1
AUTOMATIC CLUSTER ANALYSIS OF CORPUS CALLOSUM SUBDIVISIONS IN SCHIZOPHRENIA: A DIFFUSION TENSOR IMAGING STUDY Background: The corpus callosum is a major pathway connecting the two hemispheres and has an important role in relaying sensory, cognitive, and motor information throughout the brain. While MRI studies have shown mixed results, Diffusion Tensor Imaging (DTI) has proven to be a much more sensitive tool for examining integrity of white matter fiber bundles. Although seven DTI studies have shown corpus callosum abnormalities, none have examined separately anatomical subdivisions of the corpus using high-resolution 3T images combined with an automatic clustering method. Methods: 12 chronic schizophrenic men, mean age 44, and 12 normal controls, mean age 40, were scanned on a 3T magnet using a high resolution DTI protocol. DT Fiber Tractography was performed on the whole brain for each subject, and then using an automatic procedure, the tracts were grouped into clusters, according to their shapes and their spatial locations. Five fiber clusters were defined for the corpus callosum. Next, each voxel of the midsagittal slice was color-coded according to the corpus callosum clusters and fractional anisotropy, a measure of the tract integrity, was computed. Results: Reduced FA in schizophrenics as compared to normal controls in the second cluster (p = 0.051). Reduced volume in schizophrenics’ third cluster (p = 0.048). Conclusion: Our findings suggest possible abnormalities in interhemispheric connectivity between posterior parts of the frontal lobes in chronic schizophrenia. Further studies are needed to investigate the clinical role of these abnormalities. SUBJECTS: 12 Schizophrenic Patients (mean age = 44.25) and 12 Normal Controls (mean age = 40.33) Table 1 AUTOMATIC SEGMENTATION: After scans were acquired, DTI Fiber Tractography (Figure 4) was performed on the whole brain for each subject, and then using an automatic procedure, the tracts were grouped into clusters, according to a pairwise similarity function that takes into account the tracts’ shapes and spatial locations (O’Donnell, 2006). Figure 5 Five fiber clusters were defined for the Corpus Callosum. Figure 6 DTI ACQUSITION: • 3 Tesla GE system (General Electric Medical Systems, Milwaukee, WI) • Echo Planar Imaging (EPI) DTI Tensor sequence. • Acquired 51 directions with b=700, 8 baseline scans with b=0 and a spatial resolution of 1.7 x 1.7 x 1.7 ROIs: •Each voxel of the midsagittal slice was color-coded according to the cluster it belonged to. (Figure 7) •Mean FA, a measure of the fraction of magnitude of the tensor associated with anisotropic diffusion (Figure 8), for each segment was calculated, and compared between groups. (Table 2) •Volumes of the midsagittal segments were also measured and compared between Alexander AL, Tsurda JS, Parker DL (1997): Elimination of eddy current artifacts in diffusion-weighted echo-planar images: the use of bipolar gradients. Magn Reson Med, 38:1016-21. Crow TJ (1998): Schizophrenia as a transcallosal misconnection syndrome. Schizophr Res 30:111-4. Foong J, Maier M, Clark CA, Barker GJ, Miller DH, Ron MA (2000): Neuropathological abnormalities of the corpus callosum in schizophrenia: a diffusion tensor imaging study. J Neurol Neurosurg Psychiatry 68:242-4. Heid O, Deimling M, Huk W (1995): Ultra-rapid gradient echo imaging. Magn Reson Med 33:143-9. Lent R, Schmidt SL (1993): The ontogenesis of the forebrain commissures and the determination of brain asymmetries. Prog Neurobiol 40:249-76. Mori S, Crain BJ, Chacko VP, van Zijl PC (1999): Three-dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann Neurol 45:265-9. O'Donnell LJ, Kubicki M, Shenton ME, Dreusicke MH, Grimson WE, Westin CF (2006): A method for clustering white matter fiber tracts. AJNR Am J Neuroradiol 27:1032-6. O'Shea RP, Corballis PM (2003): Binocular rivalry in split-brain observers. J Vis 3:610-5. Schmahmann JD, Pandya, DN (2006): Fiber Pathways of the Brain. New York: Oxford University Press. Witelson SF (1989): Hand sex differences in the isthmus and genu of the human corpus callosum. Brain 112:799-835. Woodruff PW, McManus IC, David AS (1995): Meta-analysis of corpus callosum size in schizophrenia. J Neurol Neurosurg Psychiatry 58:457-61. •We demonstrate the method of anatomical division of the Corpus Callosum. • Our findings suggest possible abnormalities in interhemispheric connectivity between posterior parts of the frontal lobes in chronic schizophrenia. •Further studies are needed to investigate the clinical role of these abnormalities. •The Corpus Callosum is a thick band of white matter fibers connecting cortical areas of both hemispheres. (Lent 1993). •Abnormalities in the Corpus Callosum can lead to disordered transfer between the two hemispheres(O’Shea, 2003) and may explain some symptoms and abnormalities found in schizophrenia.(Crow, 1998) •Animal studies provide evidence of five anatomical subdivisions of the Corpus Callosum. (Figure 1) •MRI studies have found a reduction in area of the Corpus Callosum (Woodruff, 1995). Traditionally, Corpus Callosum has been subdivided using geometric rules (Witelson, 1989) • Diffusion Tensor Imaging (DTI), see Figure 2 and Figure 3, has proven to be a much more sensitive tool for examining integrity of white matter fiber bundles and for examining the possible functional significance of the structural abnormalities.(Foong, 2000) •Although seven DTI studies have shown Corpus Callosum abnormalities, none have examined separately anatomical subdivisions of the corpus using high-resolution 3T images combined with an automatic clustering method. ABSTRACT BACKGROUND MATERIALS AND METHODS RESULTS CONCLUSIONS REFERENCES Kate Smith(1,3), Marek Kubicki(1,3), Lauren O’Donnell(2), C.F. Westin(2), Margret Niznikiewicz(3), Robert McCarley(3), Martha Shenton(1,3). Table 1 : Subjects Figure 2 DTI offers a way to examine the amount of diffusivity of water inside the brain. Kleenex Newspaper Anisotropic Diffusion Figure 3 When water molecules aren’t restricted in a specific direction like in the Kleenex or in the grey matter the diffusivity is isotropic. When there is restriction on the molecule like in the newspaper fibers or in myelinated sheathed fiber tracts in the brain, the diffusion is anisotropic. Figure 1 Five subdivisions of the corpus callosum and their pathways; Shmahmann and Pandya, Oxford Press, 2006 (Mori et al., Ann Neurol 1999) Figure 4 The average direction of the diffusivity of the water is obtained for each voxel. Tracts are then formed by connecting the voxels with the same direction. (Courtesy, Hae-Jeong Park) Table 2 : FA Table 3 : Volumes of Subdivisions Figure 7 Mid-Sagittal ROI created from the original automatically clustered tracts. •Review of our data showed that it was a non-normal distribution so we used a non-parametric statistical method. •A two sample Mann-Whitney U test was used •The FA in the CC2 cluster came up as significantly different between schizophrenics and normals (p = 0.051) •The volume of the CC3 cluster came up as significantly different (p = 0.031) 94 94 Schizophrenics Controls p-value A ge 44.25 (8.13) 40.33 (9.64) 0.294 ParentalSES 2.45 (1.03) 2.17 (1.19) 0.543 Sex (% Male) 100% 100% - M ean (Standard Deviation) Figure 5 Automatically clustered whole brain, the Corpus Callosum is highlighted in red. 1. Psychiatry Neuroimaging Laboratory, Department of Psychiatry, Brigham and Women's Hospital, Harvard Medical School 2. MIT Computer Science and AI Lab (CSAIL), and Harvard-MIT Division of Health Sciences and Technology Laboratory of Mathematics in Imaging, MRI Division, Department of Radiology, Brigham and Women's Hospital, Harvard Medical School Clinical 3. Neuroscience Division, Laboratory of Neuroscience, Department of Psychiatry VA Boston Healthcare System, Brockton, MA and Harvard Medical School Figure 6 5 automatically clustered subdivisions of the corpus, the significantly different cluster, CC2, is highlighted in red. Figure 8 FA is a way to quantify the magnitude of diffusion inside the white matter fiber tracts of the brain. (Kuroki 2006) (Mamata et al., 2002) (Courtesy, Kindlemann) (Kindlemann)

AUTOMATIC CLUSTER ANALYSIS OF CORPUS CALLOSUM SUBDIVISIONS IN SCHIZOPHRENIA: A DIFFUSION TENSOR IMAGING STUDY Background: The corpus callosum is a major

Embed Size (px)

Citation preview

Page 1: AUTOMATIC CLUSTER ANALYSIS OF CORPUS CALLOSUM SUBDIVISIONS IN SCHIZOPHRENIA: A DIFFUSION TENSOR IMAGING STUDY Background: The corpus callosum is a major

AUTOMATIC CLUSTER ANALYSIS OF CORPUS CALLOSUM SUBDIVISIONS IN SCHIZOPHRENIA: A DIFFUSION TENSOR IMAGING STUDY

AUTOMATIC CLUSTER ANALYSIS OF CORPUS CALLOSUM SUBDIVISIONS IN SCHIZOPHRENIA: A DIFFUSION TENSOR IMAGING STUDY

Background: The corpus callosum is a major pathway connecting the two hemispheres and has an important role in relaying sensory, cognitive, and motor information throughout the brain. While MRI studies have shown mixed results, Diffusion Tensor Imaging (DTI) has proven to be a much more sensitive tool for examining integrity of white matter fiber bundles. Although seven DTI studies have shown corpus callosum abnormalities, none have examined separately anatomical subdivisions of the corpus using high-resolution 3T images combined with an automatic clustering method.

Methods: 12 chronic schizophrenic men, mean age 44, and 12 normal controls, mean age 40, were scanned on a 3T magnet using a high resolution DTI protocol. DT Fiber Tractography was performed on the whole brain for each subject, and then using an automatic procedure, the tracts were grouped into clusters, according to their shapes and their spatial locations. Five fiber clusters were defined for the corpus callosum. Next, each voxel of the midsagittal slice was color-coded according to the corpus callosum clusters and fractional anisotropy, a measure of the tract integrity, was computed.

Results: Reduced FA in schizophrenics as compared to normal controls in the second cluster (p = 0.051). Reduced volume in schizophrenics’ third cluster (p = 0.048).

Conclusion: Our findings suggest possible abnormalities in interhemispheric connectivity between posterior parts of the frontal lobes in chronic schizophrenia. Further studies are needed to investigate the clinical role of these abnormalities.

SUBJECTS:

12 Schizophrenic Patients (mean age = 44.25) and 12 Normal Controls (mean age = 40.33) Table 1

AUTOMATIC SEGMENTATION:

• After scans were acquired, DTI Fiber Tractography (Figure 4) was performed on the whole brain for each

subject, and then using an automatic procedure, the tracts were grouped into clusters, according to a pairwise

similarity function that takes into account the tracts’ shapes and spatial locations (O’Donnell, 2006). Figure 5

• Five fiber clusters were defined for the Corpus Callosum. Figure 6

DTI ACQUSITION:

• 3 Tesla GE system (General Electric Medical Systems, Milwaukee, WI)

• Echo Planar Imaging (EPI) DTI Tensor sequence.

• Acquired 51 directions with b=700, 8 baseline scans with b=0 and a spatial resolution of 1.7 x 1.7 x 1.7

ROIs:

•Each voxel of the midsagittal slice was color-coded according to the cluster it belonged to. (Figure 7)

•Mean FA, a measure of the fraction of magnitude of the tensor associated with anisotropic diffusion (Figure 8), for each segment was calculated, and compared between groups. (Table 2)

•Volumes of the midsagittal segments were also measured and compared between groups. (Table 3)

Alexander AL, Tsurda JS, Parker DL (1997): Elimination of eddy current artifacts in diffusion-weighted echo-planar images: the use of bipolar gradients. Magn Reson Med, 38:1016-21.

Crow TJ (1998): Schizophrenia as a transcallosal misconnection syndrome. Schizophr Res 30:111-4. Foong J, Maier M, Clark CA, Barker GJ, Miller DH, Ron MA (2000): Neuropathological abnormalities of the corpus callosum in schizophrenia: a diffusion tensor imaging study. J Neurol Neurosurg Psychiatry 68:242-4.

Heid O, Deimling M, Huk W (1995): Ultra-rapid gradient echo imaging. Magn Reson Med 33:143-9. Lent R, Schmidt SL (1993): The ontogenesis of the forebrain commissures and the determination of brain asymmetries. Prog Neurobiol 40:249-76.

Mori S, Crain BJ, Chacko VP, van Zijl PC (1999): Three-dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann Neurol 45:265-9.

O'Donnell LJ, Kubicki M, Shenton ME, Dreusicke MH, Grimson WE, Westin CF (2006): A method for clustering white matter fiber tracts. AJNR Am J Neuroradiol 27:1032-6. O'Shea RP, Corballis PM (2003): Binocular rivalry in split-brain observers. J Vis 3:610-5. Schmahmann JD, Pandya, DN (2006): Fiber Pathways of the Brain. New York: Oxford University Press.

Witelson SF (1989): Hand sex differences in the isthmus and genu of the human corpus callosum. Brain 112:799-835.

Woodruff PW, McManus IC, David AS (1995): Meta-analysis of corpus callosum size in schizophrenia. J Neurol Neurosurg Psychiatry 58:457-61.

•We demonstrate the method of anatomical division of the Corpus Callosum.

• Our findings suggest possible abnormalities in interhemispheric connectivity between posterior parts of the frontal lobes in chronic schizophrenia.

•Further studies are needed to investigate the clinical role of these abnormalities.

•The Corpus Callosum is a thick band of white matter fibers connecting cortical areas of both hemispheres. (Lent 1993).

•Abnormalities in the Corpus Callosum can lead to disordered transfer between the two hemispheres(O’Shea, 2003) and may explain some symptoms and abnormalities found in schizophrenia.(Crow, 1998)

•Animal studies provide evidence of five anatomical subdivisions of the Corpus Callosum. (Figure 1)

•MRI studies have found a reduction in area of the Corpus Callosum (Woodruff, 1995). Traditionally, Corpus Callosum has been subdivided using geometric rules (Witelson, 1989)

• Diffusion Tensor Imaging (DTI), see Figure 2 and Figure 3, has proven to be a much more sensitive tool for examining integrity of white matter fiber bundles and for examining the possible functional significance of the structural abnormalities.(Foong, 2000)

•Although seven DTI studies have shown Corpus Callosum abnormalities, none have examined separately anatomical subdivisions of the corpus using high-resolution 3T images combined with an automatic clustering method.

ABSTRACT

BACKGROUND

MATERIALS AND METHODS

RESULTS

CONCLUSIONS

REFERENCES

Kate Smith(1,3), Marek Kubicki(1,3), Lauren O’Donnell(2), C.F. Westin(2), Margret Niznikiewicz(3), Robert McCarley(3), Martha Shenton(1,3).

Table 1 : Subjects

Figure 2

DTI offers a way to examine the amount of diffusivity of water inside the brain.

Kleenex Newspaper

Anisotropic Diffusion

Figure 3

When water molecules aren’t restricted in a specific direction like in the Kleenex or in the grey matter the diffusivity is isotropic. When there is restriction on the molecule like in the newspaper fibers or in myelinated sheathed fiber tracts in the brain, the diffusion is anisotropic.

Figure 1

Five subdivisions of the corpus callosum and their pathways; Shmahmann and Pandya, Oxford Press, 2006

(Mori et al., Ann Neurol 1999)

Figure 4 The average direction of the diffusivity of the water is obtained for each voxel. Tracts are then formed by connecting the voxels with the same direction. (Courtesy, Hae-Jeong Park)

Table 2 : FA Table 3 : Volumes of Subdivisions

Figure 7

Mid-Sagittal ROI created from the original automatically clustered tracts.

•Review of our data showed that it was a non-normal distribution so we used a non-parametric statistical method.

•A two sample Mann-Whitney U test was used

•The FA in the CC2 cluster came up as significantly different between schizophrenics and normals (p = 0.051)

•The volume of the CC3 cluster came up as significantly different (p = 0.031)

9494

Schizophrenics Controlsp-value

Age 44.25 (8.13) 40.33 (9.64) 0.294Parental SES 2.45 (1.03) 2.17 (1.19) 0.543Sex (% Male) 100% 100% -

Mean (Standard Deviation)

Figure 5Automatically clustered whole brain, the Corpus Callosum is highlighted in red.

1. Psychiatry Neuroimaging Laboratory, Department of Psychiatry, Brigham and Women's Hospital, Harvard Medical School

2. MIT Computer Science and AI Lab (CSAIL), and Harvard-MIT Division of Health Sciences and Technology Laboratory of Mathematics in Imaging, MRI Division, Department of Radiology, Brigham and Women's Hospital, Harvard Medical School Clinical

3. Neuroscience Division, Laboratory of Neuroscience, Department of Psychiatry VA Boston Healthcare System, Brockton, MA and Harvard Medical School

Figure 6

5 automatically clustered subdivisions of the corpus, the significantly different cluster, CC2, is highlighted in red.

Figure 8FA is a way to quantify the magnitude of diffusion inside the white matter fiber tracts of the brain.

(Kuroki 2006)

(Mamata et al., 2002)

(Courtesy, Kindlemann)

(Kindlemann)