Distortion-free diffusion tensor imaging for evaluation of lumbar nerve roots : Utility of direct coronal single-shot turbo spin-echo diffusion sequence

(1)

Distortion‑free diffusion tensor imaging for evaluation of lumbar nerve roots : Utility of direct coronal single‑shot turbo spin‑echo diffusion sequence

著者坂井上之

著者別表示 SAKAI Takayuki journal or

publication title

博士論文本文Full 学位授与番号 13301甲第5285号

学位名博士（保健学）

学位授与年月日 2021‑03‑22

URL http://hdl.handle.net/2297/00065160

doi: https://doi.org/10.1016/j.mri.2018.01.003

Creative Commons : 表示 ‑ 非営利 ‑ 改変禁止 http://creativecommons.org/licenses/by‑nc‑nd/3.0/deed.ja

(2)

Contents lists available atScienceDirect

Magnetic Resonance Imaging

journal homepage:www.elsevier.com/locate/mri

Original contribution

Distortion-free di ﬀ usion tensor imaging for evaluation of lumbar nerve roots: Utility of direct coronal single-shot turbo spin-echo di ﬀ usion sequence

Takayuki Sakai

^a,b,⁎

, Kunio Doi

^c,d

, Masami Yoneyama

^e

, Atsuya Watanabe

^f,g

, Tosiaki Miyati

^b

, Noriyuki Yanagawa

^a

aDepartment of Radiology, Eastern Chiba Medical Center, 3-6-2 Okayamadai, Togane, Chiba 2838686, Japan

bFaculty of Health Sciences, Institute of Medical, Pharmaceutical and Health Sciences, Kanazawa University, 5-11-80 Kodatsuno, Kanazawa, Ishikawa 9200942, Japan

cThe University of Chicago, 5841 South Maryland Avenue, Chicago, IL 60637, USA

dGunma Prefectural College of Health Sciences, 323-1 Kamioki-machi, Maebashi, Gunma 3710052, Japan

ePhilips Japan, 2-13-37 Konan, Minato-ku, Tokyo 1088507, Japan

fDepartment of General Medical Services, Chiba University, Graduate School of Medicine, 1-8-1 Inohana, Chuo-ku, Chiba 2600856, Japan

gDepartment of Orthopedic Surgery, Eastern Chiba Medical Center, 3-6-2 Okayamadai, Togane, Chiba 2838686, Japan

A R T I C L E I N F O

Keywords:

Diﬀusion tensor imaging Single shot turbo spin echo DWI Lumbar nerve roots

Coronal section Tractography

A B S T R A C T

Purpose:Diﬀusion tensor imaging (DTI) based on a single-shot echo planer imaging (EPI-DTI) is an established method that has been used for evaluation of lumbar nerve disorders in previous studies, but EPI-DTI has problems such as a long acquisition time, due to a lot of axial slices, and geometric distortion. To solve these problems, we attempted to apply DTI based on a single-shot turbo spin echo (TSE-DTI) with direct coronal acquisition. Our purpose in this study was to investigate whether TSE-DTI may be more useful for evaluation of lumbar nerve disorders than EPI-DTI.

Materials and methods:First, lumbar nerve roots ofﬁve healthy volunteers were evaluated for optimization of imaging parameters with TSE-DTI including b-values and the number of motion proving gradient (MPG) directions. Subsequently, optimized TSE-DTI was quantitatively compared with conventional EPI-DTI by using fractional anisotropy (FA) values and visual scores in subjective visual evaluation of tractography. Lumbar nerve roots of six patients, who had unilateral neurologic symptoms in one leg, were evaluated by the optimized TSE- DTI.

Results:TSE-DTI with b-value of 400 s/mm²and 32 diﬀusion-directions could reduce the image distortion compared with EPI-DTI, and showed that the average FA values on the symptomatic side for six patients were signiﬁcantly lower than those on the non-symptomatic side (P < 0.05).

Conclusion:Tractography with TSE-DTI might show damaged areas of lumbar nerve roots without severe image distortion. TSE-DTI might improve the reproducibility in measurements of FA values for quantiﬁcation of a nerve disorder, and would become a useful tool for diagnosis of low back pain.

1. Introduction

Disc herniation and degeneration of the lumbar spine may cause symptoms such as low back pain, which is often associated with leg pain and numbness. Typically, low back pain has been diagnosed with the aid of conventional magnetic resonance imaging (MRI) [1]. How- ever, conventional MRI would not be adequate for evaluation of neurologic symptoms in the legs, because it may not clearly identify pa- thologic extraforaminal lesions or nerve root compression in the extraforaminal area [2]. It is well-known that nerve root entrapment caused by lumbar foraminal stenosis and disc herniation would be a

cause of neurologic symptoms in the legs, which can commonly occur in the extraforaminal area. Therefore, it is important to evaluate patho- logic lesions in the extraforaminal area.

Diffusion tensor imaging (DTI) based on a single-shot echo planer imaging sequence (EPI-DTI) can detect the diffusion of water molecules along nervefibers in neural tissue [3] and is promising for evaluation of lumbar nerve roots compression in the extraforaminal area. Recently, several studies have shown that EPI-DTI and tractography of human lumbar nerves can visualize and quantitatively evaluate lumbar nerves by fractional anisotropy (FA) [4–6]. FA values can be used as a quantitative assessment of diffusion indicating the preferential diffusion of

https://doi.org/10.1016/j.mri.2018.01.003

Received 17 June 2017; Received in revised form 21 January 2018; Accepted 27 January 2018

⁎Corresponding author at: Department of Radiology, Eastern Chiba Medical Center, 3-6-2 Okayamadai, Togane, Chiba 2838686, Japan.

E-mail address:sakai@tkmedical.jp(T. Sakai).

T

(3)

water molecule diﬀusion. Previous studies on EPI-DTI have indicated that FA values are lowered in peripheral nerve compression injuries, which was believed to be due to neurodegeneration, such as widening of the interstitial space [7,8], Wallerian degeneration and axonal de- myelination can result in an increased perpendicular diﬀusion vector compared with normal nerves, and peripheral nerves were quantitatively assessed by single-shot EPI as well as several methods such as readout-segmented EPI and selective-excitation EPI [9], and T2-map- ping [10]. Although EPI-DTI has a great potential for evaluation of lumbar nerve compression, it has several problems such as a long acquisition time, geometric distortion and susceptibility artifacts because of its high sensitivity to magnetic susceptibility.

DWI based on turbo spin echo (TSE-DWI), is originally based on single-shot rapid acquisition with relaxation enhancement (RARE) with phase insensitive preparation [11], has been used to reduce the image distortion and reported for the usefulness in diagnosis of middle-ear cholesteatoma [12]. In fact, TSE-DWI has not been frequently used generally in clinical studies, because it has several drawbacks such as its low signal-to-noise ratio (SNR) theoretically caused by eliminating the magnetization components along the x direction (Mx) signals which include phase errors [11], long acquisition time to suﬃciently gain the SNR, and severe image blurring due to long echo trains. Recently, an improved TSE-DWI sequence has been developed. The improved TSE- DWI sequence has several modiﬁcations from original Alsop's technique such as magnitude signal averaging, a new RF pulse shape that enables shorter echo spacing and compatibility with parallel imaging (sensitivity encoding: SENSE) to shorten the total of TSE echo train that can improve severe image blurring. Thus, we attempted to apply DTI based on a single-shot turbo spin echo sequence (TSE-DTI) to solve the problems of EPI-DTI. Our purpose in this study was to investigate whether TSE-DTI may be useful for evaluation of lumbar nerve root entrapment compared to EPI-DTI.

2. Materials and methods

All subjects were examined by use of a 1.5 T whole-body clinical system (Ingenia, Philips Healthcare, Best, The Netherlands) with a dedicated 32-channelﬂexible torso coil. DTI analyses were made by use of a Zio station2 (AMIN, Tokyo, Japan) for depiction of tractography and measurement of FA values. The institutional review board ap- proved the study, and written informed consents were obtained from all subjects.

2.1. TSE-DTI pulse sequence and DTI protocol

Fig. 1shows two sequence diagrams of the original Alsop's technique (Fig. 1a) and the improved TSE-DWI we used in this study (Fig. 1b). The improved TSE-DWI sequence includes some modiﬁca- tions from original Alsop's technique; image based signal averaging on the magnitude (image) data instead of complex (k-space) data to oﬀset the signal drop owing to phase inversions among multiple acquisitions, a narrower RF pulse shape that enables three-times shorter echo spacing (4.7 ms) compared to original sequence (14 ms), and compatibility with SENSE that leads to shorten data acquisition time and yield high image quality. Consequently, this sequence can improve image blurring and signal un-uniformities.

EPI-DTI was acquired with SENSE with factor of 2 and spectral selective fat suppression. The following imaging parameters were employed; motion proving gradients (MPG) of 15 directions, b-value of 800 s/mm², TR and echo time (TE) of 3000 and 54 ms, axial slice orientation, slice thickness and slice gap of 3.0 and 0 mm,ﬁeld-of-view (FOV) of 320 mm × 253 mm, matrices of 96 × 114, actual voxel size of 3.33 mm × 2.21 mm × 3.0 mm, calculated voxel size of 1.67 mm × 1.67 mm × 3.0 mm, 2 excitations, 50 slices, and an acquisition time of 6 min 36 s.

TSE-DTI was acquired with SENSE with factor of 3 and spectral

selective fat suppression. In addition, to reduce the total acquisition time, we applied to set coronal section. The phase encoding direction was right-left direction and phase-oversampling was applied to prevent aliasing artifacts. The following imaging parameters were employed;

MPG of 32 directions, b-value of 400 s/mm², TR and TE of 3000 and 49 ms, echo train length (ETL) including phase-oversampling of 24, coronal slice orientation, slice thickness and slice gap of 4.0 and 0 mm, FOV of 350 mm × 350 mm, matrices of 88 × 88, actual voxel size of 3.98 mm × 3.98 mm × 4.0 mm, calculated voxel size of 1.56 mm × 1.56 mm × 4.0 mm, 2 excitations, 16 slices, and an acquisition time of 6 min 36 s.

The two imaging datasets (the images series of b = 0 and that in high b-value) were applied image-based distortion correction [13] to avoid the eddy current eﬀects.

2.2. Parameter optimization based on subjective visual evaluation of tractography

Lumbar nerve roots (L4 to S1) of ﬁve healthy volunteers (one woman and four men; median age, 32.6 years; range, 24–45 years) were evaluated to optimize the imaging parameters for TSE-DTI.

In diagnosis of lumbar nerve disorder, it would be important to clearly visualize the lumbar nerves and their pathologies by using DTI and its tractography. To accurately assess the quality of tractography, there are several important points: whether the lumbar nerve roots can be visualized continuously distal to the extraforaminal area, and whether they may be visualized accurately without image distortion. To evaluate such important points fairly, the image quality of tractography was assessed by subjective visual evaluation using a pair comparison method (two alternative-forced choice method: 2AFC) by three trained observers. We compared the quality of tractography using three different b-values (200, 400, 800 s/mm²) and three diﬀerent MPG directions (6, 15, 32 directions). Since the small number of MPG directions and the low b-values basically shorten acquisition times, the actual acquisition times were kept constant by changing the number of signal averages in respective b-values/MPG directions.

Fig. 1.Sequence diagrams of the original Alsop's technique (a) and the improved TSE- DWI we used in this study (b). Improved TSE-DWI is based on the Alsop's phase-insensitive RARE technique which contains additional 90-degree pulse after the MPG and dephasing gradients in My direction to eliminate the magnetization components along the Mx direction signals which include phase errors. Furthermore, a shorter RF pulse shape enables three times shorter echo spacing compared to original sequence.

T. Sakai et al. Magnetic Resonance Imaging 49 (2018) 78–85

79

(4)

Tractography in respective parameters (three different b-values and MPG directions) of thefive volunteers was manually made by one ex- perienced observers in blind manner. To confirm the anatomical structure in putting ROIs for tractography, 3D T2-weighted TSE images (TR/TE of 1500/99 ms, matrices of 256 × 254, FOV of 270 mm × 270 mm, coronal section orientation, slice thickness and gap of 0.75/0 mm, and an acquisition time of 2 min 44 s) were used as a reference, and the ROIs were selected on the proximal dorsal root ganglion (DRG) at the levels of the L4 to S1 nerve roots. All combina- tions of paired images were randomly displayed on the right and left, including the left and right reversed, on a viewing monitor, and an observer was asked to select right or left as a superior image sub- jectively by use of the criteria based on whether lumbar nerve roots can be visualized continuously and accurately without tracking errors.

Superior image quality was given a score of 1, and the other a score of 0. The average total scores in each scan condition were defined as the relative ranking score, and the difference in relative ranking scores was evaluated statistically by steel-dwass test. Values of P < 0.05 were considered to indicate statistical significance.

2.3. Tractography and FA value comparison with conventional EPI-DTI

Subsequently, to verify whether the optimized TSE-DTI can be feasible, we compared the quality of tractography and FA values with conventional EPI-DTI. For comparison of FA values, ROIs were placed both proximally and distally to the lumbar foraminal zone at two levels on the L4 to S1 nerve roots. ROIs of proximal were at the level of middle spinal body in the intraspinal zone, and ROIs of distal were at the level of spinal disc in the extraforaminal zone (Fig. 2) [14]. In addition, SNR was measured by mean signal intensity of the ROIs divided by its standard deviation. ROIs for the SNR measurement were set on the bilateral dorsal root ganglion (DRG) at the levels of the L4 to S1 nerve roots. The visual evaluation of tractography and the diﬀerence in FA values were evaluated by the Wilcoxon test.

2.4. Comparison of image distortion

To demonstrate whether TSE-DTI can reduce the image distortion eﬀectively, we compared the actual image distortion between EPI-DTI and the optimized TSE-DTI. We quantitatively evaluated image distortion using fusion image that was made by the combination of

tractography and 3D T2-weighted TSE images. In the fusion images, position of nerve roots that was visualized by tractography and 3D T2- weighted TSE images misaligned for image distortion. We reconstructed the fusion images in coronal section to in axial section, and measured the actual distance between each center of the nerve roots that were visualized on the tractography and on the 3D T2-weighted TSE images, at the levels of L4, L5, and S1 nerve roots in the axial fusion images.

2.5. Initial clinical feasibility study

Lumbar nerve roots (L4 to S1) of six patients (three women, and three men; median age, 67.1 years; range, 56–78 years) who had unilateral neurologic symptoms in one leg, such as disturbance of per- ception, melosalgia and numbness, were evaluated by optimized TSE- DTI. We obtained tractgraphy with optimized TSE-DTI, and we evaluated the subjective image quality of tractography by the same method as that for the volunteer study. FA values were measured both proximally and distally to the bilateral lumbar foraminal zone at symptomatic levels. The diﬀerence in FA values between the symptomatic side and the non-symptomatic side was evaluated by Wilcoxon test. Values of P < 0.05 were considered to indicate statistical signiﬁcance.

3. Results

Fig. 3shows the representative iso-diffusion-weighted images of a healthy volunteer in several scan conditions including conventional EPI DTI and TSE-DTI with three different b-values. Fig. 4shows the representative tractography images of three different b-values. At a low b- value (200 s/mm²), tractography included several undesirable features such as narrowing, disruption in several parts, and breaking of its continuity. At a high b-value (800 s/mm²), tractography indicated partially thinfibers with a stenotic narrowing, nervefibers different from their original pathway, and mis-tracking of nervefibers. On the other hand, with a b-value of 400 s/mm², the background signal was suppressed, and the signal intensity was maintained at a high level.

Thus, tractography with a b-value of 400 s/mm²indicated lumbar nerve roots well more distally to extraforaminal area compared to other images with different b-values. Regarding the subjective visual evaluation, the average relative ranking score by three observers was 10.40 for a b-value of 200 s/mm², 17.00 for a b-value of 400 s/mm², and 14.60 for a b-value of 800 s/mm². The average relative ranking score Fig. 2.Illustration of the ROIs setting in both proximally and distally to the lumbar foraminal zone. The area between the inner edge of both pedicles was defined as the intraspinal zone, and the area outer to the outer edge of pedicle was defined as the extraforaminal zone.

ROIs of proximal nerves were put on the level of middle spinal body in the intraspinal zone, and ROIs of distal nerves were put on the level of the spinal disc in the extraforaminal zone.

(5)

for a b-value of 400 s/mm²was superior to those for the other b-values, although there was no signiﬁcant diﬀerence in the average ranking scores. Thus, the subjective image quality of tractography was improved with use of a medium b-value.

Fig. 5shows representative tractography images of three different MPG directions. The quality of tractography obtained with a low number of MPG directions shows poor characteristics such as breaking of its continuity and a bilateral difference. An increase in the number of MPG directions improved its continuity in tractography. Regarding the subjective visual evaluation, the average relative ranking score by three observers was 8.667 for 6 directions, 13.67 for 15 directions, and 19.67 for 32 MPG directions. The average relative ranking score in 32 directions was greater than those with other numbers of MPG directions, and there were significant differences between them (P < 0.05).

As shown above regarding the subjective visual evaluation of tractography, optimal scan parameters of TSE-DTI were selected at a b- value of 400 s/mm²and MPG of 32 directions.

Fig. 6illustrates the effect of two different imaging methods on the tractography of lumbar nerve roots of a healthy volunteer. The image quality of tractography for both of imaging methods had good continuity and bilateral symmetry. The average relative ranking score on subjective visual evaluation by three observers was 8.333 with EPI-DTI and 9.667 with the optimized TSE-DTI, which were comparable. There was no significant difference between the two imaging methods.

Table 1shows the effect of the two imaging methods (EPI-DTI and the optimized TSE-DTI) on the average FA values for thefive healthy volunteers. The FA values obtained with TSE-DTI were slightly smaller than those obtained with EPI-DTI but the behavior of the FA values of both techniques in the proximal and distal nerves indicated same trends.Table 2shows the SNR comparison of the EPI-DTI and TSE-DTI with different b-values in thefive healthy volunteers. Lower b-values in TSE-DTI, with b-value of 200 s/mm²and 400 s/mm², indicated higher SNR compared with EPI-DTI.

Table 3shows the effect of two different imaging methods on the average shift length for thefive healthy volunteers. The shift length, as an indicator of distortion, was defined as the distance of nerve roots that were visualized by use of tractography and 3D T2-weighted TSE images. The shift length of EPI-DTI was larger than that of TSE-DTI at

the levels of the L4, L5, and S1 nerve roots (Fig. 7). Especially, the shift length at level L4 with TSE-DTI was signiﬁcantly smaller than that with EPI-DTI. Thus, it is apparent that TSE-DTI can provide lower distortion compared to EPI-DTI.

Fig. 8 shows tractography of lumbar nerve roots in six patients obtained by the optimized TSE-DTI. The white arrows show the points that were suspected of a cause of neurologic disorder on clinicalﬁnd- ings. Tractography of the patients with symptomatic side of lumbar nerve roots exactly indicated abnormalities such as narrowing, deformation, and disruption. The average FA values on the symptomatic and non-symptomatic sides for the six patients by use of TSE-DTI are shown inTable 4. With the TSE-DTI, the FA values for the symptomatic side of lumbar nerve roots were signiﬁcantly lower than those of the non-symptomatic side at the proximal and distal levels (P < 0.05).

4. Discussion

In this study, we introduced a less distortion DTI technique based on improved TSE-DWI sequence. EPI-DTI commonly employs higher b- values (1000 s/mm²) in the central nervous system [15,16] and in the spinal cord [17]. Because the signal intensity in the source images in TSE-DTI is lower than that of EPI-DTI, the use of high b-values might cause the regional discontinuity and mis-tracking of tractography. Al- though the signal intensity of images with use of a low b-value (200 s/

mm²) can be high, the effect of motion probing gradient for diffusion of water molecule would be weak and it can therefore be difficult to visualize the correct course of nerves in TSE-DTI. Thus, we concluded that an optimal b-value for TSE-DTI which showed the best continuity in tractography was 400 s/mm². In this study, the b-values were examined with only three values (200, 400 and 800 s/mm²), but it is unclear which b-value is the best, for example, among b-value of 400, 500 and 600 s/mm². Because basically higher b-value would be more favorable for further clear visualization, it should be further investigated in future work.

Regarding the optimal number of MPG directions for TSE-DTI, an increase in the number of MPG directions improved its continuity and accuracy in tractography. This is because a large number of MPG directions can basically determine the direction of water molecule Fig. 3.Representative iso-diﬀusion-weighted images of a healthy volunteer in several scan conditions. (A: b-0 s/mm², B: b-200 s/mm², C: b-400 s/mm², D: b-800 s/mm², and E: coronal reformat in EPI-DTI).

Fig. 4.Representative images of the diﬀerences of b-values on tractography of lumbar nerve roots of a healthy volunteer. (A: 200 s/mm², B: 400 s/mm², and C: 800 s/mm²).

81

(6)

diﬀusion more accurately [18]. The course of lumbar nerve roots in the extraforaminal area can be complicated due to the compression of disc herniation and spine misalignment. An increase in the number of MPG directions would be useful for visualization of such complicated course of lumbar nerve roots. Thus, we concluded that an optimal number of MPG directions for TSE-DTI was 32 directions.

In comparison of conventional EPI-DTI and optimized TSE-DTI, TSE- DTI showed comparable depictablity of the nerves on tractography and slightly smaller FA values, but the behavior of the FA values of both techniques in the proximal and distal nerves indicated the same trends.

Since the imaging conditions of respective imaging methods are different (EPI-DTI: MPG of 15 directions, b-value of 800 s/mm²; optimized TSE-DTI: MPG of 32 directions, b-value of 400 s/mm²), the FA values are potentially different. Previous studies reported that change of the b- values did not significantly affect the FA values [19], but an increase in either the MPG directions and the SNR leads to the decrease of the FA values [20,21]. TSE-DTI theoretically has low signal intensity but TSE- DWI with b-value of 400 s/mm²we used in this study could keep relatively high signal intensity that thought to be resulted in higher SNR than EPI-DTI with b-value of 800 s/mm². Such SNR differences between two imaging methods might have affected the FA values. Although the FA values of two methods indicated the same trends in this study, we need to further investigate whether the FA values by using TSE-DTI datasets provide useful information as well as EPI-DTI. Furthermore, in a comparison of the image distortion (average shift length), the shift lengths at levels of L4, L5, and S1 nerve roots with TSE-DTI were smaller than those with EPI-DTI. Therefore, our study showed obviously that TSE-DTI can reduce the image distortion, especially the shift length at level L4 was significantly smaller with TSE-DTI than that with EPI- DTI. Theoretically, TSE can minimize the geometric distortion because it is insensitive to magneticfield inhomogeneity compared to EPI [12].

Less distortion with DTI images using TSE-DTI could be matched spa- tially with (anatomical) reference images for generating fusion images.

Such fusion images could be promising for easily generating tractography, but also for accurately measuring FA values. FA measurements on TSE-DTI should be further compared to conventional EPI-DTI because observers can easily recognize the nerves and put the ROIs without mis-registration. We considered that more accurate measurements of FA values and visualization of the tractography might be performed by using TSE-DTI, especially for observers who have relatively little experience in evaluating FA values on FA map or in drawing tractography (radiology residents, radiologists and clinician who are not familiar with lumbar DTI).

From our preliminary clinical results obtained by optimized TSE- DTI, the average FA values on the symptomatic side for the six patients were signiﬁcantly lower than those on the non-symptomatic side.

Additionally, the average FA values at distal levels for the six patients were higher than those at proximal levels. Hence, TSE-DTI would be able to detect nerve disorders quantitatively, as has been done with EPI- DTI. A new eﬀort in this study was the evaluation of the lumbar nerve roots by use of TSE-DTI in coronal sections. With use of TSE-DTI, it would be possible to observe visually damaged areas in tractography, Fig. 5.Representative images of the diﬀerences of the number of MPG directions on tractography of lumbar nerve roots of a healthy volunteer. (A: 6 directions, B: 15 directions, and C: 32 directions).

Fig. 6.Comparison of the tractography of lumbar nerve roots of a healthy volunteer on fusion image of 3D T2-weighted image with EPI-DTI (A) and TSE-DTI (B).

Table 1

Comparison of the average FA values between EPI-DTI and optimized TSE-DTI.

EPI-DTI TSE-DTI

FAProximal 0.466 ± 0.053 0.356 ± 0.065^⁎

FADistal 0.522 ± 0.042 0.412 ± 0.101^⁎

Values are presented as means ± SD.

⁎P < 0.05 compared with EPI-DTI.

(7)

with reduced image distortion, and also improved reproducibility of measurements of FA values for quantiﬁcation of nerve disorders. It would be particularly useful for diagnosing nerve root compression in the extraforaminal area. Lumbar nerve roots were evaluated by use of the FA values in this study. On the other hand, the perpendicular diffusivity may also be a clinically useful value [22]. The relationship between the nerve disorder and the perpendicular diﬀusivity will be investigated in a future study.

Our study has several limitations. Theﬁrst is the low signal intensity associated with TSE-DTI. The measurement accuracy of FA values would be aﬀected by the signal intensity of source images [21,23].

Hence the accuracy of the FA values obtained by TSE-DTI should be further investigated clinically. For maintaining a relatively high signal intensity with TSE-DTI, we employed the medium b-value of 400 s/

mm²in this study. An increase in the signal intensity with TSE-DTI is expected in the future with the use of a high magneticﬁeld machine such as a 3.0 T MRI and/or further improved TSE-based DWI pulse sequence. The increased signal intensity with TSE-DTI would be likely to make several beneﬁts, such as quantitative evaluation of a nerve

disorder with ADC value by use of high b-value, improved accuracy in the determination of FA values, and a reduction of the partial volume effect with high spatial resolution. The second limitation is the small number of patients investigated with TSE-DTI. For understanding the usefulness in the evaluation of lumbar nerve roots and the dependence of patients with TSE-DTI, further studies would be required for in- vestigating whether ourfindings remain valid for a large number of patients. The third limitation is the low spatial resolution associated with TSE-DTI. Although the minimum size of the average shift length was very slight compared with EPI-DTI (0.231 mm as shown in Table 3), an error should be contained from the low spatial resolution settings. The actual voxel size of TSE-DTI we used in this study was 4.0 mm³ whereas the diameter of the mostly lumbar nerve roots is under 4.0 mm². Therefore, the quantitative values in TSE-DTI such as the FA values and the shift length might be inaccurate due to partial volume effect. One possible reason for that the FA values in optimized TSE-DTI were lower than those in EPI-DTI for thefive healthy volunteers (Table 1) is that the measured ROIs in TSE-DTI might be included in adjacent tissues with low diffusion anisotropy (such as bone and muscle). Nevertheless, in some previous study [24–26], it is known that the injured nerves have considerably swelled and there might be no problem with the measurement in that cases. Since the size of S1 nerve roots are > 2 mm in diameter, the FA values might be less affected by the partial volume effect compared to L4 and L5 nerve roots. Therefore, accurate quantification of a nerve disorder would require images with high spatial resolution and without the partial volume effect. Thus, TSE-DTI would be required further investigation for increasing the accuracy in the visualization of lumbar nerve roots in tractography and in the determination of FA values in the future.

Table 2

Comparison of the SNR between EPI-DTI and TSE-DTI of several b-values.

Imaging methods EPI-DTI TSE-DTI TSE-DTI TSE-DTI

b values 800 200 400 800

MPG directions 15 32 32 32

SNR 8.395 ± 1.055 13.46 ± 3.599 11.79 ± 1.761 7.325 ± 0.788

Table 3

Comparison of the average shift length between EPI-DTI and optimized TSE-DTI.

EPI-DTI TSE-DTI

Average shift length (mm) L4 0.813 ± 0.642 0.231 ± 0.212^⁎ L5 0.744 ± 0.408 0.581 ± 0.500 S1 1.363 ± 1.138 1.075 ± 0.976 Values are presented as means ± SD.

⁎P < 0.05 compared with EPI-DTI.

Fig. 7.Comparison of the actual shift length of lumbar nerve roots at the level of L4 to S1 on fusion image of 3D T2-weighted image with EPI-DTI (A) and TSE-DTI (B). In EPI-DTI fusion image, arrow shows misaligned nerve root.

83

(8)

5. Conclusion

We introduced a less distortion DTI technique based on improved TSE-DWI sequence and evaluated the lumbar nerve roots in patients with neurologic symptoms in the legs by optimized TSE-DTI. TSE-DTI with tractography could observe visually damaged areas of lumbar nerve roots, and its FA values could detect nerve disorders. TSE-DTI may improve the reproducibility of FA measurements due to its ro- bustness for image distortion and would be promising for the assessment of low back pain.

Acknowledgments

The authors are grateful to Ryu Emura (Department of Radiology, Kimitsu Chuo Hospital) for assistance in preparation of the software program for subjective visual evaluation.

References

[1] Boden SD, Wiesel SW, et al. Lumbar spine imaging: role in clinical decision making.

J Am Acad Orthop Surg 1996 Oct;4(5):238–48.

[2] Heo DH, Lee M, Sheen SH, Cho SM, Cho YJ, Oh SM. Simple oblique lumbar magnetic resonance imaging technique and its diagnostic value for extraforaminal disc herniation. Spine 2009;34:2419–23.

[3] Mukherjee P, Chung SW, Berman JI, Hess CP, Henry RG. Diﬀusion tensor MR imaging andﬁber tractography: technical considerations. Am J Neuroradiol 2008;29:843–52.

[4] Balbi V, Budzik JF, Duhamel A, Bera-Louville A, LeThuc V, Cotton A. Tractography of lumbar nerve roots: initial results. Eur Radiol 2011;21:1153–9.

[5] Eguchi Y, Ohtori S, Orita S, et al. Quantitative evaluation and visualization of lumbar foraminal nerve root entrapment by using diﬀusion tensor imaging: preliminary results. Am J Neuroradiol 2011;32:1824–9.

[6] van der Jagt PK, Dik P, Froeling M, et al. Architectural conﬁguration and

microstructural properties of the sacral plexus: a diﬀusion tensor MRI andﬁber tractography study. NeuroImage 2012 Sep;62(3):1792–9.

[7] Khalil C, Budzik JF, Kermarrec E, Balbi V, Le Thuc V, Cotten A. Tractography of peripheral nerves and skeletal muscles. Eur J Radiol 2010;76:391–7.

[8] Andreisek G, White LM, Kassner A, Sussman MS. Evaluation of diﬀusion tensor imaging andﬁber tractography of the median nerve: preliminary results on in- trasubject variability and precision of measurements. Am J Roentgenol 2010;194:W65–72.

[9] Manoliu A, Ho M, Nanz D, et al. Diﬀusion tensor imaging of lumbar nerve roots:

comparison between fast readout-segmented and selective-excitation acquisitions.

Investig Radiol 2016 Aug;51(8):499–504.

[10] Karampinos DC, Melkus G, Shepherd TM, et al. Diﬀusion tensor imaging and T2 relaxometry of bilateral lumbar nerve roots: feasibility of in-plane imaging. NMR Biomed 2013 Jun;26(6):630–7.

[11] Alsop DC. Phase insensitive preparation of single-shot RARE: applications to diffusion imaging in humans. Magn Reson Med 1997;38:527–33.

[12] De Foer B, Vercruysse JP, Pilet B, et al. Single-shot, turbo spin-echo, diﬀusion- weighted imaging versus spin-Echo-planar, diﬀusion-weighted imaging in the de- tection of acquired middle ear cholesteatoma. Am J Neuroradiol 2006;27:1480–2.

[13] Netsch T, van Muiswinkel A. Quantitative evaluation of image-based distortion correction in diﬀusion tensor imaging. IEEE Trans Med Imaging 2004;23:789–98.

[14] Oikawa Y, Eguchi Y, et al. Diﬀusion tensor imaging of lumbar spinal nerve in subjects with degenerative lumber disorders. Magn Reson Imaging 2015;33:956–61.

[15] Andrade CS, Figueiredo KG, Valeriano C, et al. DTI-based tractography of the arc- uate fasciculus in patients with polymicrogyria and language disorders. Eur J Radiol 2015;84:2280–6.

[16] N Gajawelli, Lao Y, Apuzzo ML, et al. Neuroimaging changes in the brain in contact versus noncontact sport athletes using diﬀusion tensor imaging. World Neurosurg 2013;80:824–8.

[17] Czernicki T, Maj E, et al. Diﬀusion tensor tractography of pyramidal tracts in patients with brainstem and intramedullary spinal cord tumors: relationship with motor deﬁcits and intraoperative MEP changes. J Magn Reson Imaging 2017;46(3):715–23.

[18] Barrio-Arranz G, de Luis-Garcia R, Tristan-Vega A, Martin-Fernandez M, Aja- Fernandez S. Impact of MR acquisition parameters on DTI scalar indexes: a tractography based approach. PLoS One 2015 Oct 12;10(10):e0137905.

[19] Andreisek G, White LM, Kassner A, et al. Diﬀusion tensor imaging andﬁber tractography of the median nerve at 1.5 T: optimization of b value. Skelet Radiol 2009;38(1):51–9.

[20] Choi SI, Kang JW, Chun EJ, et al. High-resolution diffusion tensor MR imaging for evaluating myocardial anisotropy andfiber tracking at 3 T: the effect of the number of diffusion-sensitizing gradient directions. Korean J Radiol 2010;11(1):54–9.

[21] Farrell JA, Landman BA, Jones CK, et al. Effects of signal-to-noise ratio on the accuracy and reproducibility of diffusion tensor imaging-derived fractional anisotropy, mean diffusivity, and principal eigenvector measurements at 1.5 T. J Magn Reson Imaging 2007;26:756–67.

[22] Bennett IJ, Madden DJ, Vaidya CJ, et al. Age-related diﬀerences in multiple mea- sures of white matter integrity: a diﬀusion tensor imaging study of healthy aging.

Hum Brain Mapp 2010;31(3):378–90.

[23] Qin W, Yu CS, Zhang F, et al. Effects of echo time on diffusion quantification of brain white matter at 1.5 T and 3.0 T. Magn Reson Med 2009;61:755–60.

[24] Hiwatashi A, Togao O, Yamashita K, et al. Lumbar plexus in patients with chronic Fig. 8.Tractography of lumbar nerve roots from six patients by means of TSE-DTI. Nerve traces show abnormalities such as narrowing (dotted arrows), deformation (arrows), and disruption (arrowhead).

Table 4

Comparison of the average FA values between symptomatic side and non-symptomatic side of the patients acquired by optimized TSE-DTI.

Symptomatic side Non-symptomatic side

FAProximal 0.254 ± 0.057 0.406 ± 0.130^⁎

FADistal 0.320 ± 0.085 0.438 ± 0.125^⁎

⁎P < 0.05 compared with symptomatic side.

(9)

inﬂammatory demyelinating polyneuropathy: evaluation with 3D nerve-sheath signal increased with inked rest-tissue rapid acquisition of relaxation enhancement imaging (3D SHINKEI). Eur J Radiol Aug 2017;93:95–9.

[25] Goedee HS, Jongbloed BA, van Asseldonk JH, et al. A comparative study of brachial plexus sonography and magnetic resonance imaging in chronic inﬂammatory

demyelinating neuropathy and multifocal motor neuropathy. Eur J Neurol Oct 2017;24(10):1307–13.

[26] Nakamichi K, Tachibana S. Enlarged median nerve in idiopathic carpal tunnel syndrome. Muscle Nerve Nov 2000;23(11):1713–8.

85