- Email: [email protected]
Application of a new image analysis technique to study brain asymmetry in schizophrenia
Psychiatry Research: Neuroimaging 124 (2003) 25–35
Application of a new image analysis technique to study brain asymmetry in schizophrenia Clare E. Mackaya,*, Thomas R. Barrickb, Neil Robertsb, Lynn E. DeLisic, Frederik Maesd, Dirk Vandermeulend, Timothy J. Crowa Prince of Wales International Centre for SANE Research (POWIC), Department of Psychiatry, University of Oxford, Warneford Hospital, Headington, Oxford OX3 7JX, UK b Magnetic Resonance and Image Analysis Research Centre (MARIARC), Pembroke Place, University of Liverpool, Liverpool L69 3BX, UK c Department of Psychiatry, New York University School of Medicine, 650 First Avenue, New York, NY 10016, USA d Department of Electrical Engineering, Medical Image Computing, ESAT-PSI, K.U. Leuven, Belgium
a
Received 13 August 2002; received in revised form 9 May 2003; accepted 29 May 2003
Abstract The hypothesis that normal brain torque (i.e. rightward frontal and leftward occipital asymmetry) is anomalous in schizophrenia (Crow, 1997. Trends in Neuroscience, 20, 339–343) was tested by application of a novel image analysis technique on three-dimensional magnetic resonance images obtained in 26 adult patients with chronic schizophrenia (18 males, 8 females) and 24 controls (14 males, 10 females). Right and left cerebral hemisphere tissue was extracted via non-linear co-registration with a mask image, and maps were computed of inter-hemispheric differences in tissue volume in an array of columns of voxels orthogonal to the mid-plane (2D), and profiles of coronal slice volumes (1D). Furthermore, integration of two-dimensional column maps gave approximate lobar asymmetries, and occipital and frontal asymmetries were combined to give a volumetric measure of brain torque. Significant brain torque was revealed in male and female control and patient groups, and did not correlate with brain size. Frontal and occipital asymmetries were significantly correlated in all groups. Both frontal and occipital components of torque were significantly increased in males than females. Patients tended to have reduced torque, particularly the leftward occipital component. Furthermore, 3y26 patients (but no controls) had reversed torque (leftward frontal and rightward occipital asymmetry). Contrary to Crow’s hypothesis, brain torque was not significantly reduced in patients with schizophrenia relative to controls, although reversal of torque was found in three cases. Future studies with larger sample sizes should consider sexual dimorphism and specific symptoms in relation to asymmetry. 䊚 2003 Elsevier Ireland Ltd. All rights reserved.
Keywords: MRI; Psychosis; Cerebral asymmetry; Sex differences
*Corresponding author. Tel.: q44-1-865-223-779; fax: q44-1-865-244-990. E-mail address: [email protected] (C.E. Mackay). 0925-4927/03/$ - see front matter 䊚 2003 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/S0925-4927(03)00088-X
26
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
1. Introduction A method for assessment of global brain asymmetry is necessary to address the central question of whether psychosis can be considered as a primary anomaly of cerebral lateralisation (Crow, 1990, 1997). The notion of an alteration in the usual brain ‘torque’ (rightward frontal and leftward occipital asymmetry), sometimes referred to as the Yakovlevian torque (LeMay, 1976; Weinberger et al., 1982), has previously been investigated in patients with schizophrenia using width or volume measurements of the hemispheres in specific frontal (anterior) and occipital (posterior) regions. Loss of asymmetries of brain width were reported in some patients with schizophrenia by Luchins et al. (1979), but not seen in a later study by the same author (Luchins and Meltzer, 1983). Reduced width asymmetry relative to controls has been reported in early onset patients, (Crow et al., 1998; Maher et al., 1998), first episode patients (DeLisi et al., 1997) and a large series of chronic patients (Falkai et al., 1995). The first volumetric study of brain torque was by Bilder et al. (1994), who applied planimetry on successive coronal MRI sections in first episode patients and reported that torque was significantly diminished compared with findings in controls. More recently, this group reported similar reductions of lesser magnitude in patients with affective disorders (Bilder et al., 1999). Although not investigated by Bilder et al. (1994), an important aspect of cerebral torque is that a sex difference has been reported, such that males have a greater magnitude of torque than females (Bear et al., 1986; Barrick et al., 2001). Sex differences in width asymmetries have also been observed in patients with schizophrenia (e.g. DeLisi et al., 1994; Falkai et al., 1995). We suggest that the study of torque therefore has the potential to elucidate the neuroanatomical basis of sex differences associated with psychosis. We have developed new techniques for measuring cerebral asymmetry in two-dimensional maps referring to arrays of columns of voxels orthogonal to the interhemispheric plane and one-dimensional profiles of coronal slice volumes oriented along the anterior-posterior axis of three-dimensional MR images (Barrick et al., 2001; Barrick, 2003) and
applied these techniques in the study of 26 adult patients with chronic schizophrenia (18 males, 8 females) and 24 controls (14 males, 10 females). Firstly, we have investigated whether there is reduction or reversal of global asymmetry in patients with schizophrenia as compared with healthy controls, and secondly, whether there are sex differences in asymmetry that interact with diagnosis. 2. Methods 2.1. Subjects and image acquisition parameters MR images were obtained for 26 patients with DSM-IV defined schizophrenia (18 males, 8 females, mean ages33.3 ("7.11), mean age of onsets24.7 years ("6.45), mean duration of illnesss7.6 years ("2.4)) and 24 controls (14 males, 10 females, mean ages30.6"6.43). Subjects were recruited from an ongoing longitudinal study of patients with schizophrenia (see DeLisi et al., 1994, 1997). All individuals gave written informed consent before participation in the MRI study. Handedness was assessed using a 23-item scale (Annett, 1985), and only right-handed subjects were included. Images were acquired using a 1.5 T SIGNA whole body MR imaging system (General Electric, Milwaukee, WI, USA). One hundred and twentyfour coronal T1-weighted images were obtained using a three-dimensional spoiled gradient echo (SPGR) pulse sequence (TR of 24 ms, TE of 5 ms and flip angle of 408). The field of view (FOV) of the images is 24 cm, and each image refers to a 1.5-mm-thick contiguous section of tissue. 2.2. Image analysis Data were analysed on a SPARC Ultra 10 workstation (Sun Microsystems, Mountain View, CA) using MATLAB 5.3 (The Math Works, Natick, MA), KUL (Leuven, Belgium; Maes et al., 1999), in-house software LowD (MARIARC, University of Liverpool, UK; Barrick, 2003), and SPM99 (Wellcome Department of Cognitive Neurology, Institute of Neurology, London; Friston et
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
Fig. 1. Summary of image analysis steps.
al., 1995) software. Images were displayed using mri3dX (www-users.aston.ac.uky;singhkdy mri3dX). The image-processing steps are described below and are summarised in Fig. 1. Firstly, images were segmented into probability maps of grey and white matter in SPM99. This procedure incorporates affine normalisation to the standard template for registration with the grey and white matter priors and intensity inhomogeneity (bias) correction, but the segmented images were output in the original acquired space (Ashburner et al., 2000). The volumes and spatial extents of the hemispheres were calculated by application of an ROI extraction technique called KUL (Maes et al., 1999). In particular, non-rigid deformation of a template image to each study image using the demons algorithm (Thirion, 1998) yielded co-registration parameters that were applied to deform a binary mask of the cerebral hemispheres into the space of the study image. By overlaying the deformed masks on the grey and white matter segmentations, right and left hemisphere grey, white and total matter were extracted. Total volume was calculated by summing the combined probability of grey plus white matter for each voxel. For LowD (Barrick et al., 2001; Barrick, 2003), volumes were extracted for two-dimensional arrays of columns of voxels perpendicular to the midsagittal plane. In particular, the mid-sagittal slice
27
of the template space (242=174 pixels) was transformed to acquired space using an affine transformation, computed as the linear algebraic inverse of the affine transformation of subject to symmetrical template (computed from the standard template). For each hemisphere, the grey and white matter voxel contents of the segmented images were integrated in columns orthogonal to the transformed plane. This plane determines the orientation of column integration and takes account of the differences in head orientation between subjects. The integrated columns were obtained in acquired space and therefore the volume of each subject’s cerebral hemispheres was preserved. Column volumes were output on the mid-sagittal slice in template space where asymmetry measures were calculated by subtraction of left hemisphere data from corresponding right hemisphere data. Negative and positive numbers therefore represent leftward and rightward asymmetry, respectively. Absolute rather than relative asymmetry measures were used firstly, to preserve volumetric information and secondly, because relative measures would introduce increased variance in low tissue concentration areas. In addition, a mask with approximate lobar boundaries was created by drawing onto the sagittal two-dimensional array using a rendered three-dimensional image as a reference such that column voxels could be integrated for the relevant lobes (see Fig. 3). For one-dimensional analysis (Barrick et al., 2001), the superior–inferior dimension of the two-dimensional arrays was integrated to give an anterior–posterior profile of tissue volume in 242 0.781-mm-thick slices along the anterior commissure—posterior commisure line. Smoothing was applied to enable application of the Gaussian field model as a basis for correcting for multiple comparisons. Approximately equivalent smoothness of the column map and slice profile data was achieved using Full Width at Half Maximum (FWHM) isotropic Gaussian kernels of 6 and 3 mm, respectively. 2.3. Statistical analysis Statistical analysis of slice profile and column map difference images was performed in SPM99, with appropriate correction for multiple compari-
28
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
sons using random Gaussian fields (Worsley et al., 1996). Binary masks were constructed by proportional thresholding at 25% and 5% for column and slice volumes, respectively. These values were chosen because they effectively removed the boundary annulus of the column maps and the anterior and posterior limits of the slice profiles, so as to ensure that asymmetry analysis referred only to pixels of high tissue compartment concentration across all individuals. All subjects were entered in a general linear model (GLM), with sex and diagnosis as between-subject factors and asymmetry as the dependent variable. In order to test the effect of brain size on analysis of asymmetry, regression analyses were performed between both two dimensional and one dimensional asymmetry measures and total hemisphere volume. All statistical parametric maps were thresholded at P0.01 (uncorrected) and resulting foci were corrected using a spatial extent threshold of P-0.05 as has been described for voxelwise analyses (Friston et al., 1995). Statistical analysis of hemispheric volumes was conducted (in SPSS) using a repeated measures MANOVA design with between-subject factors of diagnosis (patientycontrol) and sex (maleyfemale) and a within-subject factor of hemisphere (lefty right). In the analysis of lobar volumes, an additional within-subject factor (frontalytemporaly parietalyoccipital) was applied. Analysis of the relationship between lobar asymmetry measures and hemisphere volume was conducted using Pearson’s product-moment correlation coefficient, thresholded at P-0.05. 3. Results 3.1. 2D column maps Average column asymmetry for male and female patients and controls is displayed in Fig. 1. Statistically significant leftward asymmetry in occipital lobe and rightward asymmetry in frontal lobe are clearly visible for the combined group of male and female patients and controls, but no significant group differences or interactions were found.
In addition to the torque, a striking feature of all the average column asymmetry maps displayed in Fig. 2 is a large region of significant leftward asymmetry in the inferior temporal lobe. Inspection of the T1 weighted images for individual subjects revealed that this feature has arisen from a region of high contrast susceptibility artefact on the inferior surface of both right and left temporal lobes. This region, which was not classified as either grey or white matter by automatic segmentation (see Fig. 3), had greater spatial extent on the right than the left inferior temporal lobe, causing a leftward asymmetry of the segmented volumes. Volume differences between right and left lobar volume integrations were calculated and average group data are plotted in Fig. 4. The torque effect was revealed by a significant side–lobe interaction such that the occipital lobe is larger on the left and the frontal lobe is larger on the right (Fs72.5 w46:3x, P-0.001). There was a significant side– lobe–sex interaction, indicating that sex differences in asymmetry (males greater than females) are greatest in the occipital lobe (Fs7.80 w46:3x, Ps 0.008). Post-hoc analysis revealed significant sex differences in asymmetry in occipital, frontal and temporal lobes (F)4.2 w46:3x, P-0.05). There was a non-significant trend for a side–lobe–diagnosis interaction, such that differences in asymmetry (controls more leftwardly asymmetric than patients) tended to be greatest in the occipital lobe (Fs3.3 w46:3x, Ps0.078). The relationship between respective rightward frontal and leftward occipital asymmetries (right– left) was investigated using Pearson’s productmoment correlation coefficient (see Fig. 4). Overall, rightward frontal asymmetry was significantly correlated with leftward occipital asymmetry (rsy0.77, P-0.001). No sex or diagnosis differences were observed in this relationship. 3.2. 1D slice profiles Regions of average asymmetry in slice profiles for patients and controls are displayed in Fig. 5. There was a significant torque effect (significant rightward asymmetry anteriorly and leftward asymmetry posteriorly) for all subjects together.
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
29
Fig. 2. Average column map asymmetry (left column) and standard error (right column) for male and female controls and patients are shown in rows 1–4, respectively. For average asymmetry maps, red–yellow represents increasing rightward asymmetry and blue–white represents increasing leftward asymmetry. The colour scale bars show maximum and minimum volumes in cm3.
There were no significant between-group differences. The region of artefact described above corresponds to a leftward asymmetry extending
from the Talairach y-coordinate y18 to y30, which is anterior to the leftward occipital component of torque.
30
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
Fig. 3. An example of a normalised image in the top left, then total, grey and white matter segmentations in top right, bottom left and bottom right panels, respectively. A high contrast susceptibility artefact is present on the inferior surface of both left and right temporal lobes, which are not classified into a tissue type. The artefact is more marked on the right, causing a highly significant leftward asymmetry in the inferior temporal lobe region on the two-dimensional maps, Fig. 2.
3.3. Hemisphere volumes
4. Discussion
Whole cerebral hemisphere volumes are presented in Table 1. Males had significantly larger cerebral hemisphere volumes than females (Fs 26.1 w46:1x, P-0.001), and patients tended to have smaller cerebral hemisphere volumes than controls (1.5%; Ps0.21). There was a significant main effect of side (Fs24.5 w46:1x, P-0.001) with cerebral hemisphere volume being 0.67% larger on the right than the left. There were no significant correlations between total (leftqright) hemisphere volume and hemispheric, two-dimensional, one-dimensional or lobar asymmetry (P) 0.05).
We have applied new automatic image analysis techniques developed to quantify global cerebral torque in MR images obtained for patients with chronic schizophrenia and controls. Average brain torque was significant in both patient and control groups and did not correlate with brain size. Both frontal and occipital components of torque were significantly greater in males than females (P0.05). There was no significant reduction of torque in patients relative to controls, but 3y26 patients (and no controls) had reversed torque. These results will be discussed in relation to previous investigations of brain torque in schizophrenia (for
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
31
Fig. 4. The top row shows the lobar boundary definitions used for the column maps. Data lying within each region were integrated to give lobar asymmetry volumes for each group (middle row), with male subjects in blue (darkscontrols, lightspatients) and females in red (darkscontrols, lightspatients). The bottom row shows scatter plots and regression lines between frontal and occipital asymmetry for controls (left) and patients (right).
review, see DeLisi et al., 1997) after consideration of the implications of image artefact in this dataset. Observation of the two-dimensional column maps of asymmetry in each of the groups revealed an area of highly significant leftward asymmetry in the inferior temporal lobe that had not been
observed in a previous study of healthy male and female subjects (Barrick et al., 2001). The region of leftward asymmetry corresponded to an area of susceptibility artefact, which caused misclassification of tissue during segmentation, such that it was not classified as brain matter (see Fig. 3). The
32
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
Fig. 5. Average slice profile asymmetry (left column) and standard error (right column) for male and female controls and patients are shown in rows 1–4, respectively. The x-axis on each graph is Talairach co-ordinates (mm), with posterior and anterior limits at the left and right extremes, respectively. For average asymmetry profiles, positive numbers (orange colour) represent rightward asymmetry and negative numbers (purple colour) represent leftward asymmetry.
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35 Table 1 Mean cerebral hemisphere volumes for male and female controls and patients. S.D. shown in parentheses Cerebral hemisphere volume (ml) Left Male controls Male patients Female controls Female patients
514.5 502.7 455.6 434.4
(44.0) (41.5) (37.7) (42.6)
Right
Total
516.6 507.0 458.7 437.6
(45.5) 1031.1 (89.4) (42.0) 1009.7 (83.3) (37.8) 914.3 (75.5) (42.9) 872.0 (85.4)
artefact had greater spatial extent in the right hemisphere, causing the observed leftward asymmetry in two-dimensional column maps and a corresponding region in the slice profile (Talairach y-coordinates y18 to y30). The spatial location of the artefact was such that it did not compromise the visualisation of torque in two-dimensional column maps or one-dimensional slice profiles, and did not affect the analysis of volumes in the occipital, parietal or frontal lobes. The main effects of sex and diagnosis in hemisphere volume are unlikely to be compromised by the artefact since there is no reason to suspect that the magnitude differed between groups. Furthermore, since the observed total cerebral hemisphere asymmetry was rightward, this was also not compromised by the artefactual leftward asymmetry. We have obtained evidence to suggest that brain torque is less in females than males and is reduced, but not absent, in schizophrenia. Bilder et al. (1994, 1999) reported that torque, assessed via measurement of coronal slice volume, was reduced in first episode patients compared with controls, and Sharma et al. (1999) found reduction in occipitoparietal asymmetry in putative ‘obligate carrier’ parents of patients with schizophrenia and the patients themselves using a similar method. Guerguerian and Lewine (1998) assessed torque by combining asymmetry measures from prefrontal and occipito–parietal regions and reported no main effect of diagnosis, but significant sex differences in torque (males greater than females), the magnitude of which was greater in patients with schizophrenia than controls. In our previous study (DeLisi et al., 1997), horizontal–posterior Sylvian fissure asymmetries were found to be significantly correlated in siblings with schizophrenia, consis-
33
tent with familial determination, and a similar trend was observed for postero–occipital width asymmetries. Elucidation of normal asymmetry and deviations in psychosis requires rigorous, standardised methods of analysis, and attention to sex differences in large samples. The new LowD technique enables automatic visualisation and quantification of brain torque, but is not appropriate for the detection of more local asymmetries such as those found in the peri-Sylvian region (e.g. Geschwind and Levitsky, 1968). In future work we will combine asymmetry analysis techniques in order to compare local asymmetries with global brain torque. Although we found strong correlations between the frontal and occipital components of torque, we found no significant correlation between any column map or slice profile asymmetry measures and total cerebral hemisphere volume. Despite a greater magnitude of asymmetry in males relative to females, there was no indication that increased brain size was the main contributor. Sex differences have been reported in the developmental growth rate of the cerebral hemispheres (Kretschmann et al., 1979; Giedd et al., 1996, 1997; Pfluger et al., 1999), such that brain growth is faster in females than males, but reaches a plateau earlier (Kretschmann et al., 1979). This sex difference in brain growth may cause differences in asymmetry and laterality, and also the sex differences observed in clinical characteristics of schizophrenia. Crow (1997) proposed that development of cerebral asymmetry was caused by the speciation event that gave rise to the capacity for language in Homo sapiens, and that asymmetry is anomalous in patients with schizophrenia. It is interesting that in three patients (two males, one female, all righthanded) the sign of the torque was reversed and this would have contributed to the tendency for reduced magnitude of torque in patients overall. Details of specific symptomatology were not available to the present study, but future work may benefit from examining symptom patterns in relation to brain structural anomalies. Reversal of torque has been previously reported in left-handed, but not right-handed control subjects (Bear et al., 1986), but as yet there are no studies of brain torque in mixed-handed subjects, in whom delays
34
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
in the acquisition of words and other cognitive deficits have been demonstrated (Crow et al., 1998). A fuller understanding of the relationship between sex, handedness and cerebral asymmetry, in combination with quantitative symptom rating, may be required to elucidate anomalies of cerebral asymmetry in patients with schizophrenia. Acknowledgments We acknowledge the support of the UK Medical Research Council (grant 噛 G9900348). LED is grateful to the NIMH (R01 MH-44245) and the Stanley Foundation for their support. References Annett, M., 1985. Left, Right, Hand and Brain: The Right Shift Theory. Lawrence Erlbaum Associates, Publishers, London. Ashburner, J., Andersson, J.L., Friston, K.J., 2000. Image registration using a symmetric prior in three dimensions. Human Brain Mapping 9, 212–225. Barrick, T.R., Mackay, C.E., Crow, T.J., Maes, F., Vandermeulen, D., Roberts, N., 2001. Analysis of asymmetry in 2D and 1D projections of left and right cerebral hemisphere grey and white matter. Neuroimage 13 (6), 72. Barrick, T.R. 2003. Automatic image analysis methods for quantification of cerebral asymmetry in standard space: application to analysis of average asymmetry, sexual dimorphism and investigation of asymmetry disturbances in patients with chronic schizophrenia. Ph.D. Thesis, University of Liverpool. Bear, D.M., Schiff, D., Saver, J., Greenberg, M., Freeman, R., 1986. Quantitative analysis of cerebral asymmetry: fronto– occipital correlation, sexual dimorphism and association with handedness. Archives of Neurology 43, 598–603. Bilder, R.M., Wu, H., Bogerts, B., Ashtari, M., Robinson, D., Woerner, M., Lieberman, J.A., Degreef, G., 1999. Cerebral volume asymmetries in schizophrenia and mood disorders: a quantitative magnetic resonance imaging study. International Journal of Psychophysiology 34, 197–205. Bilder, R.M., Wu, H., Bogerts, B., Degreef, G., Ashtari, M., Alvir, J.M.J., Snyder, P.J., Lieberman, J.A., 1994. Absence of regional hemispheric volume asymmetries in first episode schizophrenia. American Journal of Psychiatry 151, 1437–1447. Crow, T.J., 1990. Temporal lobe asymmetries as the key to the etiology of schizophrenia. Schizophrenia Bulletin 16, 433–443. Crow, T.J., 1997. Schizophrenia as a failure of hemispheric dominance for language. Trends in Neuroscience 20, 339–343.
Crow, T.J., Crow, L.R., Done, D.J., Leask, S.J., 1998. Relative hand skill predicts academic ability: global deficits at the point of hemispheric indecision. Neuropsychologia 36, 1275–1282. DeLisi, L.E., Hoff, A.L., Neale, C., Kushner, M., 1994. Asymmetries in the superior temporal lobe in male and female first-episode schizophrenic patients: measures of the planum temporale and superior temporal gyrus by MRI. Schizophrenia Research 12, 19–28. DeLisi, L.E., Sakuma, M., Kushner, M., Finer, D.L., Hoff, A.L., Crow, T.J., 1997. Anomalous cerebral asymmetry and language processing in schizophrenia. Schizophrenia Bulletin 23, 255–271. Falkai, P., Schneider, T., Greve, B., Klieser, E., Bogerts, B., 1995. Reduced frontal and occipital lobe asymmetry on the CT scans of schizophrenic patients: its specificity and clinical significance. Journal of Neural Transmission-General Section 99, 63–77. Friston, K.J., Holmes, A.P., Worsley, K.J., Poline, J.-B., Frith, C.D., Frackowiak, R.S., 1995. Statistical parametric maps in functional imaging: a general approach. Human Brain Mapping 2, 189–210. Geschwind, N., Levitsky, W., 1968. Human brain: left-right asymmetry in temporal speech region. Science 161, 186–187. Giedd, J.N., Castellanos, F.X., Rajapakse, J.C., Vaituzis, A.C., Rapoport, J.L., 1997. Sexual dimorphism of the developing brain. Progress in Neuropsychopharmacology and Biological Psychiatry 21, 1185–1201. Giedd, J.N., Snell, J.W., Lange, N., Rajapakse, J.C., Casey, B.J., Kosuch, P.I., Vaitzis, A.C., Vauss, Y.C., Hamburger, S.D., Kaysen, D., Rapoport, J.L., 1996. Quantitative magnetic resonance imaging of human brain development: ages 4–18. Cerebral Cortex 6, 551–560. Guerguerian, R., Lewine, R.R., 1998. Brain torque and sex differences in schizophrenia. Schizophrenia Research 30, 175–181. Kretschmann, H.F., Schleicher, A., Wingert, F., Zilles, K., Loeblich, H.-J., 1979. Human brain growth in the 19th and 20th century. Journal of the Neurological Sciences 40, 169–188. LeMay, M., 1976. Morphological cerebral asymmetries of modern man, fossil man and non-human primate. Annals of the New York Academy of Sciences 280, 349–369. Luchins, D.J., Meltzer, H.Y., 1983. A blind, controlled study of occipital asymmetry in schizophrenia. Psychiatry Research 10, 87–95. Luchins, D.J., Weinberger, D.R., Wyatt, R.J., 1979. Schizophrenia evidence for a subgroup with reversed cerebral asymmetry. Archives of General Psychiatry 36, 1309–1311. Maes, F., Van Leemput, K., DeLisi, L.E., Vandermeulen D., Suetens, P., 1999. Quantification of cerebral grey and white matter asymmetry from MRI. Proceedings of the 2nd International Conference on Medical Image Computing and Computer-Assisted Intervention. In Taylor C. and Colchester A. (Eds.). MICCAI’99, Lecture Notes in Computer Science, Springer, Cambridge, UK, vol. 1679, pp. 348–357.
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35 Maher, B.A., Manschreck, T.C., Yurgelun-Todd, D.A., Tsuang, M.T., 1998. Hemispheric asymmetry of frontal and temporal grey matter and age of onset in schizophrenia. Biological Psychiatry 44, 413–417. Pfluger, T., Weil, S., Weis, S., Vollmar, C., Heiss, D., Egger, J., Scheck, R., Hahn, K., 1999. Normative volumetric data of the developing hippocampus in children based on magnetic resonance imaging. Epilepsia 40, 414–423. Sharma, T.M., Lancaster, E., Sigmundsson, T., Lewis, S.W., Takei, N., Gurling, H.D.M., Barta, P., Pearlson, G.D., Murray, R.M., 1999. Lack of normal pattern of cerebral asymmetry in familial schizophrenic patients and their relatives—
35
the Maudsley family study. Schizophrenia Research 40, 111–120. Thirion, J.P., 1998. Image matching as a diffusion process: an analogy with Maxwell’s demons. Medical Image Analysis 2, 243–260. Weinberger, D.R., Luchins, D.J., Morihisa, J., Wyatt, R.J., 1982. Asymmetrical volumes of the right and left frontal and occipital regions of the human brain. Annals of Neurology 11, 97–100. Worsley, K.J., Marratt, S., Neelin, P., Vandal, A.C., 1996. A unified statistical approach for determining significant signals in images or cerebral activation. Human Brain Mapping 4, 58–73.
Application of a new image analysis technique to study brain asymmetry in schizophrenia Clare E. Mackaya,*, Thomas R. Barrickb, Neil Robertsb, Lynn E. DeLisic, Frederik Maesd, Dirk Vandermeulend, Timothy J. Crowa Prince of Wales International Centre for SANE Research (POWIC), Department of Psychiatry, University of Oxford, Warneford Hospital, Headington, Oxford OX3 7JX, UK b Magnetic Resonance and Image Analysis Research Centre (MARIARC), Pembroke Place, University of Liverpool, Liverpool L69 3BX, UK c Department of Psychiatry, New York University School of Medicine, 650 First Avenue, New York, NY 10016, USA d Department of Electrical Engineering, Medical Image Computing, ESAT-PSI, K.U. Leuven, Belgium
a
Received 13 August 2002; received in revised form 9 May 2003; accepted 29 May 2003
Abstract The hypothesis that normal brain torque (i.e. rightward frontal and leftward occipital asymmetry) is anomalous in schizophrenia (Crow, 1997. Trends in Neuroscience, 20, 339–343) was tested by application of a novel image analysis technique on three-dimensional magnetic resonance images obtained in 26 adult patients with chronic schizophrenia (18 males, 8 females) and 24 controls (14 males, 10 females). Right and left cerebral hemisphere tissue was extracted via non-linear co-registration with a mask image, and maps were computed of inter-hemispheric differences in tissue volume in an array of columns of voxels orthogonal to the mid-plane (2D), and profiles of coronal slice volumes (1D). Furthermore, integration of two-dimensional column maps gave approximate lobar asymmetries, and occipital and frontal asymmetries were combined to give a volumetric measure of brain torque. Significant brain torque was revealed in male and female control and patient groups, and did not correlate with brain size. Frontal and occipital asymmetries were significantly correlated in all groups. Both frontal and occipital components of torque were significantly increased in males than females. Patients tended to have reduced torque, particularly the leftward occipital component. Furthermore, 3y26 patients (but no controls) had reversed torque (leftward frontal and rightward occipital asymmetry). Contrary to Crow’s hypothesis, brain torque was not significantly reduced in patients with schizophrenia relative to controls, although reversal of torque was found in three cases. Future studies with larger sample sizes should consider sexual dimorphism and specific symptoms in relation to asymmetry. 䊚 2003 Elsevier Ireland Ltd. All rights reserved.
Keywords: MRI; Psychosis; Cerebral asymmetry; Sex differences
*Corresponding author. Tel.: q44-1-865-223-779; fax: q44-1-865-244-990. E-mail address: [email protected] (C.E. Mackay). 0925-4927/03/$ - see front matter 䊚 2003 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/S0925-4927(03)00088-X
26
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
1. Introduction A method for assessment of global brain asymmetry is necessary to address the central question of whether psychosis can be considered as a primary anomaly of cerebral lateralisation (Crow, 1990, 1997). The notion of an alteration in the usual brain ‘torque’ (rightward frontal and leftward occipital asymmetry), sometimes referred to as the Yakovlevian torque (LeMay, 1976; Weinberger et al., 1982), has previously been investigated in patients with schizophrenia using width or volume measurements of the hemispheres in specific frontal (anterior) and occipital (posterior) regions. Loss of asymmetries of brain width were reported in some patients with schizophrenia by Luchins et al. (1979), but not seen in a later study by the same author (Luchins and Meltzer, 1983). Reduced width asymmetry relative to controls has been reported in early onset patients, (Crow et al., 1998; Maher et al., 1998), first episode patients (DeLisi et al., 1997) and a large series of chronic patients (Falkai et al., 1995). The first volumetric study of brain torque was by Bilder et al. (1994), who applied planimetry on successive coronal MRI sections in first episode patients and reported that torque was significantly diminished compared with findings in controls. More recently, this group reported similar reductions of lesser magnitude in patients with affective disorders (Bilder et al., 1999). Although not investigated by Bilder et al. (1994), an important aspect of cerebral torque is that a sex difference has been reported, such that males have a greater magnitude of torque than females (Bear et al., 1986; Barrick et al., 2001). Sex differences in width asymmetries have also been observed in patients with schizophrenia (e.g. DeLisi et al., 1994; Falkai et al., 1995). We suggest that the study of torque therefore has the potential to elucidate the neuroanatomical basis of sex differences associated with psychosis. We have developed new techniques for measuring cerebral asymmetry in two-dimensional maps referring to arrays of columns of voxels orthogonal to the interhemispheric plane and one-dimensional profiles of coronal slice volumes oriented along the anterior-posterior axis of three-dimensional MR images (Barrick et al., 2001; Barrick, 2003) and
applied these techniques in the study of 26 adult patients with chronic schizophrenia (18 males, 8 females) and 24 controls (14 males, 10 females). Firstly, we have investigated whether there is reduction or reversal of global asymmetry in patients with schizophrenia as compared with healthy controls, and secondly, whether there are sex differences in asymmetry that interact with diagnosis. 2. Methods 2.1. Subjects and image acquisition parameters MR images were obtained for 26 patients with DSM-IV defined schizophrenia (18 males, 8 females, mean ages33.3 ("7.11), mean age of onsets24.7 years ("6.45), mean duration of illnesss7.6 years ("2.4)) and 24 controls (14 males, 10 females, mean ages30.6"6.43). Subjects were recruited from an ongoing longitudinal study of patients with schizophrenia (see DeLisi et al., 1994, 1997). All individuals gave written informed consent before participation in the MRI study. Handedness was assessed using a 23-item scale (Annett, 1985), and only right-handed subjects were included. Images were acquired using a 1.5 T SIGNA whole body MR imaging system (General Electric, Milwaukee, WI, USA). One hundred and twentyfour coronal T1-weighted images were obtained using a three-dimensional spoiled gradient echo (SPGR) pulse sequence (TR of 24 ms, TE of 5 ms and flip angle of 408). The field of view (FOV) of the images is 24 cm, and each image refers to a 1.5-mm-thick contiguous section of tissue. 2.2. Image analysis Data were analysed on a SPARC Ultra 10 workstation (Sun Microsystems, Mountain View, CA) using MATLAB 5.3 (The Math Works, Natick, MA), KUL (Leuven, Belgium; Maes et al., 1999), in-house software LowD (MARIARC, University of Liverpool, UK; Barrick, 2003), and SPM99 (Wellcome Department of Cognitive Neurology, Institute of Neurology, London; Friston et
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
Fig. 1. Summary of image analysis steps.
al., 1995) software. Images were displayed using mri3dX (www-users.aston.ac.uky;singhkdy mri3dX). The image-processing steps are described below and are summarised in Fig. 1. Firstly, images were segmented into probability maps of grey and white matter in SPM99. This procedure incorporates affine normalisation to the standard template for registration with the grey and white matter priors and intensity inhomogeneity (bias) correction, but the segmented images were output in the original acquired space (Ashburner et al., 2000). The volumes and spatial extents of the hemispheres were calculated by application of an ROI extraction technique called KUL (Maes et al., 1999). In particular, non-rigid deformation of a template image to each study image using the demons algorithm (Thirion, 1998) yielded co-registration parameters that were applied to deform a binary mask of the cerebral hemispheres into the space of the study image. By overlaying the deformed masks on the grey and white matter segmentations, right and left hemisphere grey, white and total matter were extracted. Total volume was calculated by summing the combined probability of grey plus white matter for each voxel. For LowD (Barrick et al., 2001; Barrick, 2003), volumes were extracted for two-dimensional arrays of columns of voxels perpendicular to the midsagittal plane. In particular, the mid-sagittal slice
27
of the template space (242=174 pixels) was transformed to acquired space using an affine transformation, computed as the linear algebraic inverse of the affine transformation of subject to symmetrical template (computed from the standard template). For each hemisphere, the grey and white matter voxel contents of the segmented images were integrated in columns orthogonal to the transformed plane. This plane determines the orientation of column integration and takes account of the differences in head orientation between subjects. The integrated columns were obtained in acquired space and therefore the volume of each subject’s cerebral hemispheres was preserved. Column volumes were output on the mid-sagittal slice in template space where asymmetry measures were calculated by subtraction of left hemisphere data from corresponding right hemisphere data. Negative and positive numbers therefore represent leftward and rightward asymmetry, respectively. Absolute rather than relative asymmetry measures were used firstly, to preserve volumetric information and secondly, because relative measures would introduce increased variance in low tissue concentration areas. In addition, a mask with approximate lobar boundaries was created by drawing onto the sagittal two-dimensional array using a rendered three-dimensional image as a reference such that column voxels could be integrated for the relevant lobes (see Fig. 3). For one-dimensional analysis (Barrick et al., 2001), the superior–inferior dimension of the two-dimensional arrays was integrated to give an anterior–posterior profile of tissue volume in 242 0.781-mm-thick slices along the anterior commissure—posterior commisure line. Smoothing was applied to enable application of the Gaussian field model as a basis for correcting for multiple comparisons. Approximately equivalent smoothness of the column map and slice profile data was achieved using Full Width at Half Maximum (FWHM) isotropic Gaussian kernels of 6 and 3 mm, respectively. 2.3. Statistical analysis Statistical analysis of slice profile and column map difference images was performed in SPM99, with appropriate correction for multiple compari-
28
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
sons using random Gaussian fields (Worsley et al., 1996). Binary masks were constructed by proportional thresholding at 25% and 5% for column and slice volumes, respectively. These values were chosen because they effectively removed the boundary annulus of the column maps and the anterior and posterior limits of the slice profiles, so as to ensure that asymmetry analysis referred only to pixels of high tissue compartment concentration across all individuals. All subjects were entered in a general linear model (GLM), with sex and diagnosis as between-subject factors and asymmetry as the dependent variable. In order to test the effect of brain size on analysis of asymmetry, regression analyses were performed between both two dimensional and one dimensional asymmetry measures and total hemisphere volume. All statistical parametric maps were thresholded at P0.01 (uncorrected) and resulting foci were corrected using a spatial extent threshold of P-0.05 as has been described for voxelwise analyses (Friston et al., 1995). Statistical analysis of hemispheric volumes was conducted (in SPSS) using a repeated measures MANOVA design with between-subject factors of diagnosis (patientycontrol) and sex (maleyfemale) and a within-subject factor of hemisphere (lefty right). In the analysis of lobar volumes, an additional within-subject factor (frontalytemporaly parietalyoccipital) was applied. Analysis of the relationship between lobar asymmetry measures and hemisphere volume was conducted using Pearson’s product-moment correlation coefficient, thresholded at P-0.05. 3. Results 3.1. 2D column maps Average column asymmetry for male and female patients and controls is displayed in Fig. 1. Statistically significant leftward asymmetry in occipital lobe and rightward asymmetry in frontal lobe are clearly visible for the combined group of male and female patients and controls, but no significant group differences or interactions were found.
In addition to the torque, a striking feature of all the average column asymmetry maps displayed in Fig. 2 is a large region of significant leftward asymmetry in the inferior temporal lobe. Inspection of the T1 weighted images for individual subjects revealed that this feature has arisen from a region of high contrast susceptibility artefact on the inferior surface of both right and left temporal lobes. This region, which was not classified as either grey or white matter by automatic segmentation (see Fig. 3), had greater spatial extent on the right than the left inferior temporal lobe, causing a leftward asymmetry of the segmented volumes. Volume differences between right and left lobar volume integrations were calculated and average group data are plotted in Fig. 4. The torque effect was revealed by a significant side–lobe interaction such that the occipital lobe is larger on the left and the frontal lobe is larger on the right (Fs72.5 w46:3x, P-0.001). There was a significant side– lobe–sex interaction, indicating that sex differences in asymmetry (males greater than females) are greatest in the occipital lobe (Fs7.80 w46:3x, Ps 0.008). Post-hoc analysis revealed significant sex differences in asymmetry in occipital, frontal and temporal lobes (F)4.2 w46:3x, P-0.05). There was a non-significant trend for a side–lobe–diagnosis interaction, such that differences in asymmetry (controls more leftwardly asymmetric than patients) tended to be greatest in the occipital lobe (Fs3.3 w46:3x, Ps0.078). The relationship between respective rightward frontal and leftward occipital asymmetries (right– left) was investigated using Pearson’s productmoment correlation coefficient (see Fig. 4). Overall, rightward frontal asymmetry was significantly correlated with leftward occipital asymmetry (rsy0.77, P-0.001). No sex or diagnosis differences were observed in this relationship. 3.2. 1D slice profiles Regions of average asymmetry in slice profiles for patients and controls are displayed in Fig. 5. There was a significant torque effect (significant rightward asymmetry anteriorly and leftward asymmetry posteriorly) for all subjects together.
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
29
Fig. 2. Average column map asymmetry (left column) and standard error (right column) for male and female controls and patients are shown in rows 1–4, respectively. For average asymmetry maps, red–yellow represents increasing rightward asymmetry and blue–white represents increasing leftward asymmetry. The colour scale bars show maximum and minimum volumes in cm3.
There were no significant between-group differences. The region of artefact described above corresponds to a leftward asymmetry extending
from the Talairach y-coordinate y18 to y30, which is anterior to the leftward occipital component of torque.
30
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
Fig. 3. An example of a normalised image in the top left, then total, grey and white matter segmentations in top right, bottom left and bottom right panels, respectively. A high contrast susceptibility artefact is present on the inferior surface of both left and right temporal lobes, which are not classified into a tissue type. The artefact is more marked on the right, causing a highly significant leftward asymmetry in the inferior temporal lobe region on the two-dimensional maps, Fig. 2.
3.3. Hemisphere volumes
4. Discussion
Whole cerebral hemisphere volumes are presented in Table 1. Males had significantly larger cerebral hemisphere volumes than females (Fs 26.1 w46:1x, P-0.001), and patients tended to have smaller cerebral hemisphere volumes than controls (1.5%; Ps0.21). There was a significant main effect of side (Fs24.5 w46:1x, P-0.001) with cerebral hemisphere volume being 0.67% larger on the right than the left. There were no significant correlations between total (leftqright) hemisphere volume and hemispheric, two-dimensional, one-dimensional or lobar asymmetry (P) 0.05).
We have applied new automatic image analysis techniques developed to quantify global cerebral torque in MR images obtained for patients with chronic schizophrenia and controls. Average brain torque was significant in both patient and control groups and did not correlate with brain size. Both frontal and occipital components of torque were significantly greater in males than females (P0.05). There was no significant reduction of torque in patients relative to controls, but 3y26 patients (and no controls) had reversed torque. These results will be discussed in relation to previous investigations of brain torque in schizophrenia (for
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
31
Fig. 4. The top row shows the lobar boundary definitions used for the column maps. Data lying within each region were integrated to give lobar asymmetry volumes for each group (middle row), with male subjects in blue (darkscontrols, lightspatients) and females in red (darkscontrols, lightspatients). The bottom row shows scatter plots and regression lines between frontal and occipital asymmetry for controls (left) and patients (right).
review, see DeLisi et al., 1997) after consideration of the implications of image artefact in this dataset. Observation of the two-dimensional column maps of asymmetry in each of the groups revealed an area of highly significant leftward asymmetry in the inferior temporal lobe that had not been
observed in a previous study of healthy male and female subjects (Barrick et al., 2001). The region of leftward asymmetry corresponded to an area of susceptibility artefact, which caused misclassification of tissue during segmentation, such that it was not classified as brain matter (see Fig. 3). The
32
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
Fig. 5. Average slice profile asymmetry (left column) and standard error (right column) for male and female controls and patients are shown in rows 1–4, respectively. The x-axis on each graph is Talairach co-ordinates (mm), with posterior and anterior limits at the left and right extremes, respectively. For average asymmetry profiles, positive numbers (orange colour) represent rightward asymmetry and negative numbers (purple colour) represent leftward asymmetry.
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35 Table 1 Mean cerebral hemisphere volumes for male and female controls and patients. S.D. shown in parentheses Cerebral hemisphere volume (ml) Left Male controls Male patients Female controls Female patients
514.5 502.7 455.6 434.4
(44.0) (41.5) (37.7) (42.6)
Right
Total
516.6 507.0 458.7 437.6
(45.5) 1031.1 (89.4) (42.0) 1009.7 (83.3) (37.8) 914.3 (75.5) (42.9) 872.0 (85.4)
artefact had greater spatial extent in the right hemisphere, causing the observed leftward asymmetry in two-dimensional column maps and a corresponding region in the slice profile (Talairach y-coordinates y18 to y30). The spatial location of the artefact was such that it did not compromise the visualisation of torque in two-dimensional column maps or one-dimensional slice profiles, and did not affect the analysis of volumes in the occipital, parietal or frontal lobes. The main effects of sex and diagnosis in hemisphere volume are unlikely to be compromised by the artefact since there is no reason to suspect that the magnitude differed between groups. Furthermore, since the observed total cerebral hemisphere asymmetry was rightward, this was also not compromised by the artefactual leftward asymmetry. We have obtained evidence to suggest that brain torque is less in females than males and is reduced, but not absent, in schizophrenia. Bilder et al. (1994, 1999) reported that torque, assessed via measurement of coronal slice volume, was reduced in first episode patients compared with controls, and Sharma et al. (1999) found reduction in occipitoparietal asymmetry in putative ‘obligate carrier’ parents of patients with schizophrenia and the patients themselves using a similar method. Guerguerian and Lewine (1998) assessed torque by combining asymmetry measures from prefrontal and occipito–parietal regions and reported no main effect of diagnosis, but significant sex differences in torque (males greater than females), the magnitude of which was greater in patients with schizophrenia than controls. In our previous study (DeLisi et al., 1997), horizontal–posterior Sylvian fissure asymmetries were found to be significantly correlated in siblings with schizophrenia, consis-
33
tent with familial determination, and a similar trend was observed for postero–occipital width asymmetries. Elucidation of normal asymmetry and deviations in psychosis requires rigorous, standardised methods of analysis, and attention to sex differences in large samples. The new LowD technique enables automatic visualisation and quantification of brain torque, but is not appropriate for the detection of more local asymmetries such as those found in the peri-Sylvian region (e.g. Geschwind and Levitsky, 1968). In future work we will combine asymmetry analysis techniques in order to compare local asymmetries with global brain torque. Although we found strong correlations between the frontal and occipital components of torque, we found no significant correlation between any column map or slice profile asymmetry measures and total cerebral hemisphere volume. Despite a greater magnitude of asymmetry in males relative to females, there was no indication that increased brain size was the main contributor. Sex differences have been reported in the developmental growth rate of the cerebral hemispheres (Kretschmann et al., 1979; Giedd et al., 1996, 1997; Pfluger et al., 1999), such that brain growth is faster in females than males, but reaches a plateau earlier (Kretschmann et al., 1979). This sex difference in brain growth may cause differences in asymmetry and laterality, and also the sex differences observed in clinical characteristics of schizophrenia. Crow (1997) proposed that development of cerebral asymmetry was caused by the speciation event that gave rise to the capacity for language in Homo sapiens, and that asymmetry is anomalous in patients with schizophrenia. It is interesting that in three patients (two males, one female, all righthanded) the sign of the torque was reversed and this would have contributed to the tendency for reduced magnitude of torque in patients overall. Details of specific symptomatology were not available to the present study, but future work may benefit from examining symptom patterns in relation to brain structural anomalies. Reversal of torque has been previously reported in left-handed, but not right-handed control subjects (Bear et al., 1986), but as yet there are no studies of brain torque in mixed-handed subjects, in whom delays
34
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35
in the acquisition of words and other cognitive deficits have been demonstrated (Crow et al., 1998). A fuller understanding of the relationship between sex, handedness and cerebral asymmetry, in combination with quantitative symptom rating, may be required to elucidate anomalies of cerebral asymmetry in patients with schizophrenia. Acknowledgments We acknowledge the support of the UK Medical Research Council (grant 噛 G9900348). LED is grateful to the NIMH (R01 MH-44245) and the Stanley Foundation for their support. References Annett, M., 1985. Left, Right, Hand and Brain: The Right Shift Theory. Lawrence Erlbaum Associates, Publishers, London. Ashburner, J., Andersson, J.L., Friston, K.J., 2000. Image registration using a symmetric prior in three dimensions. Human Brain Mapping 9, 212–225. Barrick, T.R., Mackay, C.E., Crow, T.J., Maes, F., Vandermeulen, D., Roberts, N., 2001. Analysis of asymmetry in 2D and 1D projections of left and right cerebral hemisphere grey and white matter. Neuroimage 13 (6), 72. Barrick, T.R. 2003. Automatic image analysis methods for quantification of cerebral asymmetry in standard space: application to analysis of average asymmetry, sexual dimorphism and investigation of asymmetry disturbances in patients with chronic schizophrenia. Ph.D. Thesis, University of Liverpool. Bear, D.M., Schiff, D., Saver, J., Greenberg, M., Freeman, R., 1986. Quantitative analysis of cerebral asymmetry: fronto– occipital correlation, sexual dimorphism and association with handedness. Archives of Neurology 43, 598–603. Bilder, R.M., Wu, H., Bogerts, B., Ashtari, M., Robinson, D., Woerner, M., Lieberman, J.A., Degreef, G., 1999. Cerebral volume asymmetries in schizophrenia and mood disorders: a quantitative magnetic resonance imaging study. International Journal of Psychophysiology 34, 197–205. Bilder, R.M., Wu, H., Bogerts, B., Degreef, G., Ashtari, M., Alvir, J.M.J., Snyder, P.J., Lieberman, J.A., 1994. Absence of regional hemispheric volume asymmetries in first episode schizophrenia. American Journal of Psychiatry 151, 1437–1447. Crow, T.J., 1990. Temporal lobe asymmetries as the key to the etiology of schizophrenia. Schizophrenia Bulletin 16, 433–443. Crow, T.J., 1997. Schizophrenia as a failure of hemispheric dominance for language. Trends in Neuroscience 20, 339–343.
Crow, T.J., Crow, L.R., Done, D.J., Leask, S.J., 1998. Relative hand skill predicts academic ability: global deficits at the point of hemispheric indecision. Neuropsychologia 36, 1275–1282. DeLisi, L.E., Hoff, A.L., Neale, C., Kushner, M., 1994. Asymmetries in the superior temporal lobe in male and female first-episode schizophrenic patients: measures of the planum temporale and superior temporal gyrus by MRI. Schizophrenia Research 12, 19–28. DeLisi, L.E., Sakuma, M., Kushner, M., Finer, D.L., Hoff, A.L., Crow, T.J., 1997. Anomalous cerebral asymmetry and language processing in schizophrenia. Schizophrenia Bulletin 23, 255–271. Falkai, P., Schneider, T., Greve, B., Klieser, E., Bogerts, B., 1995. Reduced frontal and occipital lobe asymmetry on the CT scans of schizophrenic patients: its specificity and clinical significance. Journal of Neural Transmission-General Section 99, 63–77. Friston, K.J., Holmes, A.P., Worsley, K.J., Poline, J.-B., Frith, C.D., Frackowiak, R.S., 1995. Statistical parametric maps in functional imaging: a general approach. Human Brain Mapping 2, 189–210. Geschwind, N., Levitsky, W., 1968. Human brain: left-right asymmetry in temporal speech region. Science 161, 186–187. Giedd, J.N., Castellanos, F.X., Rajapakse, J.C., Vaituzis, A.C., Rapoport, J.L., 1997. Sexual dimorphism of the developing brain. Progress in Neuropsychopharmacology and Biological Psychiatry 21, 1185–1201. Giedd, J.N., Snell, J.W., Lange, N., Rajapakse, J.C., Casey, B.J., Kosuch, P.I., Vaitzis, A.C., Vauss, Y.C., Hamburger, S.D., Kaysen, D., Rapoport, J.L., 1996. Quantitative magnetic resonance imaging of human brain development: ages 4–18. Cerebral Cortex 6, 551–560. Guerguerian, R., Lewine, R.R., 1998. Brain torque and sex differences in schizophrenia. Schizophrenia Research 30, 175–181. Kretschmann, H.F., Schleicher, A., Wingert, F., Zilles, K., Loeblich, H.-J., 1979. Human brain growth in the 19th and 20th century. Journal of the Neurological Sciences 40, 169–188. LeMay, M., 1976. Morphological cerebral asymmetries of modern man, fossil man and non-human primate. Annals of the New York Academy of Sciences 280, 349–369. Luchins, D.J., Meltzer, H.Y., 1983. A blind, controlled study of occipital asymmetry in schizophrenia. Psychiatry Research 10, 87–95. Luchins, D.J., Weinberger, D.R., Wyatt, R.J., 1979. Schizophrenia evidence for a subgroup with reversed cerebral asymmetry. Archives of General Psychiatry 36, 1309–1311. Maes, F., Van Leemput, K., DeLisi, L.E., Vandermeulen D., Suetens, P., 1999. Quantification of cerebral grey and white matter asymmetry from MRI. Proceedings of the 2nd International Conference on Medical Image Computing and Computer-Assisted Intervention. In Taylor C. and Colchester A. (Eds.). MICCAI’99, Lecture Notes in Computer Science, Springer, Cambridge, UK, vol. 1679, pp. 348–357.
C.E. Mackay et al. / Psychiatry Research: Neuroimaging 124 (2003) 25–35 Maher, B.A., Manschreck, T.C., Yurgelun-Todd, D.A., Tsuang, M.T., 1998. Hemispheric asymmetry of frontal and temporal grey matter and age of onset in schizophrenia. Biological Psychiatry 44, 413–417. Pfluger, T., Weil, S., Weis, S., Vollmar, C., Heiss, D., Egger, J., Scheck, R., Hahn, K., 1999. Normative volumetric data of the developing hippocampus in children based on magnetic resonance imaging. Epilepsia 40, 414–423. Sharma, T.M., Lancaster, E., Sigmundsson, T., Lewis, S.W., Takei, N., Gurling, H.D.M., Barta, P., Pearlson, G.D., Murray, R.M., 1999. Lack of normal pattern of cerebral asymmetry in familial schizophrenic patients and their relatives—
35
the Maudsley family study. Schizophrenia Research 40, 111–120. Thirion, J.P., 1998. Image matching as a diffusion process: an analogy with Maxwell’s demons. Medical Image Analysis 2, 243–260. Weinberger, D.R., Luchins, D.J., Morihisa, J., Wyatt, R.J., 1982. Asymmetrical volumes of the right and left frontal and occipital regions of the human brain. Annals of Neurology 11, 97–100. Worsley, K.J., Marratt, S., Neelin, P., Vandal, A.C., 1996. A unified statistical approach for determining significant signals in images or cerebral activation. Human Brain Mapping 4, 58–73.