ORIGINAL ARTICLE

Three-dimensional fast imaging employing steady-stateacquisition MRI and its diagnostic value for lumbarforaminal stenosis

Osamu Nemoto • Akira Fujikawa • Atsuko Tachibana

Received: 27 July 2013 / Accepted: 16 November 2013

� Springer-Verlag France 2013

Abstract The aim of this study was to evaluate the use-

fulness of three-dimensional (3D) fast imaging employing

steady-state acquisition (3D FIESTA) in the diagnosis of

lumbar foraminal stenosis (LFS). Fifteen patients with LFS

and 10 healthy volunteers were studied. All patients met the

following criteria: (1) single L5 radiculopathy without

compressive lesion in the spinal canal, (2) pain reproduction

during provocative radiculography, and (3) improvement of

symptoms after surgery. We retrospectively compared the

symptomatic nerve roots to the asymptomatic nerve roots on

fast spin-echo (FSE) T1 sagittal, FSE T2 axial and recon-

stituted 3D FIESTA images. The j values for interobserver

agreement in determining the presence of LFS were 0.525 for

FSE T1 sagittal images, 0.735 for FSE T2 axial images,

0.750 for 3D FIESTA sagittal, 0.733 for axial images, and

0.953 for coronal images. The sensitivities and specificities

were 60 and 86 % for FSE T1 sagittal images, 27 and 91 %

for FSE T2 axial images, 60 and 97 % for 3D FIESTA sag-

ittal images, 60 and 94 % for 3D FIESTA axial images, and

100 and 97 % for 3D FIESTA coronal images, respectively.

3D FIESTA can provide more reliable and additional

information for the running course of lumbar nerve root,

compared with conventional magnetic resonance imaging.

Particularly, use of 3D FIESTA coronal images enables

accurate diagnosis for LFS.

Keywords Three-dimensional fast imaging

employing steady-state acquisition � Lumbar

foraminal stenosis � Magnetic resonance imaging

Introduction

Lumbar foraminal stenosis (LFS) unfortunately results in

failed back surgery syndrome and is the cause of continued

postoperative pain. Magnetic resonance imaging (MRI)

evaluation of lumbar nerve root is routinely performed with

fast spin-echo (FSE) T1 sagittal and FSE T2 axial images,

but inattentive evaluation may fail to discover pathologic

foraminal lesions since it may not clearly demonstrate

nerve root entrapment in the foraminal zone. Several

methods to prevent overlooking a diagnosis of LFS have

been proposed [1–3]. Among them, provocative radicu-

lography and nerve root block have been useful as golden

standard. However, it is invasive and difficult to examine

several levels simultaneously. Therefore, new diagnostic

imaging technique to detect lumbar nerve root entrapment

in the foraminal zone is urgently required.

With recent advantages of MRI in gradient system and

power amplifier technology leading to higher gradient

amplitude and higher slew rates, gradient-echo (GRE)

sequences have become fast and robust. One type of fast

GRE sequence is a steady-state sequence, in which longi-

tudinal magnetization and transverse magnetization are

kept constant with each cycle. Steady-state sequences have

proved to be useful in a variety of applications, including

imaging of the heart and vessels. First described in 1986 as

fast imaging employing steady-state precession, steady-

state-free precession sequences are currently known by

various synonyms: true FISP, balanced fast field echo and

fast imaging employing steady-state acquisition (FIESTA)

[4–7]. Recently, it has been reported that three-dimensional

(3D) FIESTA is useful for demonstrating the anatomy and

pathology of cranial nerves and spinal cord [8, 9]. How-

ever, application of 3D FIESTA to peripheral nerve such as

lumbar nerve root to improve visualization has not been

O. Nemoto (&) � A. Fujikawa � A. Tachibana

Department of Orthopaedic Surgery, Japanese Self Defense

Forces Central Hospital, 1-2-24, Ikejiri, Setagaya-ku,

Tokyo, Japan

e-mail: drsamusio@world.ocn.ne.jp

123

Eur J Orthop Surg Traumatol

DOI 10.1007/s00590-013-1377-9

investigated. The purpose of our study was to evaluate the

efficacy of 3D FIESTA for depicting lumbar nerve root

entrapment in the foraminal zone by comparing to the

conventional MRI sequences.

Materials and methods

From 2009 to 2011, a total of 24 consecutive patients, who

underwent surgical exploration for LFS refractory to con-

servative treatment, were screened for this study. The diag-

nosis was based on the neurological symptoms and a

combination of diagnostic images including radiographs,

computerized tomography, and MRI. Only those patients

who met the following criteria were included—(1) unilateral

single-level L5 radiculopathy without compressive lesions

in the spinal canal, (2) pain reproduction during provocative

radiculography and transient pain relief after nerve root

block for agreement of side and level, and (3) improvement

of symptoms after surgery. Of the 24 patients screened, 15

patients were suitable for inclusion. They consisted of 15

males, with an average age of 51.9 years (range

41–70 years). Transforaminal lumbar interbody fusion with

pedicle screwing system was performed, and the nerve root

entrapment in the foraminal zone was observed at surgery in

all patients. A total of 10 normal volunteers (8 men and 2

women) with a mean age of 49.8 years (range 37–66 years)

were included as controls. At the screening interview, these

normal volunteers did not report leg symptoms that sug-

gested LFS, and they did not show evidence of leg signs

during physical examination.

All volunteers and patients underwent two-dimensional

(2D) MRI with a 1.5-T scanner (Signa infinity Excite; GE

Medical Systems, Milwaukee, WI, USA) using a spine

array coil. Conventional axial, and sagittal MRI used FSE

T1-weighted (TR/TE 550/7.9 ms) and T2-weighted (TR/

TE 4,000/102 ms) sequences with the following parame-

ters: slice thickness 4 mm; slice gap 1 mm; field of view

32 cm for sagittal images and 15 cm for axial images;

matrix 320 9 224; flip angle 90�; excitations 3. Additional

3D FIESTA images were acquired with following acqui-

sition parameters: TR = 6 ms, TE = 1.3 ms, flip

angle = 45�, no intersection gap, slice thickness = 1 mm,

FOV = 18 9 18 cm, matrix = 320 9 320. Multiplanar

image reconstruction was performed using an advantage

workstation (AW 4.3; GE Medical Systems, Milwaukee,

WI, USA), and all 3D images were reconstructed in the 3

planes (sagittal, axial, and coronal). The additional time

taken for the 3D FIESTA acquisition was 5 min. On con-

ventional 2D MRI, LFS] grade 2 was defined as positive,

according to the classification reported by Lee et al. [10].

The evaluation by 3D FIESTA was performed by tracing

the entire running course of the nerve root from intracanal

to extraforaminal on consecutive sliced images. The

marked entrapment of the nerve root with running course

abnormality and surrounding fat obliteration was judged as

positive. The symptomatic L5 nerve roots of the patients

(n = 15) and the asymptomatic L5 nerve roots (contralat-

eral asymptomatic L5 nerve roots in the patients, n = 15;

bilateral L5 nerve roots in the normal volunteers, n = 20)

were evaluated on FSE T1 sagittal, FSE T2 axial, and

reconstituted 3 planes of 3D FIESTA images. The MRI

images were reviewed randomly and independently by an

experienced neuroradiologist and senior spinal surgeon

who did not participate in the care of the patients. They

were aware that images were obtained for the evaluation of

the lumbosacral spinal disorders but were blinded to any

clinical, diagnostic, or treatment-related information and

only asked to decide whether the patients were positive or

negative for LFS. After this independent evaluation was

performed, discrepancies were resolved through discus-

sion, and observations agreed on were used to analyze the

data. The reliability of the evaluation was estimated using

the j statistics for interobserver reliability. The agreement

was calculated using SPSS (ver. 13.0; SPSS Inc., Chicago,

IL, USA) and rated as follows: poor, j = 0–0.2; fair,

j = 0.21–0.4; moderate, j = 0.41–0.6; substantial

j = 0.61–0.8; and excellent, j[ 0.81.

Results

Interobserver reliability in the detection of nerve root

entrapment by MRI is summarized in Table 1. All sequences

except FSE T1 showed optimal reliability. Among them,

excellent agreement was demonstrated on 3D FIESTA

coronal images. Both on conventional 2D MRI and 3D

FIESTA images, exiting nerve root shown as low signal

intensity is normally surrounded by fat tissue and runs

obliquely downward from intracanal to foraminal and ex-

traforaminal zone without angulation. Because of the slice

gap, conventional sagittal and axial MRI images cannot

demonstrate the entire running course of the nerve root

(Fig. 1). In contrast, due to the thin slice thickness without

Table 1 Interobserver agreement in evaluating L5 nerve entrapment

Sequences j value

FSE T1 sagittal 0.525

FSE T2 axial 0.735

3D FIESTA sagittal 0.750

3D FIESTA axial 0.733

3D FIESTA coronal 0.953

FSE indicates fast spin-echo

FIESTA indicates fast imaging employing steady-state acquisition

Eur J Orthop Surg Traumatol

123

slice gap and the high signal-to-noise ratio, 3D FIESTA

sequences can reveal an excellent full view of the nerve root

from foraminal to extraforaminal continuously (Fig. 2).

Diagnostic performance in evaluating L5 nerve root

entrapment using conventional 2D MRI and 3D FIESTA is

presented in Table 2. The sensitivity of FSE T2 axial images

was poor. Although the specificities were high, the sensi-

tivities of 3D FIESTA sagittal, axial, and FSE T1 sagittal

images were suboptimal. Among them, 3D FIESTA coronal

images revealed excellent diagnostic performance of LFS

(Fig. 3).

Discussion

LFS is defined as the narrowing of the bony exit of the

nerve root, and its incidence is reported to be between 8

and 11 % [11]. In general, common and important findings

on MRI images related to the diagnosis of LFS have been

type of stenosis, amount of fat obliteration, and presence of

Fig. 1 Sagittal T1 (a) and axial T2 (b) images of conventional MRI

with a 48-year-old normal volunteer show asymptomatic bilateral L5

nerve roots from intracanal to foraminal zone. Because of the slice

gap, entire running course of L5 nerve root is not clearly

demonstrated

Fig. 2 Consecutive 3D FIESTA coronal images of a 51-year-old

normal volunteer can depict the entire fine view of bilateral L5 nerve

roots (arrow) from intracanal to foraminal zone

Table 2 Detection of L5 nerve root entrapment in the foraminal zone

FSE 3D FIESTA

T1

Sagittal

T2

Axial

Sagittal Axial Coronal

Normal volunteers 3/20 1/20 0/20 0/20 0/20

Patients

Asymptomatic

nerve root

2/15 2/15 1/15 2/15 1/15

Symptomatic

nerve root

9/15 4/15 9/15 9/15 15/15

Sensitivity (%) 60 26 60 60 100

Specificity (%) 86 91 97 94 97

Positive predictive

value

64 57 90 82 94

Negative predictive

value

83 74 85 85 100

FSE indicates fast spin-echo

FIESTA indicates fast imaging employing steady-state acquisition

Eur J Orthop Surg Traumatol

123

nerve root entrapment. However, these findings are occa-

sionally missed since conventional sagittal and axial MRI

images cannot depict the entire running course of the nerve

root because of the slice gap and its poor resolution.

Oblique MRI has been reported to be useful in detecting

LFS [12]. However, oblique images along intervertebral

foramina as well as several MR findings mentioned by Lee

et al. [13] have not been helpful in patients with spinal

deformities such as scoliosis or severe lordosis. Addition-

ally, it takes much time so as to perform bilateral oblique

images in each patient. On the contrary, some authors

reported that conventional MRI images do not detect LFS

with any certainty because false-positive findings may be

frequently observed [14, 15]. There is a possibility that in

cases in which the acquisition of images has not been

parallel or perpendicular to the disc, different sizes of nerve

roots, or an asymmetric position of the nerve roots in cases

without nerve root entrapment might be diagnosed as LFS.

Thus, there are many conflicting reports as to the sensi-

tivity and specificity of the conventional MRI sequences

for the diagnosis of LFS.

Recently, usefulness of MR myelography (MRM) and

diffusion tensor imaging (DTI) as diagnostic tool for LFS

has been reported. MRM was developed in late 1980s and

has become one of popular auxiliary diagnostic methods

used in brain and spinal lesion. MRM has several merits

Fig. 3 Conventional FSE T1 sagittal (a) and FSE T2 axial (b) images

cannot prove definite L5 nerve root entrapment of a 53-year-old male

with left L5/S foraminal stenosis. Coronal 3D FIESTA images

(c) clearly depict the entrapment of left L5 nerve root by bony spur

and bulged disc in the foraminal zone

Eur J Orthop Surg Traumatol

123

over conventional contrast media using myelography: non-

invasiveness, no drug adverse reaction, and so on. In the

evaluation of lumbar nerve root, study results by Kim et al.

[16] and Aota et al. [15] showed that MRM is helpful in

differentiating symptomatic nerve root from asymptomatic

nerve root. Although high specificity was observed in these

reports, the sensitivity of running course abnormality was

65.5 %. Accordingly, Kim et al. [16] described that MRM

alone was not good enough as diagnostic modality for LFS.

Diffusion-weighted imaging has been widely used clini-

cally in the evaluation of the central nervous system.

Several studies have shown that DTI is useful for the

evaluation and visualization of peripheral nerves [17–19].

Eguchi et al. [20] reported that DTI can clearly show

tractograms of lumbar nerves and determine fractional

anisotropy values of the nerve roots in patients and healthy

volunteers. However, that tracts might be apparently

missing in tractograms does not necessarily indicate loss of

nerve fibers. Accordingly, significance of nerve root dis-

ruption on the tractograms is not clearly understood. In

addition, there exist other limitations such as directional

information loss as a result of the partial volume effect,

indefinite relationship between the number of tracts visu-

alized by DTI and actual volume of nerve fiber trajectories,

and so on. Considering these shortcomings, simple and

practical diagnostic modality to evaluate the entrapment of

lumbar nerve root is needed.

Recently, several studies have shown that 3D FIESTA

can provide an excellent visualization of neurovascular

tissues [8, 21, 22]. 3D FIESTA sequence is a fully balanced

steady coherent imaging pulse sequence, and perfectly

balanced gradient at the end of each TR interval of hori-

zontal magnetic realignment phase, which results in neural

structures returning low signal intensity. Attenuation of

transverse magnetization re-gathered at the end of each TR

time creates good T2 contrast in the 3D FIESTA sequence,

and slowly flowing liquid generates high signal intensity.

Additionally, 3D FIESTA has a good signal-to-noise ratio

that allows the evaluation of small structures or lesions. In

this study, to visualize L5 nerve root entrapment, conven-

tional 2D MRI and 3D FIESTA sequences were performed

on 15 surgically treated patients with definite LFS and 10

normal volunteers. Our results show that compared with

conventional 2D MRI, 3D FIESTA imaging could provide

more precise information of the lumbar nerve root

entrapment in the foraminal zone. A key reason for the

difference between conventional 2D MRI and 3D FIESTA

might be distinction in MRI imaging techniques. Using

reconstituted images without slice gap and high spatial

resolution, 3D FIESTA can visualize the continuous full

view of the lumbar nerve root. Particularly, 3D FIESTA

coronal images offer excellent diagnostic performance

compared to sagittal or axial images. Kim et al. [16] also

described the usefulness of coronal views for detecting

foraminal L5 root entrapment in their report with 3D

MRM. Since craniocaudal compression-type LFS is better

visualized on coronal images, it might be affected by its

prevalence in our patient population. Another possibility

may be selection bias of the patient in this retrospective

study. With respect to avoid misdiagnosis of symptomatic

nerve root, provocative radiculography is still valuable. In

patients with multiple involvements of nerve roots on the

different levels, 3D FIESTA cannot identify the symp-

tomatic nerve roots and we have to perform the provocative

radiculography. However, this method is invasive, and it is

often difficult to evaluate the accurate localization of nerve

root entrapment by the leak of contrast media. Therefore, in

case with unilateral single-level radiculopathy, 3D FIESTA

can be an excellent modality for replacing invasive pro-

vocative radiculography.

To strengthen the superiority of the 3D FIESTA coronal

images, further investigations involving larger patient

cohorts are needed. There are several limitations in the

present study. The first limitation is the small sample size

of 15 patients with symptomatic LFS, but the LFS that

meets the criteria in this study is much less common.

Second, we could not repeat the 3D FIESTA following

surgery because of spinal instrumentation artifacts.

In conclusion, 3D FIESTA imaging revealed an excel-

lent full view of the lumbar nerve roots, compared with

conventional MRI. Particularly, 3D FIESTA coronal ima-

ges can provide more reliable information in the diagnosis

of LFS, in addition to helping with surgical treatment

planning. Considering the results of this study, we suggest

that 3D FIESTA should be included in the routine MRI

protocol for the investigation into LFS.

Conflicts of interest None.

References

1. Jenkins JR, Rauch A (1994) Magnetic resonance imaging of

entrapment of lumbar nerve roots in spondylolytic spondylolis-

thesis. J Bone Joint Surg Am 76:1643–1648

2. Segnarbieux F, Van de Kelft E, Candon E, Bitoun J, Frerebeau P

(1994) Disco-computed tomography in extraforaminal and

foraminal lumbar disc herniation: influence on surgical approa-

ches. Neurosurgery 34:643–647

3. Taguchi T, Kawai S, Hashiguchi T (2002) Reassessment of the

diagnostic value of selective lumbosacral radiculography. J Neu-

roradiol 29:122–127

4. Oppelt A, Graumann R, Barfu H, Hartl WS (1986) Fisp-a new

fast MRI sequence. Electromedica 54:15–18

5. Haacke EM, Tkach JA (1990) Fast MR techniques and their

clinical applications. Am J Roentogenol 155:951–964

6. Duerk JL, Lewin JS, Wendt M, Petrsigle C (1988) Remember

true FISP? A high SNR, near 1 second imaging method for T2-

like contrast in interventional MRI at 0.2T. J Magn Reson

Imaging 8:203–208

Eur J Orthop Surg Traumatol

123

7. Chung YC, Merkle EM, Lewin JS, Shomk JR, Duerk JL (1999)

Fast T(2)-weighted imaging by PSIF at 0.2T for interventional

MTI. Magn Reson Med 42:335–344

8. Amemiya S, Aoki S, Ohtomo K (2009) Cranial nerve assessment

in cavernous sinus tumors with contrast-enhanced 3D fast-

imaging employing steady-state acquisition MR imaging. Neu-

roradiol 51:467–470

9. Raval M, Kumari R, Dung AAD, Guglani B, Gupta N, Gupta R

(2010) MRI findings in Hirayama disease. Indian J Radiol

Imaging 20(4):245–249

10. Lee S, Lee JW, Yeom JS et al (2010) A practical MRI grading

system for lumbar foraminal stenosis. Am J Roentogenol

194:1095–1098

11. Jenis L, An H (2000) Spine update: lumbar foraminal stenosis.

Spine 25:389–394

12. Heo DH, Lee MS, Sheen SH, Cho SM, Cho YJ, Oh SM (2009)

Simple oblique lumbar magnetic resonance imaging technique

and its diagnostic value for extraforaminal disc herniation. Spine

22:2419–2423

13. Lee C, Rauschning W, Glenn W (1988) Lateral lumbar spinal canal

stenosis: classification, pathologic anatomy, surgical decompres-

sion. Spine 13:313–320

14. Kunogi J, Hasue M (1991) Diagnosis and operative treatment of

intraforaminal and extraforaminal nerve root compression. Spine

16:1312–1320

15. Aota Y, Niwa T, Yoshikawa K, Fujiwara A, Asada T, Saito T

(2007) Magnetic resonance imaging and magnetic resonance

myelography in the presurgical diagnosis of lumbar foraminal

stenosis. Spine 32:896–903

16. Kim SB, Jang JS, Lee SH (2009) Morphologic changes of L5 root at

coronal source images of MR myelography in cases of foraminal or

extraforaminal compression. J Korean Neurosurg Soc 46:11–15

17. Khalil C, Hancart C, Le Thuc V, Chantelot C, Chechin D, Cotton

A (2008) Diffusion tensor imaging and tractography of the

median nerve in carpal tunnel syndrome: preliminary results. Eur

Radiol 18:2283–2291

18. Lehmann HC, Zhang J, Mori S, Sheikh KA (2010) Diffusion

tensor imaging to assess axonal regeneration in peripheral nerves.

Exp Neurol 223:238–244

19. Takagi T, Nakamura M, Yamada M et al (2009) Visualization of

peripheral nerve degeneration and regeneration monitoring with

diffusion tensor tractography. Neuroimage 44:884–892

20. Eguchi Y, Ohtori S, Orita S et al (2011) Quantitative evaluation

and visualization of lumbar foraminal nerve root entrapment by

using diffusion tensor imaging: preliminary results. Am J Neu-

roradiol 32:1824–1829

21. Zhou Q, Liu ZL, Qu CC, Ni SL, Xue F, Zeng QS (2012) Pre-

operative demonstration of neurovascular relationship in tri-

geminal neuralgia by using 3D FIESTA sequence. J Magn Reson

Imaging 30:666–671

22. Li C, Li Y, Zhang D, Yang Z, Wu L (2012) 3D-FIESTA MRI at

3T demonstrating branches of the intraparotid facial nerve, par-

otid ducts and relation with benign parotid tumours. Clin Radiol

67:1078–7082

Eur J Orthop Surg Traumatol

123