Tractography of Median Nerve Musculoskeletal Imaging Original Research Neslihan Kabakci 1 Bengi Gürses 1 Zeynep Firat 1 Ali Bayram 1 Aziz Müfit Uluğ 2,3 Arzu Kovanlıkaya 1 İlhami Kovanlıkaya 1 Kabakci N, Gürses B, Firat Z, et al. Keywords: diffusion tensor imaging, median nerve, MRI DOI:10.2214/AJR.07.2423 Received January 29, 2007; accepted after revision May 13, 2007. Preliminary data presented at the 2006 Annual Scientific Meeting of the European Society for Magnetic Resonance in Medicine and Biology, Warsaw, Poland. 1 Department of Radiology, Yeditepe University Hospital, Devlet Yolu Ankara Cad. 102-104, 34752 Kozyataği, Istanbul, Turkey. Address correspondence to N. Kabakci (nkabakci@yeditepe.edu.tr). 2 Department of Biomedical Engineering, Yeditepe University School of Engineering, Istanbul, Turkey. 3 Department of Radiology, Weill Medical College of Cornell University, New York, NY. AJR 2007; 189:923 927 0361 803X/07/1894 923 American Roentgen Ray Society Tensor Imaging and Tractography of Median Nerve: Normative s OBJECTIVE. The purposes of this study were to visualize the human median nerve on diffusion tensor imaging and to determine the normal fractional anisotropy (FA) value and apparent diffusion coefficient (ADC) of the normal median nerve. SUBJECTS AND METHODS. The wrists of 20 healthy volunteers and of two patients with carpel tunnel syndrome were examined with a 3-T MRI system with a standard eight-channel sensitivity-encoding head coil. tensor imaging was performed with a spin-echo echo-planar sequence. A T1-weighted sequence was performed for anatomic reference. After tractography, the FA value and ADC of the whole nerve were calculated automatically. Manual focal measurements also were obtained at the levels of the flexor retinaculum, wrist, and forearm. RESULTS. We visualized the median nerve with MR diffusion tensor tractography and followed the nerve for approximately 77.5 mm. We found the normative diffusion values of the median nerve were an FA of 0.709 ± 0.046 (SD) and an ADC of 1.016 ± 0.129 10 3 mm 2 /s. There was a statistically significant difference between the FA values obtained at the level of the flexor retinaculum and the values obtained from the other parts of the median nerve (p < 0.0001). We found a decrease in FA value (p < 0.01) and an increase in ADC (p < 0.05) with advancing age. CONCLUSION. The normative diffusion values of the human median nerve can be used as a reference in evaluation, diagnosis, and follow-up of entrapment, trauma, and regeneration of the median nerve. he median nerve is one of three T main nerves of the forearm. It arises from the lateral and medial cords of the brachial plexus (C6 T1). At the wrist level, it passes under the flexor retinaculum deep in relation to the flexor digitorum superficialis tendons through the carpal tunnel and divides into digital and muscular branches distal in relation to the flexor retinaculum. Several entrapment and compression syndromes affect these nerves of the forearm. Carpal tunnel syndrome (CTS) is the most common peripheral neuropathy of the upper extremity resulting from dysfunction of the median nerve. CTS is characterized by numbness in the first three digits and the radial aspect of the fourth digit and by thenar atrophy. There are several diagnostic methods for CTS, such as the Phalen maneuver, Flick test, and electromyography [1, 2]. Although the sensitivity and specificity of MRI in the diagnosis of CTS are low (sensitivity, 23 96%; specificity, 39 87%), a few signs, such as nerve enlargement, nerve flattening, and increased nerve signal intensity, do occur [1]. With application of the appropriate magnetic field gradients, MRI can be sensitized to the thermally driven random motion (diffusion) of water molecules in the direction of the field gradient. This technique is called diffusion-weighted imaging (DWI) [3]. Many materials have intrinsic structural properties that hinder diffusion so that diffusivity is greater in some directions than in others. This property is known as anisotropy. If there is no directional variation in diffusion rate, diffusion is said to be isotropic. Biologic tissues often are anisotropic because structures such as cell membranes and large protein molecules restrict the motion of water molecules. This property is called restricted diffusion. DWI usually shows diffusion information in one direction. In an anisotropic sample, diffusion tensor imaging (DTI) is required to fully characterize diffusion. In theory, to determine all elements of the diffusion tensor, at least six independent measurements with diffusion gradients applied sequentially along six noncollinear directions are required [4 7]. The direction of maximum diffusivity has been shown to coincide with the fiber tract orientation [6, 7]. In white matter fiber bun- AJR:189, October 2007 923
Fig. 1 26-year-old healthy subject who underwent tractography. A, Coronal color-coded map shows median nerve (arrows). B, Axial color-coded map obtained with aid of coronal map shows median nerve (arrow). C, Anatomic T1-weighted reference MR image confirms localization of median nerve. dles, water molecules move more along the fibers than in other directions [8]. With special fiber tracking software, it is possible to visualize neural tracts with DTI data [3]. Peripheral nerve imaging can be useful in the diagnosis of nerve compression, inflammation, trauma, and neuropathy and in the postoperative follow-up of nerve repair procedures. MR neurographic methods with fat-saturated heavily T2-weighted sequences or STIR T2- weighted sequences also can be used for peripheral nerve imaging [9]. DTI has been used mostly in imaging of the CNS to visualize white matter tracts. Several studies [10 12] have shown the clinical applications of DTI in neuronal disorders such as stroke, epilepsy, and CNS tumors. In a few studies [8, 12, 13], peripheral nerves have been visualized with DTI. For use of DTI in the diagnosis of pathologic conditions affecting peripheral nerves, the normative values of these nerves must be determined. In this study, our aims were to visualize the human median nerve with DTI and to determine the normal fractional anisotropy (FA) value and apparent diffusion coefficient (ADC) for use in the diagnosis of CTS, the most common peripheral neuropathy. Subjects and Methods Imaging was performed on the wrists of 21 volunteers (six men, 14 women; mean age, 27.5 years; A range, 21 35 years) with no history of medication use, surgery, or neurologic disorder related to the median nerve. Informed consent was obtained from all subjects. We also imaged the wrists of two patients with CTS to explore the applicability of the method to pathologic conditions. The examinations were performed on a 3-T MRI system (Intera Achieva, Philips Medical Systems) with a maximum gradient amplitude of 30 mt/m, slew rate of 150 T/m/s, and an eightchannel sensitivity-encoding head coil. During imaging, the coil was positioned in the center of the magnet bore. The subject was in a prone position with the hands over the head within the coil and immobilized with cushions, sandbags, and bandages. This positioning produced little discomfort for the subjects. A single-shot spin-echo echo-planar DTI sequence was used with the following parameters: TR/TE, 4,600/90; flip angle, 90 ; field of view, 140 mm; matrix size, 128 128; number of signals averaged, 3. weighting with a b value of 1,000 s/mm 2 was applied in 32 directions. An image without diffusion gradients also was acquired. Total sequence duration was 7 minutes 49 seconds. The data were obtained from 35 axial slices of 4-mm thickness with no gap. For anatomic reference, a T1-weighted axial sequence was obtained (382/20; flip angle, 90 ; field of view, 140 mm; number of signals averaged, 2). After DTI data were transferred to a PC, manufacturer-supplied software (PRIDE, Philips Medical Systems) was used for fiber tracking. The first step was to use color-coded maps in the coronal plane to B locate the nerve (Fig. 1A). Circular regions of interest (ROIs) were placed in the anatomic location of the median nerve expected on the basis of information from coronal and axial color-coded maps (Fig. 1B). ROI placement was done at two levels: the distal radioulnar joint and one of the most proximal slices. The locations of the coded fibers were confirmed by use of the anatomic location on T1-weighted reference images (Fig. 1C). A circular ROI larger than the nerve was used for this purpose (Fig. 2A). The anisotropy threshold was 0.3. After anatomic confirmation of the median nerve tract, the number of coded fibers, mean fiber length, mean FA value, and mean ADC were calculated with the software. The ADC calculated was one third of the trace of the diffusion tensor. On the color-coded map, three points along the tract of the median nerve in the axial and coronal planes were determined for focal FA measurements. One of the measurements was obtained at the level of flexor retinaculum. The other measurements were obtained at the wrist and forearm levels. At the wrist level, FA was measured from the median nerve at the level of musculotendinous junctions of the flexor digitorum and palmaris longus muscles. At the forearm level, the measurement was obtained approximately 5 cm proximal to the flexor retinaculum in the center of the flexor digitorum muscle bundles. For these measurements, ROIs smaller than the nerve were selected to avoid partial volume effects (Fig. 2B). Statistical analyses were performed with commercially available software (SPSS 11.0, SPSS). C 924 AJR:189, October 2007
Tractography of Median Nerve The mean FA values and ADCs obtained from the whole median nerve and focal measurements were compared by Friedman variance analysis and Wilcoxon s signed rank test. The comparison of FA values and ADCs on the basis of age was evaluated with Spearman s rank correlation test. A value of p < 0.05 was considered significant. Results The images of three of the 21 volunteer subjects were initially discarded because of the presence of motion artifacts. We repeated imaging of two of these subjects and included their data in the analysis. As a result, the median nerves of 20 subjects were visualized with DTI and fiber tracking. Figures 3 6 show the tractographic images of the median nerves of selected subjects. A Fig. 2 21-year-old healthy subject who underwent tractography. A, Axial color-coded map shows that for tractography, region of interest is larger than nerve. B, Axial color-coded map shows that for focal fractional anisotropy measurement, region of interest is smaller than nerve. Fig. 4 21-year-old healthy subject. Tractographic Fig. 5 29-year-old healthy subject. Tractographic The number and length of the tracked fibers, mean FA value, and mean ADC are summarized in Table 1. The mean number of tracked fibers was 1,417, and the mean length was 77.5 mm (range, 52.86 100.40 mm). The FA values of the normal median nerve ranged from 0.69 to 0.80. The mean FA value for all of the subjects was 0.709 ± 0.046 (SD). The ADCs of normal median nerves were between 0.74 and 1.27 10 3 mm 2 /s. The mean ADC was 1.016 ± 0.129 10 3 mm 2 /s. The mean FA value was 0.72 ± 0.05 for the women and 0.69 ± 0.01 for men. The mean ADC for the women was 0.97 ± 0.10 and that for the men was 1.12 ± 0.12. There was no statistically significant difference between the values for men and those for women. There were, however, significant differences with advancing age: a B Fig. 3 26-year-old healthy subject. Tractographic Fig. 6 27-year-old healthy subject. Tractographic significant increase in ADC (Spearman s rank correlation, p < 0.05; ρ = 0.547) and a significant decrease in FA value (Spearman s rank correlation, p <0.01; ρ =0.726). The mean FA value and ADC of focal measurements of the median nerve at the flexor retinaculum, wrist, and forearm levels are shown in Table 2. At the flexor retinaculum level, the mean FA value was 0.59 ± 0.07 and the mean ADC, 0.97 ± 0.03 10 3 mm 2 /s. At the wrist level, the mean FA value was 0.72 ± 0.07 and the mean ADC, 0.95 ± 0.02 10 3 mm 2 /s. At the forearm level, the mean FA value was 0.71 ± 0.08 and the mean ADC, 0.98 ± 0.02 10 3 mm 2 /s. There was a statistically significant difference between the FA value obtained from the flexor retinaculum level and those obtained from the other parts of the me- AJR:189, October 2007 925
TABLE 1: Number and Length of Tracked Fibers, Mean, and Mean for Median Nerves (n =20) Subject No. Age (y) No. of Fibers Fiber Length (mm) Mean Mean 1 31 1,428 81.10 0.67 1.01 2 27 496 65.40 0.66 0.97 3 27 192 52.86 0.69 1.27 4 21 738 96.10 0.75 0.91 5 28 779 74.50 0.70 0.96 6 23 978 80.40 0.74 0.74 7 26 2,855 59.17 0.69 1.17 8 32 2,108 67.60 0.67 1.02 9 30 2,882 66.10 0.74 0.94 10 34 139 78.40 0.64 1.16 11 27 514 64.20 0.69 1.24 12 21 696 73.60 0.74 0.88 13 29 1,444 82.50 0.66 1.04 14 35 770 92.30 0.68 1.08 15 25 1,492 110.40 0.71 1.00 16 25 313 61.05 0.79 1.02 17 26 4,907 77.43 0.77 0.90 18 29 1,998 87.17 0.74 1.04 19 22 1,694 81.25 0.80 0.90 20 32 1,922 100.00 0.67 1.08 Mean 27.5 1,417 77.5 0.709 ± 0.046 1.016 ± 0.129 dian nerve (p < 0.0001). No statistical difference was observed between ADCs measured at the three anatomic locations. In the two patients with CTS, the mean FA values of the median nerve were 0.41 and 0.44. These values were both 2 SDs below the normal value. Discussion DTI has been used mostly in visualization of white matter fibers in the CNS [7]. The results of our study show that DTI and fiber tracking at 3 T can be used to visualize the peripheral nerves. In a study with three healthy volunteers, Skorpil et al. [14] used DTI with fiber tracking on a 1.5-T MRI system to image the sciatic nerves in vivo. Hiltunen et al. [8] later used a 3-T MRI system to study the median, ulnar, and radial nerves and the tibial and peroneal nerves of six healthy subjects. Meek et al. [13] were the first to image abnormalities in the peripheral nerve with DTI. In a postoperative evaluation 3 months after fascicular nerve repair, those authors visualized the median nerve, which could be tracked only up to the transection site 1 month after the procedure. That report showed DTI to be capable of depicting living nerve fibers. Further prospective large series are needed, however, to assess the value of DTI in nerve repair surgery. Several indexes can be used for quantitative analysis of DWI and DTI data. We used ADC and FA value. ADC is a scalar value reflecting molecular diffusivity under motion restriction and is independent of magnetic field strength [8, 15, 16]. Like T1- and T2-weighted relaxation, diffusivity is thought to be an intrinsic tissue parameter and has been used primarily in the diagnosis of acute cerebral ischemia [5]. The other parameter, FA value, is one of the most common anisotropy metrics. It ranges from 0 to 1, where 0 is isotropic and 1 is fully anisotropic. The direction of maximum diffusivity is mapped with red, green, and blue color coding in which brightness is modulated by FA. The result is a summary map from which the degree of anisotropy and local fiber direction can be determined [7]. In our study, the normal FA value and ADC of the median nerve were defined. To our knowledge, only one study [9] has shown the FA value and ADC of this nerve. In that study, the FA values of four subjects were between 0.69 and 0.89, and the ADCs ranged from 0.94 to 1.36 10 3 mm 2 /s. In our study, the whole median nerve tract had a mean FA value of 0.709 ± 0.046 and a mean ADC of 1.016 ± 0.129 10 3 mm 2 /s. We believe that these normative values may play an important role in the diagnosis of CTS because, as shown in this study, focal FA measurement of the nerve at the flexor retinaculum level showed a significant difference from values at other levels. Under the flexor retinaculum, the position of the median nerve is constant among the flexor tendons. Therefore, the motion of water molecules through the myelin sheath can be affected, causing FA values to vary. By a similar mechanism, CTS also can cause a change in FA values. FA values can aid in follow-up after nerve repair procedures. We found that although there was a difference in FA values, ADCs did not differ significantly in different parts of the median nerve. This finding shows that diffusion anisotropy indexes may be more sensitive than ADC in the evaluation of nerves for pathologic conditions. In the two patients with CTS in this study, we found the mean FA values of human median nerve were 0.41 and 0.44, both 2 SDs below the normal value. In this study, the age-related change in anisotropy indexes found in the median nerve was similar to previously reported age-related anisotropy changes in the brain [17, 18]. Although the oldest subject in our study was only 35 years old, we observed that with increasing age, FA value in normal human median nerves decreased (p < 0.01) and ADC increased (p > 0.05). A decrease in the number of myelinated fibers in the peripheral nerve with advancing age has been reported [19, 20]. A progressive age-related decrease in conduction velocity of the median nerve after the age of 20 years also has been reported [20, 21]. We believe that the changes in FA value and ADC may be associated with the decrease in fiber number with advancing age. These results indicate that patient age may have to be considered in comparing the anisotropy indexes of healthy subjects with those of persons with peripheral nerve disorders. Age-matched control groups may be necessary in such studies. Imaging of the peripheral nerves with DTI is thought to have difficulties such as differentiating the nerves from surrounding structures such as ligaments and muscles. Because muscle fibers are shorter than nerve fibers and have a relatively low anisotropy index, this difficulty can be overcome with use of fiber length and degree of anisot- 926 AJR:189, October 2007
Tractography of Median Nerve TABLE 2: Focal and s for Median Nerve (n = 20) Subject No. Flexor Retinaculum Wrist Forearm 1 0.73 0.56 0.77 0.93 0.76 0.91 2 0.61 0.75 0.78 0.77 0.67 1.05 3 0.52 0.43 0.67 1.05 0.65 1.42 4 0.50 1.16 0.77 0.93 0.70 1.19 5 0.59 0.95 0.77 1.14 0.78 1.16 6 0.55 0.89 0.74 0.86 0.62 0.91 7 0.60 0.61 0.68 0.83 0.64 1.29 8 0.64 0.99 0.76 0.95 0.73 0.98 9 0.60 0.14 0.74 1.19 0.69 0.96 10 0.47 1.26 0.78 0.97 0.75 1.02 11 0.61 1.33 0.71 1.42 0.71 0.91 12 0.58 1.61 0.61 0.41 0.69 0.63 13 0.63 1.35 0.75 0.86 0.77 0.88 14 0.52 0.85 0.63 1.15 0.70 1.24 15 0.73 0.96 0.77 1.00 0.68 1.04 16 0.55 0.89 0.74 0.86 0.62 0.91 17 0.69 0.93 0.78 0.84 0.89 0.84 18 0.58 1.19 0.67 1.37 0.61 0.72 19 0.66 1.36 0.51 0.92 0.87 0.80 20 0.52 1.18 0.72 0.68 0.65 0.74 Mean 0.59 ± 0.07 0.97 ± 0.03 0.72 ± 0.07 0.95 ± 0.02 0.71 ± 0.08 0.98 ± 0.02 ropy [8]. For this purpose, we set the anisotropy threshold at 0.3. The color-coded maps, especially in the coronal plane, combined with anatomic reference views made it easy to localize median nerve tracts. There are several disadvantages to the technique used. The major limitation is a relatively long data acquisition time, which makes the sequence more susceptible to motion artifacts. Especially in postoperative follow-up, image degradation due to motion artifacts can be a serious problem because of the uncomfortable positioning for median nerve visualization. Another limitation is that spatial resolution is low, even with a 3-T MRI system. In our study, the in-plane resolution was 1.09 1.09 mm 2, which can be improved with dedicated wrist coils. In conclusion, we found that in DTI of 20 healthy subjects, the FA value of normal median nerve was 0.709 ± 0.046 and the ADC was 1.016 ± 0.129 10 3 mm 2 /s. Our findings suggest that FA value and ADC can be used in the diagnosis of CTS because focal FA measurement of the nerve at the flexor retinaculum level showed a significant difference from FA values in other parts of the nerve. Age-related changes in anisotropy also should be taken into account in evaluation of the median nerve. References 1. Andreisek G, Crook DW, Burg D, Marincek B, Weishaupt D. Peripheral neuropathies of the median, radial, and ulnar nerves: MR imaging features. RadioGraphics 2006; 26:1267 1287 2. Sallomi D, Janzen DL, Munk PL, Connell DG, Tirman PF. Muscle denervation patterns in upper limb nerve injuries: MR imaging findings and anatomic basis. AJR 1998; 171:779 784 3. Stieltjes B, Kaufmann WE, van Zijl PC, et al. tensor imaging and axonal tracking in the human brainstem. Neuroimage 2001; 14:723 735 4. Zhang S, Laidlow DH, Kindlmann G. tensor MRI visualization. In: Hansen CD, Johnson CR. The visualization handbook. Oxford, UK: Elsevier, 2005:327 340 5. Bammer R. Basic principles of diffusion-weighted imaging. Eur J Radiol 2003; 45:169 184 6. Basser P, Pierpaoli C. Microstructural and physiological features of tissues elucidated by quantitative-diffusion-tensor MRI. J Magn Reson B 1996; 111:209 219 7. Jellison BJ, Field AS, Medow J, et al. tensor imaging of cerebral white matter: a pictorial review of physics, fiber tract anatomy, and tumor imaging patterns. Am J Neuroradiol 2004; 25:356 369 8. Hiltunen J, Suortti T, Arvela S, Seppa M, Joensuu R, Hari R. tensor imaging and tractography of distal peripheral nerves at 3T. Clin Neurophysiol 2005; 116:2315 2323 9. Bendszus M, Stoll G. Technology insight: visualizing peripheral nerve injury using MRI. Nat Clin Pract Neurol 2005; 1:45 53 10. Le Bihan D, van Zijl P. From the diffusion coefficient to the diffusion tensor. NMR Biomed 2002; 15:431 434 11. Raybaud C, Shroff M, Rutka JT, Chuang SH. Imaging surgical epilepsy in children. Childs Nerv Syst 2006; 22:786 809 12. Inoue T, Ogasawara K, Beppu T, Ogawa A, Kabasawa H. tensor imaging for preoperative evaluation of tumor grade in gliomas. Clin Neurol Neurosurg 2005; 107:174 180 13. Meek MF, Stenekes MW, Hoogduin HM, Nicolai JP. In vivo three-dimensional reconstruction of human median nerves by diffusion tensor imaging. Exp Neurol 2006; 198:479 482 14. Skorpil M, Karlsson M, Nordell A. Peripheral nerve diffusion tensor imaging. Magn Reson Imaging 2004; 22:743 745 15. Masutani Y, Aoki S, Abe O, Hayashi N, Otomo K. MR diffusion tensor imaging: recent advance and new techniques for diffusion tensor visualization. Eur J Radiol 2003; 46:53 66 16. Melhem ER, Mori S, Mukundan G, Kraut MA, Pomper MG, van Zijl PC. tensor MR imaging of the brain and white matter tractography. AJR 2002; 178:3 16 17. Abe O, Aoki S, Hayashi N, et al. Normal aging in the central nervous system: quantitative MR diffusiontensor analysis. Neurobiol Aging 2002; 23:433 441 18. Pfefferbaum A, Sullivan EV, Hedehus M, Lim KO, Adalsteinsson E, Moseley M. Age-related decline in brain white matter anisotropy measured with spatially corrected echo-planar diffusion tensor imaging. Magn Reson Med 2000; 44:259 268 19. Vital A, Vital C, Rigal B, Decamps A, Emeriau JP, Galley P. Morphological study of the aging human peripheral nerve. Clin Neuropathol 1990; 9:10 15 20. Wang FC, de Pasqua V, Delwaide PJ. Age-related changes in fastest and slowest conducting axons of thenar motor units. Muscle Nerve 1999; 22:1022 1029 21. Oh SJ. Physiologic factors affecting nerve conduction. In: Oh SJ, ed. Clinical electromyography: nerve conduction studies. Philadelphia, PA: Lippincott Williams & Wilkins, 2003:334 337 AJR:189, October 2007 927