The Neuroradiology Journal 20: 139-147, 2007

www. centauro. it

Influence of User-Defined Parameters on Diffusion Tensor Tractography of the Corticospinal Tract P.M. PARIZEL*, V. VAN ROMPAEY**, R.VAN LOOCK**, W. VAN HECKE***, J.W. VAN GOETHEM*, A. LEEMANS***, J. SIJBERS*** * Department of Radiology and Medical Imaging, University Hospital Antwerp; Edegem, Belgium ** Faculty of Medicine, University of Antwerp; Antwerp, Belgium *** Vision Lab, Department of Physics, University of Antwerp; Antwerp, Belgium

Key words: white matter, diffusion tensor imaging, fiber tracking parameters

SUMMARY – This study discusses the influence of user-defined parameters on fiber tracking results obtained from a standard deterministic streamline tractography algorithm. Diffusion tensor imaging with fiber tractography was performed in five healthy volunteers. A region of interest was highlighted in the ventral part of the pons at the level of the middle cerebellar peduncle. The parameters studied were angle threshold, fractional anisotropy threshold, step length and number of seed samples per voxel. Changes in fiber tracts were described for increasing values per parameter. Increasing the angle threshold resulted in more and longer fibers. A higher fractional anisotropy threshold resulted in decreased length and fiber tracts that were not representative. Increasing the step length decreased the fiber continuity and altered its position. A higher number of seed samples per voxel resulted in a higher fiber tract density. When interpreting diffusion tensor images, the reader should understand the influence of user-defined settings on the results, and should be aware of the inter-dependency of fiber tracking parameters.

Introduction Diffusion tensor imaging (DTI) is an advanced magnetic resonance (MR) technique characterizing the amount and orientation of water self-diffusion 1-3. Fiber-tract trajectories in soft fibrous tissues, such as nerves, muscles, ligaments, tendons, etc., can be generated from the acquired DTI data sets. This is referred to as diffusion tensor tracking (DTT) or fiber tractography (FT) 4-5. DTI, combined with FT, is the only approach available to visualize white matter in vivo and non-invasively, and can provide exceptional information on the architecture of white matter 6-8. Dense packing of axons and their axonal membranes hinder water diffusion significantly perpendicular to the long axis of the fibers relative to the parallel direction. Directionally-dependent diffusion is therefore called anisotropic diffusion. On the other hand, if diffusion is the

same in all directions, it is called isotropic diffusion 2. Several anisotropy measures can be used, one of the most common being fractional anisotropy (FA) 9. The direction of maximum diffusivity, i.e. the preferred local displacement direction of the water molecules, has been shown to coincide with the white matter fiber tract orientation 10. This information can be described with a mathematic model of diffusion in three-dimensional (3D) space, called the diffusion tensor. Actual mapping of the fiber pathways by FT is achieved by consecutively connecting this voxelbased directional information in a reproducible 3D manner 4. To facilitate the 3D visualization, the direction of maximum diffusivity can be mapped for instance by using red, green and blue (RGB) color channels defining the sagittal, coronal and axial directions, respectively, combined with a color-brightness modulated by the FA. In this way, a convenient summary map 139

Influence of User-Defined Parameters on Diffusion Tensor Tractography of the Corticospinal Tract

can be obtained from which the degree of anisotropy and the local fiber direction can easily be determined 11-12, 6. The clinical applications of DTI are diverse and promising. It is expected that diffusion tensor imaging will become an important tool for the study of brain anatomy 13-14, development of the human brain 15-17, diagnosis of various white matter abnormalities, such as amyotrophic lateral sclerosis 18, stroke 19, multiple sclerosis 20-23, brain tumors 24-25, etc. With the advent of DTI, a rigorous formulation of the full 3D Gaussian diffusion process was established providing not only a quantitative measure for diffusion anisotropy (like FA), but also the corresponding predominant directions of water diffusion. This framework has led to a proliferation of FT algorithms with varying degrees of complexity and emphasizing different aspects to extricate the structural connectivity 26-29. Although FT allows this fiber architecture to be visualized and explored, objective and reliable quantitative results remain difficult to obtain due to the complex multi-parameter fiber tracking problem, the low signalto-noise ratio (SNR) of the DT-MRI data sets, the partial volume effect (i.e., powder averaging) and the lack of ground truth 30-32. Slightly changing one of the user-defined fiber tracking parameters, such as angle threshold, fractional anisotropy threshold, etc. can yield very different results. Only a few references are available on the influence of these parameters on DTT data 33-35. The purpose of this study is to discuss the influence of several user-defined parameters on the fiber tracking results, as obtained from a standard deterministic streamline tractography algorithm 36. The parameters we studied were angle threshold, FA threshold, step length, number of seed samples per voxel. We also briefly mention the effect of two parameters, i.e. radius of the tube and decimate factor, which do not affect the fiber tracking calculation as such, but can be tuned for optimal display quality. It should be noted that the effect of changing fiber tracking parameters is well known to the experienced 'tractography-user', but a systematic overview of the resulting effect of these parameter settings is very useful for the novice user. Acquaintance with these parameters and a sound judgment in their influence on the images is required for a better understanding of fiber tractography results. 140

P.M. Parizel

Materials and Methods Acquisition

We performed diffusion tensor imaging on five healthy volunteers (one woman and four men, aged between 20 and 21 years). Informed consent was obtained. The subjects had no history of neurologic disorder. Imaging was performed on a clinical 1.5 Tesla unit (Magnetom Sonata Siemens AG, Erlangen, Germany) with echo-planar readout. The DTI dataset protocol took 12 minutes and 18 seconds, it included b-values of 0 and 700 s/mm2; 60 gradient directions; bandwidth, 1302 Hz/Px; matrix size, 128×128; field of view, 256 mm; voxel size, 2.0×2.0×2.0 mm3. The sequence generated 60 slices, of 2 mm thickness, in the axial plane. The other relevant parameters were: TR/TE, 10400/100 ms; relative SNR ratio, 1.00; echo spacing, 0.83 ms. Imaging was performed using a standard circularly polarized (CP) head coil. FT parameter settings

In all volunteer datasets, a region of interest (ROI) was determined in the axial plane of the low b-value images using a computational environment (DTI Task Card, Massachusetts General Hospital, Boston, MA) by an experienced neuroradiologist. The ROI contained the ventral part of the pons at the level of the middle cerebellar peduncles (as shown in figure 1). As default values for the fiber tracking parameters, we used: angle threshold (AT), 35°; FA threshold, 0.2; step length (SL), 2; and display parameters: number of seed samples per voxel (SPV), 2; radius of the tube (RT), 0.4 mm and decimate factor (DF), i.e. fiber tract subsampling percentage: 0%. SL represents the size of step that is made to connect the different positions of the tract pathway. Fibers crossing the ROI (as illustrated in figure 2) were then tracked and displayed into one of three chosen positions: anterior-posterior projection, right oblique and left oblique. Different values per parameter were introduced, each time leaving the other five parameters in the default value. We varied the AT from 10° to 89° in steps of 5°. FA threshold was analyzed from 0 to 0.5 in steps of 0.05; the SL from 0.1 to 10 in steps of 0.5; SPV from 1 to 5 in steps of 1. Each tracking image was numbered and saved on a hard drive, and all images were analyzed anonymously according to each pa-

www. centauro. it

Figure 1 The region of interest (blue) on a low b-value image is situated in the ventral part of the pons at the level of the brachium pontis.

The Neuroradiology Journal 20: 139-147, 2007

Figure 2 Diffusion tensor tracking (DTT) representation of the left corticospinal tract, seen in left anterior oblique (LAO) projection (for reasons of clarity, the right corticospinal tract was cropped). Color coding represents orientation (not the direction) of fibers: blue for superior-inferior, red for left-right, green for anterior-posterior. In the upper part of the figure, the corona radiata is seen (cr). The fiber tracts converge at the level of the posterior limb of the internal capsule (ic). At the level of the brainstem, the following structures can be identified: pontocerebellar fibers (pc) and brachium pontis (bp).

Figure 3 Influence of the angle threshold on DTT. LAO view of the left corticospinal tract. When the angle threshold is increased, the number of fiber tracts also increases. The drawback is that the number of spurious tracts also augments. We suggest using values between 30° and 40° to provide sufficient fiber density while minimalizing the number of spurious tracts.

Figure 4 Influence of the fractional anisotropy (FA) threshold on DTT. LAO view of the left corticospinal tract. When the FA threshold is increased, the number of fiber tracts decreases. We suggest using values between 0.05 and 0.20 to provide sufficient fiber density while minimalizing the number of spurious tracts.

rameter. Data analysis was performed using a blinded method, i.e. without specific information of the variables at first. Visual assessment of fiber tracts was performed in consensus by three observers.

The changes observed at increasing values per variable were recorded by using its designated number. This gave us the opportunity to assess the influence of all four fiber tracking parameters. 141

Influence of User-Defined Parameters on Diffusion Tensor Tractography of the Corticospinal Tract

Results We studied the effect of the different parameters on diffusion tensor fiber tracking, including the angle threshold, FA threshold, step length, and the number of seed samples per voxel. In addition we present the effect of the radius of the tube and the decimate factor on the display quality and rendering complexity. Variations in angle threshold (AT) affect the quality and quantity of the tract pathways. We varied AT values between 10° and 89°, with steps of 5°, and calculated the corresponding fiber tracts (example shown in figure 3). Increasing the AT settings results in the visualization of more and longer fibers. Below an AT value of 40° the corticospinal and cerebellar tracts appeared to increase in length and quantity for an increasing AT. Further increasing the AT to levels above 40°, did not affect visualization of the corticospinal tract, but did show perpendicular fibers which may arise from prolongation of cerebellar tracts, but which more likely are to be considered spurious tracts. The FA threshold parameter ranges from 0 (defined as maximal isotropy) to 1 (maximal anisotropy). If the FA threshold parameter is set at values higher than 0.5, no fibers are to be seen. Therefore, we varied FA threshold values from 0 to 0.5, in steps of 0.05 (example shown in figure 4). Overall, there is a decrease in the quantity of fibers seen outside the corticospinal tracts at higher fractional anisotropy threshold values. The length of the corticospinal fibers decreased slightly up to 0.15. At higher values, length and quality of the corticospinal fibers decreased further. Step length (SL) predefined values ranged from 0.1 to 10, considering intervals of 0.5 (example shown in figure 5). Overall, we observed a decrease in the quantity of fibers with increasing SL. Position of the fiber was not constant, though the number of fibers at that particular position had not changed. Values higher than 1.5 resulted in the gradual disappearance of perpendicular spurious tracts. Additionally, values between 1.5 and 4 showed a decrease in length of some corticospinal fibers. The number of seed samples per voxel (SPV) alters the density of fibers artificially. We studied values from 1 to 5 respecting intervals of 1 (figure 6). We observed higher density of fibers at higher values. The radius of the tube (RT) is an artificial parameter, not influencing the calculated tract pathways. We used values from 0.1 mm to 1 142

P.M. Parizel

mm, with 0.1 mm intervals (example shown in figure 7). The RT changed the aspect of the fibers, i.e. higher values resulted in increased diameter of all fibers, changing the degree of quality for the visual perception of these fibers. There are no differences in thickness of the actual fibers, i.e. the RT does not correlate with a certain feature of the fiber, e.g. real axonal diameters do not correlate with the displayed diffusion tensor fiber diameter. The decimate factor (DF) affects the number of sample points that has to be drawn for each tract. We analyzed values from 0% adding up 10% till 90% (example shown in figure 8). Higher percentage of DF resulted in less sample points to be drawn. This made rendering faster, but decreased display quality, hereby making tracts less smooth above values of 70% (figure 9). Discussion FT is a promising technique based on diffusion tensor imaging, which can provide unique information to visualize and study fiber tract architecture in white matter. It also has the potential to enhance our understanding of connectivity and fiber organization in the central and peripheral nervous system in vivo and non-invasively 5,27. DTI has proven to be useful in the study of brain development or changes due to aging, as well as in the diagnosis of several types of brain disorders 37, including stroke 38, epilepsy 39, multiple sclerosis 40-41, HIV-1 infection 42, Parkinson’s disease 43, ischaemic leukoaraiosis 44; metabolic disorders such as X-linked adrenoleukodystrophy 45 and Krabbe disease 46; neuropsychiatric disorders such as schizophrenia 47, geriatric depression, alcoholism 48, Alzheimer’s disease 49-50 , and obsessive-compulsive disorder 51. Many other disorders are being studied. Clinical applications of DTT or FT are more limited. Reports have been published regarding the localization of lacunar infarctions by correlation with clinical symptoms 19, assessment of severity and recovery in lenticulostriate infarcts 52; detection of early pathological changes in multiple sclerosis 21; prediction of visual defects following anterior temporal lobe resection 53 ; neurosurgical planning in patients with cerebral tumors tracing the passage of functionally relevant fiber tracts along the tumors 54-55. Moreover DTT may provide additional findings beyond those seen with conventional MRI in

www. centauro. it

The Neuroradiology Journal 20: 139-147, 2007

Fig. 5 Influence of step length on DTT. LAO view of the left corticospinal tract. When the step length is increased, the number of fiber tracts decreases. We suggest using values between 1.5 and 4 mm to provide sufficient fiber density while minimalizing the number of spurious tracts.

Fig. 6 Influence of the number of samples per voxel on DTT. LAO view of the left corticospinal tract. When the number of samples per voxel is increased, the number of fiber tracts also increases. The drawback is that this increase in samples per voxel requires more computation time. We suggest using 2 as value to provide sufficient fiber density.

Fig. 7 Influence of the radius of the tube on DTT. LAO view of the left corticospinal tract. When the radius of the tube is increased, the diameter of fiber tracts also increases. We suggest using values between 0.200 mm and 0.400 mm to provide a clear view on the fiber tracts.

Fig. 8 Influence of the decimate factor on DTT. LAO view of the left corticospinal tract. When the decimate factor is increased, rendering becomes faster, but display quality decreases. We suggest using 0% as value to provide an optimal display quality.

central nervous system anomalies like callosal agenesis, cortical dysplasia, periventricular leukomalacia and Joubert syndrome 56. Although DTT is useful, it remains an ‘artificial’ representation of reality. A fiber tract does not physically represent a single fiber tract, but only indicates fiber direction 32. Fiber tracking results are highly sensitive to the choice of parameter settings. Therefore, it is important to understand the influence of these parameters on the DTT result. Though the topic DTI has

generated much literature 57, to our knowledge few references have investigated the influence of different fiber tracking parameter settings. We studied six parameters and only found relevant literature on fractional anisotropy threshold, inner product threshold 33,29 and step size 30, and non-quantitative literature on tube radius 31. Using the angular deviation between successive step directions, the angle threshold (AT) terminates the fiber tract propagation when 143

Influence of User-Defined Parameters on Diffusion Tensor Tractography of the Corticospinal Tract

P.M. Parizel

Fig. 9 Influence of the decimate factor on DTT. LAO view of the left corticospinal tract. Fiber tracts become angular and less smooth above values of 70%.

angles exceed the value chosen as AT. For example, if fiber tract curvature equals an angle of 90° within a distance of 1 mm, it is likely that from this point on the fiber tract is erroneous, and therefore, the AT stops the fiber tract at this point. Related terms are angular deflection threshold and inner product threshold, using the scalar product of successive elements. We observed that the AT affects the quality and quantity of fiber tracts, i.e. higher AT settings result in the visualization of more and longer fibers, but also is more likely to introduce spurious tracts, as described above. For example, AT values above 40° showed perpendicular fibers, likely to be spurious tracts, excluding these values as optimal parameter settings. Below 40°, increasing AT values appeared to increase the corticospinal and cerebellar tracts in length and quantity. Several authors mention AT values, e.g. 35° or 41° 58,14. Default AT values in this study were 35°. On the basis of our findings, we conclude that values in the range of [30°-40°] provide sufficient fiber density minimizing the amount of spurious tracts. Nevertheless, the found track is not identical to the full true anatomical representation of the fiber bundle. Local tract orientation can only be validated through anisotropy. In regions of highly anisotropic white matter, fiber tracts follow the principal eigenvector. It is assumed that fiber tracts based on low values of FA are less reliable 4. The FA threshold is based on this prin144

ciple. When the FA value is below a certain threshold, the fiber tract stops its propagation. That is why FA threshold is called a termination threshold. We saw that increasing the FA threshold values decreases the quantity of fibers seen outside the corticospinal tract. Spurious tracts seen without using an FA threshold disappeared at values of 0.05 and higher. The length of the corticospinal fibers decreased slightly when the FA threshold was raised to 0.10 and 0.15. At higher values, i.e. above 0.20, length and quality of the corticospinal fibers decreased faster. The FA threshold has been previously studied 29 and an optimal value of 0.20 was suggested. Several authors mention FA threshold, e.g. 0.35, 0.30, 0.27, 0.20, 0.15 59-61,14,62 . Based on our findings, values of 0.05, 0.10, 0.15 and 0.20 are acceptable. The default FA threshold was 0.2. Note that these findings may only correlate with the corticospinal tract, other FA threshold values may be more useful in other white matter regions 29. This may be, among other factors due to track size and/or local noise conditions. The choice of step length (SL) is determined by the size of the voxel. It can been defined as the gap between the boundaries of the voxel at the step entry and exit points 26. Smaller SL values make it more difficult to reach the angle threshold because the angle between two connection elements will decrease. If the SL increases, the quantity of fibers decreases. The position of the fibers was not constant, al-

www. centauro. it

though the number of fibers at a particular position had not changed. Parameter values below 1.5 mm show several spurious tracts. Spurious tracts diminished using values higher than 1.5 mm, but above 4 mm, fiber length and quality decreased. The default SL value was 2 mm. On the basis of our work, we suggest to use only values between 1.5 mm and 4 mm. The number of seed samples per voxel (SPV) alters the density of fibers artificially, meaning that higher values result in more fibers. For example, if this parameter is too small, fine details of a dataset may not be reconstructed; On the other hand, increasing its value too much, increases the required computation time dramatically. Values of four and five SPV take at least one minute of processing on our workstation. Considering the fact that this parameter is a ’super sampling factor’, higher values only result in an artificial increase of fiber quantity. The absolute quantity of fibers is not relevant, although it may influence the resolving power. To avoid a trade-off in increase of computation, we used a default SPV value of two. The radius of the tube (RT) is also an artificial parameter. RT changes the aspect of the fibers, with higher values resulting in an increased diameter of all fibers. The in-vivo axonal diameter does not correlate with the displayed diffusion tensor fiber diameter. Note that RT does not correlate with a certain feature of the fiber. Because RT is a subjective parameter, it is not possible to state an optimal value. We suggest values between 0.200 mm and 0.400 mm, because lower values 0.200 mm result in an irregular outline of the fiber tracts and values above 0.400 mm result in fiber tracts that merge. The decimate factor (DF) affects the number of sample points that has to be drawn, i.e. subsampling. This makes rendering faster, but decreases the display quality, which might impede the detailed visualization of small tract features. With DF values above 70% the tracts became angular and less smooth. The default decimate factor value was 0% and based on our findings we can agree with this value because no sacrifices have to be made on the output quality. Conclusion In summary, the following aspects, regarding FA threshold, angle threshold, step length, and number of seed points within a voxel, should

The Neuroradiology Journal 20: 139-147, 2007

be taken under consideration for tracking fiber pathways: 1. FT computation times are higher for: a lower FA threshold (tracking continues longer), a lower step length (more computations per unit of distance), a higher angle threshold (less physical constraint on the fiber’s shape), and a higher number of seed points per voxel (a higher number of tracts are being calculated). The latter one having a linear dependency (doubling the parameter value, doubles computation time) and the highest influence. Computation times for display rendering can be decreased by increasing the decimate factor (subsampling of the fiber tracts). 2. The actual shape of the tract is determined by the step length (defining the position of the next tract sample point in the data set). 3. The length of the tract mainly depends on the FA threshold and the angle threshold: for instance, tracking continues longer for lower FA and for higher angle thresholds. 4. It is very important to know that the step length and angle threshold are inter-dependent: changing the step size (for instance to save computation time when tracking from a large ROI) automatically requires adjustment of the angle threshold. If not, the physical curvature constraints, determined by these two parameters, will be changed unintentionally. Generally, it should be noted that several more technical issues have an indirect influence on the tracking results, such as the applied interpolation field, partial volume effect, geometric artifacts due to acquisition imperfections, etc. which are beyond the scope of the clinical end-user. Our work has shown that it is necessary to be aware of the inter-dependency of several parameters in diffusion tensor fiber tracking. In general, experience is necessary to get a better feel with DTT, in order to make relevant comparisons between normal individuals and patients. There is no such thing as an optimal parameter setting for DTT, and trade-offs have to be made between theoretical goals and what is possible in clinical practice. Optimizing parameter settings is an arbitrary and subjective process. Moreover, it is likely that these optimal values are different according to the region that was being studied, and additionally according to the equipment and sequences being used. With the presented example of the corticospinal tracts, we hope that our work will illustrate the influence of diffusion tensor tracking parameters on the quality of the fiber tracking results. 145

Influence of User-Defined Parameters on Diffusion Tensor Tractography of the Corticospinal Tract

P.M. Parizel

References 1 Basser PJ, Le Bihan D: Fiber orientation mapping in an anisotropic medium with NMR diffusion spectroscopy. In: Proceedings of the 11th Annual Meeting of the SMRM, Berlin 1992: 1221. 2 Basser PJ, Mattiello J, Le Bihan D: MR diffusion tensor spectroscopy and imaging. Biophys J 66: 259-267, 1994. 3 Basser PJ: New histological and physiological stains derived from diffusion-tensor MR images. Ann NY Acad Sci. 820: 123-138, 1997. 4 Mori S, van Zijl PC: Fiber tracking: principles and strategies - a technical review. NMR Biomed. 15: 46880, 2002. 5 Basser PJ, Pajevic S, Pierpaoli C et Al: In vivo fiber tractography using DT-MRI data. Magn Reson Med. 44: 625-632, 2000. 6 Pierpaoli C, Jezzard P, Basser PJ et Al: Diffusion tensor MR imaging of human brain. Radiology 201: 673648, 1996. 7 Makris N, Worth AJ, Sorensen AG et Al: Morphometry of in vivo human white matter association pathways with diffusion weighted magnetic resonance imaging. Ann Neurol 42: 951-962, 1997. 8 Beaulieu C, Allen PS: Determinants of anisotropic water diffusion in nerves. Magn Reson Med 32: 592-601, 1994. 9 Basser PJ, Pierpaoli C: Microstructural and physiological features of tissues elucidated by quantitative diffusion tensor MRI. J Magn Reson 111: 209-219, 1996. 10 Moseley ME, Cohen Y, Kucharczyk J et Al: Diffusionweighted MR imaging of anisotropic water diffusion in cat central nervous system. Radiology 176: 439-446, 1990. 11 Douek P, Turner R, Pekar J et Al: MR color mapping of myelin fiber orientation. J Comput Assist Tomogr 15: 923-929, 1991. 12 Pajevic S, Pierpaoli C: Color schemes to represent the orientation of anisotropic tissues from diffusion tensor data: application to white matter fiber tract mapping in the human brain. Magn Reson Med 42: 526-540, 1999. 13 Jellison BJ, Field AS, Medow J et Al: Diffusion tensor imaging of cerebral white matter: a pictorial review of physics, fiber tract anatomy, and tumor imaging patterns. AJNR Am J Neuroradiol 25: 356-69, 2004. 14 Wakana S, Jiang H, Nagae-Poetscher LM et Al: Fiber tract-based atlas of human white matter anatomy. Radiology 230: 77-87, 2004. 15 Huppi P, Maier S, Peled S et Al: Microstructural development of human newborn cerebral white matter assessed in vivo by diffusion tensor magnetic resonance imaging. Pediatr Res 44: 584-590, 1998. 16 Mukherjee P, Miller JH, Shimony JS et Al: Diffusiontensor MR imaging of gray and white matter development during normal human brain maturation. Am J Neuroradiol 23: 1445-1456, 2002. 17 Abe O, Aoki S, Hayashi N et Al: Normal aging in the central nervous system: quantitative MR diffusion-tensor analysis. Neurobiology of aging 23: 433-441, 2002. 18 Abe O, Yamada H, Masutani Y et Al: Amyotrophic lateral sclerosis: diffusion tractography and voxel-based analysis. NMR Biomed 17: 411-416, 2004. 19 Lee JS, Han MK, Kim SH et Al: Fiber tracking by diffusion tensor imaging in corticospinal tract stroke: Topographical correlation with clinical symptoms. Neuroimage 26: 771-6, 2005. 20 Wilson M, Tench CR, Morgan PS et Al: Pyramidal tract mapping by diffusion tensor magnetic resonance imaging in multiple sclerosis: improving correlations with disability. J Neurol Neurosurg Psychiatry 74: 203-207, 2003. 21 Lin X, Tench CR, Morgan PS et Al: ‘Importance sampling’ in MS: use of diffusion tensor tractography to

146

22

23

24

25 26 27 28 29

30 31 32

33 34

35 36 37 38

39

40

41

quantify pathology related to specific impairment. J Neurol Sci 237: 13-9, 2005. Hasan KM, Gupta RK, Santos RM et Al: Diffusion tensor fractional anisotropy of the normal-appearing seven segments of the corpus callosum in healthy adults and relapsing-remitting multiple sclerosis patients. J Magn Reson Imaging 21: 735-743, 2005. Gallo A, Rovaris M, Riva R et Al: Diffusion-tensor magnetic resonance imaging detects normal-appearing white matter damage unrelated to short-term disease activity in patients at the earliest clinical stage of multiple sclerosis. Arch Neurol 62: 803-8, 2005. Inglis BA, Neubauer D, Yang L et Al: Diffusion tensor MR imaging and comparative histology of glioma engrafted in the rat spinal cord. Am J Neuroradiol 20: 713-6, 1999. Mori S, Frederiksen K, van Zijl PC et Al: Brain white matter anatomy of tumor patients evaluated with diffusion tensor imaging. Ann Neurol 51: 377-80, 2002. Mori S, Crain BJ, Chacko VP et Al: Three dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann Neurol 45: 265-269, 1999. Bammer R, Acar B, Moseley ME: In vivo MR tractography using diffusion imaging. Eur J Radiol 45: 223-234, 2003. Conturo TE, Lori NF, Cull TS et Al: Tracking neuronal fiber pathways in the living human brain. Proc Natl Acad Sci U S A. 96: 10422-7, 1999. Huang H, Zhang J, van Zijl PC et Al: Analysis of noise effects on DTI-based tractography using the brute-force and multi-ROI approach. Magn Reson Med 52: 559-65, 2004. Leemans A, Sijbers J, Verhoye M et Al: Mathematical framework for simulating diffusion tensor MR neural fiber bundles. Magn Reson Med 53: 944-53, 2005. Anderson AW: Theoretical analysis of the effects of noise on diffusion tensor imaging. Magn Reson Med 46: 1174-88, 2001. Tournier JD, Calamante F, King MD et Al: Limitations and requirements of diffusion tensor fiber tracking: an assessment using simulations. Magn Reson Med 47: 701-8, 2002. Stieltjes B, Kaufmann WE, van Zijl PC et Al: Diffusion tensor imaging and axonal tracking in the human brainstem. Neuroimage 14: 723-35, 2001. Kunimatsu A, Aoki S, Masutani Y et Al: The optimal trackability threshold of fractional anisotropy for diffusion tensor tractography of the corticospinal tract. Magn Reson Med Sci 3: 11-7, 2004. McGraw T, Vemuri BC, Chen Y et Al: DT-MRI denoising and neuronal fiber tracking. Med Image Anal 8: 95111, 2004. Basser PJ, Pajevic S, Pierpaoli C, Duda J, Aldroubi A: In vivo fiber tractography using DT-MRI data. Magn Reson Med. 44: 625-632, 2000. Masutani Y, Aoki S, Abe O et Al: MR diffusion tensor imaging: recent advance and new techniques for diffusion tensor visualization. Eur J Radiol 46: 53-66, 2003. Lie C, Hirsch JG, Rossmanith C et Al: Clinicotopographical correlation of corticospinal tract stroke: a color-coded diffusion tensor imaging study. Stroke 35: 86-92, 2004. Eriksson SH, Rugg-Gunn FJ, Symms MR et Al: Diffusion tensor imaging in patients with epilepsy and malformations of cortical development. Brain 124: 617-26, 2001. Ciccarelli O, Werring DJ, Wheeler-Kingshott CA, et Al: Investigation of MS normal-appearing brain using diffusion tensor MRI with clinical correlations. Neurology 56: 926-33, 2001. Filippi M, Cercignani M, Inglese M et Al: Diffusion tensor magnetic resonance imaging in multiple sclerosis. Neurology 56: 304-11, 2001.

www. centauro. it

42 Pomara N, Crandall DT, Choi SJ et Al: White matter abnormalities in HIV-1 infection: diffusion tensor imaging study. Psychiat Res 106: 15-24, 2001. 43 Yoshikawa K, Nakata Y, Yamada K et Al: Early pathological changes in the parkinsonian brain demonstrated by diffusion tensor MRI. J Neurol Neurosurg Psychiatry 75: 481-4, 2004. 44 O’Sullivan M, Morris RG, Huckstep B et Al: Diffusion tensor MRI correlates with executive dysfunction in patients with ischaemic leukoaraiosis. J Neurol Neurosurg Psychiatry 75: 441-7, 2004. 45 Ito R, Melhem ER, Mori S et Al: Diffusion tensor brain MR imaging in X-linked cerebral adrenoleukodystrophy. Neurology 56: 544-7, 2001. 46 Guo AC, Petrella JR, Kurtzberg J et Al: Evaluation of white matter anisotropy in Krabbe disease with diffusion tensor MR imaging: initial experience. Radiology 21: 809-15, 2001. 47 Kubicki M, McCarley R, Westin CF et Al: A review of diffusion tensor imaging studies in schizophrenia. J Psychiatr Res 41: 15-30, 2005. 48 Lim KO, Helpern JA: Neuropsychiatric applications of DTI - a review. NMR Biomed 15: 587-93, 2002 49 Choi SJ, Lim KO, Monteiro I et Al: Diffusion tensor imaging of frontal white matter microstructure in early Alzheimer’s disease: a preliminary study. J Geriatr Psychiatry Neurol 18: 12-9, 2005. 50 Rose SE, Chen F, Chalk JB et Al: Loss of connectivity in Alzheimer’s disease: an evaluation of white matter tract integrity with color-coded MR diffusion tensor imaging. J Neurol Neurosurg Psychiatry 69: 528-30, 2000. 51 Szeszko PR, Ardekani BA, Ashtari M et Al: White matter abnormalities in obsessive-compulsive disorder: a diffusion tensor imaging study. Arch Gen Psychiatry 62: 782-90, 2005. 52 Konishi J, Yamada K, Kizu O et Al: MR tractography for the evaluation of functional recovery from lenticulostriate infarcts. Neurology 64: 108-13, 2005. 53 Powell HW, Parker GJ, Alexander DC et Al: MR tractography predicts visual field defects following temporal lobe resection. Neurology 65: 596-9, 2005. 54 Yu CS, Li KC, Xuan Y et Al: Diffusion tensor tractography in patients with cerebral tumors: A helpful technique for neurosurgical planning and postoperative assessment. Eur J Radiol 56: 197-204, 2005. 55 Clark CA, Barrick TR, Murphy MM et Al: White matter fiber tracking in patients with space-occupying lesions of the brain: a new technique for neurosurgical planning? Neuroimage 20: 1601-8, 2003. 56 Lee SK, Kim DI, Kim J et Al: Diffusion-tensor MR imaging and fiber tractography: a new method of describing aberrant fiber connections in developmental CNS anomalies. Radiographics 25: 53-68, 2005. 57 Field AS: Diffusion tensor imaging at the crossroads: fiber tracking meets tissue characterization in brain tumors. Am J Neuroradiol 26: 2168-9, 2005.

The Neuroradiology Journal 20: 139-147, 2007

58 Seghier ML, Lazeyras F, Vuilleumier P et Al: Functional magnetic resonance imaging and diffusion tensor imaging in a case of central poststroke pain. J Pain 6: 208-12, 2005. 59 Akai H, Mori H, Aoki S et Al: Diffusion tensor tractography of gliomatosis cerebri: fiber tracking through the tumor. J Comput Assist Tomogr 29: 127-9, 2005. 60 Wakana S, Nagae-Poetscher LM, Jiang H et Al: Macroscopic orientation component analysis of brain white matter and thalamus based on diffusion tensor imaging. Magn Reson Med 53: 649-57, 2005. 61 Jones DK, Simmons A, Williams SC et Al: Non-invasive assessment of axonal fiber connectivity in the human brain via diffusion tensor MRI. Magn Reson Med 42: 37-41, 1999.

Paul M. Parizel, MD, PhD Department of Radiology Universitair Ziekenhuis Antwerpen University of Antwerp Wilrijkstraat 10 B-2650 Edegem Belgium Tel.: + 32 3 821 35 32 Fax: + 32 3 821 45 32 E-mail: [email protected]

147

Influence of user-defined parameters on diffusion tensor tractography of the corticospinal tract.

This study discusses the influence of user-defined parameters on fiber tracking results obtained from a standard deterministic streamline tractography...
384KB Sizes 0 Downloads 0 Views