- Email: [email protected]
A New Semi-quantitative Approach for Analysing 3T Diffusion Tensor Imaging of Optic Fibres and Its Clinical Evaluation in Glaucoma
Technical Report
A New Semi-quantitative Approach for Analysing 3T Diffusion Tensor Imaging of Optic Fibres and Its Clinical Evaluation in Glaucoma Tobias Engelhorn, MD, Sultan Haider, MSc, Georg Michelson, MD, Arnd Doerfler, MD Rationale and Objectives: Diffusion tensor imaging can depict rarefaction of the optical fibers. Manual segmentation is time consuming. The purposes of the study were (1) to present a new semiquantitative segmentation approach for analyzing 3-T diffusion tensor imaging of optical fibers and (2) to clinically test the new approach by comparing optic fiber rarefaction in patients with glaucoma to that in agematched, healthy controls. Materials and Methods: To perform semiautomated and quantitative segmentation of the optical radiation, a Mathcad-based software program was developed. The results were compared to the manual evaluation of the images performed by two experienced neuroradiologists. The eyes of 42 subjects (22 patients with glaucoma and 20 controls) aged 37 to 86 years were assessed in full ophthalmologic examinations. Magnetic resonance imaging was performed using a 3-T high-field scanner. Results: The evaluation using the new approach matched 94% with manually acquired rarefaction of the optical radiation; Cronbach’s a was >0.81 for calculation of the manually and semiautomatically derived volumes. Conclusion: The new approach seems to be robust and is clearly faster compared to the more tedious manual segmentation. Using diffusion tensor imaging at 3 T, it could be shown that there was increasing atrophy of the optical radiation (fourth neuron) with increasing age in patients with glaucoma. Compared to age-matched, healthy patients, more pronounced atrophy of the fourth neuron was found in patients with glaucoma. Key Words: 3T; DTI; MRI; glaucoma. ªAUR, 2010
W
hether glaucoma-related rarefaction of the optical radiation exists is still dubious, because there are no experimental or clinical studies dealing with this topic. Loss of ganglion cells in the retina and their axons representing the third neuron and atrophy of the fourth neuron have been described in diseases such as traumatic injury and glaucoma (1–4). The axons of various retinal ganglion cell subtypes, differing in specific morphology and function, exit the eyeball and finally converge to anatomically distinguishable layers of the lateral geniculate nucleus (LGN) (2), where a loss of neural cells has also been described for glaucoma (3).
Acad Radiol 2010; 17:1313–1316 From the Departments of Neuroradiology (T.E., A.D.) and Ophthalmology (G.M.), University Erlangen-Nuremberg, Schwabachanlage 6, D-91054 Erlangen, Germany; and Siemens AG Healthcare Sector, Erlangen, Germany (S.H.). Drs Engelhorn and Haider contributed equally to this work and thus share first authorship. Received March 24, 2010; accepted April 27, 2010. Address correspondence to: T.E. e-mail: [email protected] ªAUR, 2010 doi:10.1016/j.acra.2010.04.017
Diffusion tensor imaging (DTI) can depict the optic nerve (third neuron) and optical radiation (fourth neuron) (5–7). However, manual analysis is time consuming. We used DTI with a newly developed semiautomatic segmentation approach to assess rarefaction of the fourth neuron in 20 control subjects aged 45 to 83 years and in 22 severely ill patients with glaucoma aged 37 to 86 years.
MATERIALS AND METHODS Eyes were assessed in full ophthalmologic examinations. Available additional examinations were referred to for evaluation, including Heidelberg Retinal Tomography (Heidelberg Engineering, Heidelberg, Germany), automated perimetry (Octopus 101 dG2; Interzeag, Schlieren, Switzerland), spatial-temporal contrast sensitivity (frequency doubling test; Carl Zeiss Meditec AG, Jena, Germany), nonmydriatic fundus images (nonmyd-alpha 45; Kowa Optimed, Inc, Torrance, CA), and measurement of intraocular pressure. Magnetic resonance imaging was performed using a 3-T high-field scanner (Magnetom Tim Trio; Siemens Healthcare AG, Erlangen, Germany) with a gradient field strength up to 1313
ENGELHORN ET AL
Academic Radiology, Vol 17, No 10, October 2010
Figure 1. The various steps of the algorithm for segmentation and quantification of the optical radiation are shown. In panel A2, the green channel from the diffusion tensor image from panel A1 is extracted for the red-green-blue (RGB) images, and threshold operation is performed in the green channel of every selected slice. The white pixel in panel A2, G image, represents green values above a computed threshold value, which is based on the image’s histogram. In panel A3, the optical radiation is segmented on the basis of the binarized image. The selected region of interest is the part of optical radiation with fiber tracts emerging from the lateral geniculate nucleus to the primary visual cortex. The resulting area is calculated with pixel per pixel summation. In panel A4, the volume is calculated as area times slice thickness. To determine the overall optical radiation volume, the different slice volumes are summed.
45 mT/m (72 mT/m effective). Anatomic data were obtained in a T1-weighted three-dimensional magnetization-prepared rapid gradient-echo sequence (repetition time, 900 ms; echo time, 3 ms; field of view, 23 23 cm; acquisition matrix size, 512 256 reconstructed to 512 512; reconstructed axial plans with 1.2-mm slice thickness). DTI was performed in the axial plane with 4-mm slice thickness and no interslice separation using a single-shot, spin-echo, echoplanar imaging diffusion tensor sequence covering the whole visual pathway (repetition time, 3400 ms; echo time, 93 ms; field of view, 23 23 cm; acquisition matrix size, 256 256 reconstructed to 512 512; number of signal averages, 7; partial Fourier acquisition, 60%). Diffusion weighting with a maximal b factor of 1000 s/mm2 was carried out along 15 icosahedral directions complemented by one scan with b = 0. Data sets were automatically corrected for imaging distortions and coregistered in reference to T1-weighted magnetization-prepared rapid gradient-echo images. These and further calculations, such as determining the independent elements of the diffusion tensor, deriving the corresponding eigenvalues and eigenvectors, and reconstructing and volume rendering fibers, were performed using dedicated software (Neuro 3D; Siemens Healthcare AG). 1314
For manual segmentation by two experienced neuroradiologists (T.E. and A.D.) and quantification of the optical radiation, the seed regions for the tracking algorithm were selected on the coregistered T1-weighted images that were overlaid on the DTI data. As a start region, an area consisting of approximately 12 to 18 voxels was chosen covering the LGN (5). Therewith, fiber tracts emerging from the LGN into the primary visual cortex were investigated. The resulting fiber pathways were evaluated visually for integrity and accuracy of the reconstructed fibers (ie, the course on the coregistered anatomic T1-weighted data was compared to the known anatomy of the visual pathway). In addition, volume rendering of the LGN and fiber tracts was performed by outlining the fourth neurons on each DTI slice by hand to calculate rarefaction. T2-weighted images were evaluated to exclude white-matter signal abnormalities from microinfracts that may involve the optical radiations and influence the results. For semiautomated segmentation and quantification of the optical radiation, a Mathcad-based software program (Parametric Technology Corporation, Needham, MA) was developed (Fig 1). The program selects slices in which optical radiation is present by making use of the pattern recognition,
Academic Radiology, Vol 17, No 10, October 2010
3 TESLA DTI OF OPTIC FIBRES
atrophy of the fourth neuron in patients with glaucoma (Fig 3).
DISCUSSION
Figure 2. Age-related rarefaction of the optical radiation (fourth neuron) in healthy controls is shown.
head organ dimension (ie, before actual image processing), location of the optical radiation in relation to the head size, and area of the largest green concentration (horizontal direction) as a starting point. Furthermore, the region of interest is selected for both sides. The green channel is extracted for the red-green-blue image and threshold operation in the green channel of every selected image. The white pixel represents green values above a computed threshold value, which is based on the image’s histogram. The optical radiation is segmented on the basis of the binarized image. The resulting area is calculated with a simple pixel per pixel summation. In the next step, the volume is calculated as area times slice thickness. To receive the overall optical radiation volume, the different slice volumes are summed.
RESULTS The semiautomated segmentation and quantification of the optical radiation was feasible in all patients and controls. The evaluation matched 94% with manual evaluation for all 42 patients. Cronbach’s a was >0.81 for calculation of the manually and semiautomatically derived volumes. The mean evaluation time of semiautomatic segmentation including the time for data transfer was significantly shorter (38 minutes) compared to manual segmentation (91 minutes). The ophthalmologic examinations revealed no significant effects in the eyes in all 20 controls, and severe glaucoma was observed in 22 severely ill patients with glaucoma. Agerelated optical atrophy of up to 30% was detected in healthy patients. Age-related rarefaction for healthy patients is shown in Figure 2. For healthy patients, there was high correlation between a decrease in the volume of the fourth neuron of both sides with increasing age (r = 0.82). For healthy patients, both right-side and left-side optical radiation fibers were similar and differed only in small ranges up to 300 mm3 (no significant differences in all test subjects). For severely ill patients with glaucoma, total optical fiber volume was up to 38% less compared to healthy patients. Compared to agematched healthy patients, we found more pronounced
Applying DTI reconstructions to the human visual system, visualization of the fourth neuron is possible noninvasively (5–7). A new semiquantitative segmentation approach with error 5% was implemented and compared to manual segmentation and quantification. As a result, in only two of 42 patients was the error of the automatic evaluation greater than 5%, demonstrating high accuracy of this method. In addition to its high accuracy, this method resulted in significantly faster assessment (38 vs 91 minutes), making this technique interesting for ophthalmologic studies using DTI. Manual volume measurements are based on tractography, which is influenced by the choice of flip angle and angle thresholds, whereas semiautomated volume measurements are based on colored flip angle maps, resulting in robust findings and high reproducibility. To our best knowledge, this is the first study to report the in vivo visualization of glaucoma and normal age-related changes of the optical radiation (fourth neuron) applying DTI fiber tracking using a high-field 3-T magnetic resonance scanner. Both methods demonstrated increasing atrophy of optical radiation with increasing age in controls: the volume of the optical radiation of subjects aged 75 to 83 years compared to those aged 45 to 54 years was decreased by >30%. Compared to age-matched controls, the total volume of optical radiation (fourth neuron) in patients with glaucoma was significantly reduced by up to 38%. A disadvantage of this technique is that the very last part of the optical radiation (intersection in the optical cortex) is not completely assessable, because in this region, the optic fibers spread in several directions. Also, the optical radiation starts at the LGN, proceeds posteriorly by dividing into three bundles: anterior or Meyer’s loop, central, and posterior bundles. It might also be a problem to separate the complete optical radiation within Meyer’s loop, because there is pronounced spreading when these fibers curve around the temporal horn and atrium (8). To overcome this problem, even higher resolution is necessary. Consequently, the described approach misses some parts of the optical radiation, but interobserver reliability was very high.
CONCLUSION The new approach for analyzing DTI of the optical radiation seems to be robust and is clearly faster compared to manual segmentation. Using DTI at 3 T, we could demonstrate that there is ongoing atrophy of the optical radiation with increasing age. Compared to age-matched controls, we found significant pronounced atrophy of the fourth neuron 1315
ENGELHORN ET AL
Academic Radiology, Vol 17, No 10, October 2010
Figure 3. Comparison of age-related rarefaction of the optical radiation (fourth neuron) in severely ill patients with glaucoma and healthy subjects.
in patients with glaucoma. Hence, this technique could have great clinical impact in the monitoring of patients with glaucoma. Further studies are needed to determine whether there is a correlation between the clinical extent of glaucoma and its effect on DTI-derived rarefaction of the optical radiation. REFERENCES 1. Conti AC, Raghupathi R, Trojanowski JQ, et al. Experimental brain injury induces regionally distinct apoptosis during the acute and delayed posttraumatic period. J Neurosci 1998; 18:5663–5672. 2. Dacey DM. Circuitry for color coding in the primate retina. Proc Natl Acad Sci U S A 1996; 93:582–588.
1316
3. Yu¨cel YH, Zhang Q, Weinreb RN, et al. Atrophy of relay neurons in magnoand parvocellular layers in the lateral geniculate nucleus in experimental glaucoma. Invest Ophthalmol Vis Sci 2001; 42:3216–3222. 4. Weber AJ, Chen H, Hubbard WC, et al. Experimental glaucoma and cell size, density, and number in the primate lateral geniculate nucleus. Invest Ophthalmol Vis Sci 2000; 41:1370–1379. 5. Staempfli P, Rienmueller A, Reischauer C, et al. Reconstruction of the human visual system based on DTI fiber tracking. J Magn Reson Imaging 2007; 26:886–893. 6. Taoka T, Sakamoto M, Iwasaki S, et al. Diffusion tensor imaging in cases with visual field defect after anterior temporal lobectomy. AJNR Am J Neuroradiol 2005; 26:797–803. 7. Shimony JS, Burton H, Epstein AA, et al. Diffusion tensor imaging reveals white matter reorganization in early blind humans. Cereb Cortex 2006; 16: 1653–1661. 8. Meyer A. The connections of the occipital lobes and the presents statues of the cerebral visual affections. Trans Assoc Phys 1907; 22:7–23.
A New Semi-quantitative Approach for Analysing 3T Diffusion Tensor Imaging of Optic Fibres and Its Clinical Evaluation in Glaucoma Tobias Engelhorn, MD, Sultan Haider, MSc, Georg Michelson, MD, Arnd Doerfler, MD Rationale and Objectives: Diffusion tensor imaging can depict rarefaction of the optical fibers. Manual segmentation is time consuming. The purposes of the study were (1) to present a new semiquantitative segmentation approach for analyzing 3-T diffusion tensor imaging of optical fibers and (2) to clinically test the new approach by comparing optic fiber rarefaction in patients with glaucoma to that in agematched, healthy controls. Materials and Methods: To perform semiautomated and quantitative segmentation of the optical radiation, a Mathcad-based software program was developed. The results were compared to the manual evaluation of the images performed by two experienced neuroradiologists. The eyes of 42 subjects (22 patients with glaucoma and 20 controls) aged 37 to 86 years were assessed in full ophthalmologic examinations. Magnetic resonance imaging was performed using a 3-T high-field scanner. Results: The evaluation using the new approach matched 94% with manually acquired rarefaction of the optical radiation; Cronbach’s a was >0.81 for calculation of the manually and semiautomatically derived volumes. Conclusion: The new approach seems to be robust and is clearly faster compared to the more tedious manual segmentation. Using diffusion tensor imaging at 3 T, it could be shown that there was increasing atrophy of the optical radiation (fourth neuron) with increasing age in patients with glaucoma. Compared to age-matched, healthy patients, more pronounced atrophy of the fourth neuron was found in patients with glaucoma. Key Words: 3T; DTI; MRI; glaucoma. ªAUR, 2010
W
hether glaucoma-related rarefaction of the optical radiation exists is still dubious, because there are no experimental or clinical studies dealing with this topic. Loss of ganglion cells in the retina and their axons representing the third neuron and atrophy of the fourth neuron have been described in diseases such as traumatic injury and glaucoma (1–4). The axons of various retinal ganglion cell subtypes, differing in specific morphology and function, exit the eyeball and finally converge to anatomically distinguishable layers of the lateral geniculate nucleus (LGN) (2), where a loss of neural cells has also been described for glaucoma (3).
Acad Radiol 2010; 17:1313–1316 From the Departments of Neuroradiology (T.E., A.D.) and Ophthalmology (G.M.), University Erlangen-Nuremberg, Schwabachanlage 6, D-91054 Erlangen, Germany; and Siemens AG Healthcare Sector, Erlangen, Germany (S.H.). Drs Engelhorn and Haider contributed equally to this work and thus share first authorship. Received March 24, 2010; accepted April 27, 2010. Address correspondence to: T.E. e-mail: [email protected] ªAUR, 2010 doi:10.1016/j.acra.2010.04.017
Diffusion tensor imaging (DTI) can depict the optic nerve (third neuron) and optical radiation (fourth neuron) (5–7). However, manual analysis is time consuming. We used DTI with a newly developed semiautomatic segmentation approach to assess rarefaction of the fourth neuron in 20 control subjects aged 45 to 83 years and in 22 severely ill patients with glaucoma aged 37 to 86 years.
MATERIALS AND METHODS Eyes were assessed in full ophthalmologic examinations. Available additional examinations were referred to for evaluation, including Heidelberg Retinal Tomography (Heidelberg Engineering, Heidelberg, Germany), automated perimetry (Octopus 101 dG2; Interzeag, Schlieren, Switzerland), spatial-temporal contrast sensitivity (frequency doubling test; Carl Zeiss Meditec AG, Jena, Germany), nonmydriatic fundus images (nonmyd-alpha 45; Kowa Optimed, Inc, Torrance, CA), and measurement of intraocular pressure. Magnetic resonance imaging was performed using a 3-T high-field scanner (Magnetom Tim Trio; Siemens Healthcare AG, Erlangen, Germany) with a gradient field strength up to 1313
ENGELHORN ET AL
Academic Radiology, Vol 17, No 10, October 2010
Figure 1. The various steps of the algorithm for segmentation and quantification of the optical radiation are shown. In panel A2, the green channel from the diffusion tensor image from panel A1 is extracted for the red-green-blue (RGB) images, and threshold operation is performed in the green channel of every selected slice. The white pixel in panel A2, G image, represents green values above a computed threshold value, which is based on the image’s histogram. In panel A3, the optical radiation is segmented on the basis of the binarized image. The selected region of interest is the part of optical radiation with fiber tracts emerging from the lateral geniculate nucleus to the primary visual cortex. The resulting area is calculated with pixel per pixel summation. In panel A4, the volume is calculated as area times slice thickness. To determine the overall optical radiation volume, the different slice volumes are summed.
45 mT/m (72 mT/m effective). Anatomic data were obtained in a T1-weighted three-dimensional magnetization-prepared rapid gradient-echo sequence (repetition time, 900 ms; echo time, 3 ms; field of view, 23 23 cm; acquisition matrix size, 512 256 reconstructed to 512 512; reconstructed axial plans with 1.2-mm slice thickness). DTI was performed in the axial plane with 4-mm slice thickness and no interslice separation using a single-shot, spin-echo, echoplanar imaging diffusion tensor sequence covering the whole visual pathway (repetition time, 3400 ms; echo time, 93 ms; field of view, 23 23 cm; acquisition matrix size, 256 256 reconstructed to 512 512; number of signal averages, 7; partial Fourier acquisition, 60%). Diffusion weighting with a maximal b factor of 1000 s/mm2 was carried out along 15 icosahedral directions complemented by one scan with b = 0. Data sets were automatically corrected for imaging distortions and coregistered in reference to T1-weighted magnetization-prepared rapid gradient-echo images. These and further calculations, such as determining the independent elements of the diffusion tensor, deriving the corresponding eigenvalues and eigenvectors, and reconstructing and volume rendering fibers, were performed using dedicated software (Neuro 3D; Siemens Healthcare AG). 1314
For manual segmentation by two experienced neuroradiologists (T.E. and A.D.) and quantification of the optical radiation, the seed regions for the tracking algorithm were selected on the coregistered T1-weighted images that were overlaid on the DTI data. As a start region, an area consisting of approximately 12 to 18 voxels was chosen covering the LGN (5). Therewith, fiber tracts emerging from the LGN into the primary visual cortex were investigated. The resulting fiber pathways were evaluated visually for integrity and accuracy of the reconstructed fibers (ie, the course on the coregistered anatomic T1-weighted data was compared to the known anatomy of the visual pathway). In addition, volume rendering of the LGN and fiber tracts was performed by outlining the fourth neurons on each DTI slice by hand to calculate rarefaction. T2-weighted images were evaluated to exclude white-matter signal abnormalities from microinfracts that may involve the optical radiations and influence the results. For semiautomated segmentation and quantification of the optical radiation, a Mathcad-based software program (Parametric Technology Corporation, Needham, MA) was developed (Fig 1). The program selects slices in which optical radiation is present by making use of the pattern recognition,
Academic Radiology, Vol 17, No 10, October 2010
3 TESLA DTI OF OPTIC FIBRES
atrophy of the fourth neuron in patients with glaucoma (Fig 3).
DISCUSSION
Figure 2. Age-related rarefaction of the optical radiation (fourth neuron) in healthy controls is shown.
head organ dimension (ie, before actual image processing), location of the optical radiation in relation to the head size, and area of the largest green concentration (horizontal direction) as a starting point. Furthermore, the region of interest is selected for both sides. The green channel is extracted for the red-green-blue image and threshold operation in the green channel of every selected image. The white pixel represents green values above a computed threshold value, which is based on the image’s histogram. The optical radiation is segmented on the basis of the binarized image. The resulting area is calculated with a simple pixel per pixel summation. In the next step, the volume is calculated as area times slice thickness. To receive the overall optical radiation volume, the different slice volumes are summed.
RESULTS The semiautomated segmentation and quantification of the optical radiation was feasible in all patients and controls. The evaluation matched 94% with manual evaluation for all 42 patients. Cronbach’s a was >0.81 for calculation of the manually and semiautomatically derived volumes. The mean evaluation time of semiautomatic segmentation including the time for data transfer was significantly shorter (38 minutes) compared to manual segmentation (91 minutes). The ophthalmologic examinations revealed no significant effects in the eyes in all 20 controls, and severe glaucoma was observed in 22 severely ill patients with glaucoma. Agerelated optical atrophy of up to 30% was detected in healthy patients. Age-related rarefaction for healthy patients is shown in Figure 2. For healthy patients, there was high correlation between a decrease in the volume of the fourth neuron of both sides with increasing age (r = 0.82). For healthy patients, both right-side and left-side optical radiation fibers were similar and differed only in small ranges up to 300 mm3 (no significant differences in all test subjects). For severely ill patients with glaucoma, total optical fiber volume was up to 38% less compared to healthy patients. Compared to agematched healthy patients, we found more pronounced
Applying DTI reconstructions to the human visual system, visualization of the fourth neuron is possible noninvasively (5–7). A new semiquantitative segmentation approach with error 5% was implemented and compared to manual segmentation and quantification. As a result, in only two of 42 patients was the error of the automatic evaluation greater than 5%, demonstrating high accuracy of this method. In addition to its high accuracy, this method resulted in significantly faster assessment (38 vs 91 minutes), making this technique interesting for ophthalmologic studies using DTI. Manual volume measurements are based on tractography, which is influenced by the choice of flip angle and angle thresholds, whereas semiautomated volume measurements are based on colored flip angle maps, resulting in robust findings and high reproducibility. To our best knowledge, this is the first study to report the in vivo visualization of glaucoma and normal age-related changes of the optical radiation (fourth neuron) applying DTI fiber tracking using a high-field 3-T magnetic resonance scanner. Both methods demonstrated increasing atrophy of optical radiation with increasing age in controls: the volume of the optical radiation of subjects aged 75 to 83 years compared to those aged 45 to 54 years was decreased by >30%. Compared to age-matched controls, the total volume of optical radiation (fourth neuron) in patients with glaucoma was significantly reduced by up to 38%. A disadvantage of this technique is that the very last part of the optical radiation (intersection in the optical cortex) is not completely assessable, because in this region, the optic fibers spread in several directions. Also, the optical radiation starts at the LGN, proceeds posteriorly by dividing into three bundles: anterior or Meyer’s loop, central, and posterior bundles. It might also be a problem to separate the complete optical radiation within Meyer’s loop, because there is pronounced spreading when these fibers curve around the temporal horn and atrium (8). To overcome this problem, even higher resolution is necessary. Consequently, the described approach misses some parts of the optical radiation, but interobserver reliability was very high.
CONCLUSION The new approach for analyzing DTI of the optical radiation seems to be robust and is clearly faster compared to manual segmentation. Using DTI at 3 T, we could demonstrate that there is ongoing atrophy of the optical radiation with increasing age. Compared to age-matched controls, we found significant pronounced atrophy of the fourth neuron 1315
ENGELHORN ET AL
Academic Radiology, Vol 17, No 10, October 2010
Figure 3. Comparison of age-related rarefaction of the optical radiation (fourth neuron) in severely ill patients with glaucoma and healthy subjects.
in patients with glaucoma. Hence, this technique could have great clinical impact in the monitoring of patients with glaucoma. Further studies are needed to determine whether there is a correlation between the clinical extent of glaucoma and its effect on DTI-derived rarefaction of the optical radiation. REFERENCES 1. Conti AC, Raghupathi R, Trojanowski JQ, et al. Experimental brain injury induces regionally distinct apoptosis during the acute and delayed posttraumatic period. J Neurosci 1998; 18:5663–5672. 2. Dacey DM. Circuitry for color coding in the primate retina. Proc Natl Acad Sci U S A 1996; 93:582–588.
1316
3. Yu¨cel YH, Zhang Q, Weinreb RN, et al. Atrophy of relay neurons in magnoand parvocellular layers in the lateral geniculate nucleus in experimental glaucoma. Invest Ophthalmol Vis Sci 2001; 42:3216–3222. 4. Weber AJ, Chen H, Hubbard WC, et al. Experimental glaucoma and cell size, density, and number in the primate lateral geniculate nucleus. Invest Ophthalmol Vis Sci 2000; 41:1370–1379. 5. Staempfli P, Rienmueller A, Reischauer C, et al. Reconstruction of the human visual system based on DTI fiber tracking. J Magn Reson Imaging 2007; 26:886–893. 6. Taoka T, Sakamoto M, Iwasaki S, et al. Diffusion tensor imaging in cases with visual field defect after anterior temporal lobectomy. AJNR Am J Neuroradiol 2005; 26:797–803. 7. Shimony JS, Burton H, Epstein AA, et al. Diffusion tensor imaging reveals white matter reorganization in early blind humans. Cereb Cortex 2006; 16: 1653–1661. 8. Meyer A. The connections of the occipital lobes and the presents statues of the cerebral visual affections. Trans Assoc Phys 1907; 22:7–23.