with schizophrenia: a diffusion tensor imaging study. Neuroreport 14, 1833–1836. Suzuki, M., Nohara, S., Hagino, H., Kurokawa, K., Yotsutsuji, T., Kawasaki, Y., Takahashi, T., Matsui, M., Watanabe, N., Seto, H., Kurachi, M., 2002. Regional changes in brain gray and white matter in patients with schizophrenia demonstrated with voxelbased analysis of MRI. Schizophrenia Research 55, 41–54. Talairach, J., Tournoux, P., 1988. Co-plana Stereotaxic Atlas of the Human Brain. Thieme, New York. Vogt, B.A., Rosene, D.L., Pandya, D.N., 1979. Thalamic and cortical afferents differentiate anterior from posterior cingulate cortex in the monkey. Science 204, 205–207. Vogt, B.A., Finch, D.M., Olson, C.R., 1992. Functional heterogeneity in cingulate cortex: the anterior executive and posterior evaluative regions. Cerebral Cortex 2, 435–443. Wakana, S., Jiang, H., Nagae-Poetscher, L.M., van Zijl, P.C., Mori, S., 2004. Fiber tract-based atlas of human white matter anatomy. Radiology 230, 77–87.

Wang, F., Sun, Z., Du, X., Wang, X., Cong, Z., Zhang, H., Zhang, D., Hong, N., 2003. A diffusion tensor imaging study of middle and superior cerebellar peduncle in male patients with schizophrenia. Neuroscience Letters 348, 135–138. Weinberger, D.R., Berman, K.F., 1996. Prefrontal function in schizophrenia: confounds and controversies. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 351, 1495–1503. Wolkin, A., Choi, S.J., Szilagyi, S., Sanfilipo, M., Rotrosen, J.P., Lim, K.O., 2003. Inferior frontal white matter anisotropy and negative symptoms of schizophrenia: a diffusion tensor imaging study. American Journal of Psychiatry 160, 572–574. Wright, I.C., McGuire, P.K., Poline, J.B., Travere, J.M., Murray, R.M., Frith, C.D., Frackowiak, R.S., Friston, K.J., 1995. A voxel-based method for the statistical analysis of gray and white matter density applied to schizophrenia. Neuroimage 2, 244–252.