Magnetic Resonance Imaging of the Peripheral Nerves and Fascicles of the Knee Using Double Echo Steady State Sequence at 7 Tesla

Abstract

Objective

To evaluate the applicability of the double echo steady state (DESS) sequence at 7 tesla (7T) for high-resolution imaging of the peripheral nerves and fascicles of the knee.

Materials and Methods

We prospectively included 32 healthy participants (mean age 39 ± 14 years, 20 females). The patients underwent 7T magnetic resonance imaging (MRI) of the knee using proton density turbo spin-echo fat suppression (PD-TSE FS), three-dimensional DESS (3D-DESS), and higher in-plane resolution DESS (DESS_HR) sequences. The signal-to-noise ratios (SNRs) of the peroneal nerve (PN) and tibial nerve (TN) and contrast-to-noise ratios (CNRs) between the nerves and adjacent fat, vessels, and muscles were quantitatively measured by two readers and averaged. Five radiologists qualitatively assessed the overall image quality, pulsatile flow artifacts, and visualization of the PN and its branches, the TN, and the saphenous nerve (SN) using a five-point Likert-type scale, with the results averaged. The results of the three image sequences were compared.

Results

The SNR for the TNs in the DESS_HR sequence were lower than those in the PD-TSE FS (P < 0.001) and 3D-DESS (P = 0.024) sequences, whereas the SNR for the PNs did not differ significantly across the three sequences. The DESS_HR sequence exhibited superior TN- or PN-to-fat and PN-to-muscle CNR values when compared with the PD-TSE FS and 3D-DESS sequences (P ≤ 0.016). The TN- and PN-to-vessel CNR values in the DESS_HR and PD-TSE FS sequences were higher than those in the 3D-DESS sequence (P ≤ 0.001). Qualitative assessments revealed fewer pulsatile artifacts in 3D-DESS than in DESS_HR and PD-TSE FS (P < 0.001), with DESS_HR exhibiting fewer artifacts than PD-TSE FS (P = 0.035). DESS_HR excelled in visualizing the common PN, TN, and SN when compared with other sequences (P < 0.001), whereas 3D-DESS provided superior visualization of PN branches when compared with other sequences (P ≤ 0.042).

Conclusion

The DESS sequence at 7T MRI enhances visualization of peripheral nerves and fascicular structures around the knee.

INTRODUCTION

The peroneal nerve (PN), tibial nerve (TN), and saphenous nerve (SN) are the primary peripheral nerves located in the vicinity of the knee joint and are vulnerable to injury from external forces [1,2,3] or surgical interventions, such as medial arthrotomy, meniscectomy, arthroscopic anterior cruciate ligament repair, and total knee replacement [4]. Damage to these nerves can cause various clinical symptoms such as pain, numbness, and impaired lower-extremity function [5]. Although ultrasonography is commonly employed for imaging peripheral nerves [6], its limited penetration makes it ineffective for visualizing small or deep-seated nerves, such as the sural and musculocutaneous or deeper branches of the PN, TN, and SN [7].

Magnetic resonance imaging (MRI) is a valuable tool for examining peripheral nerves and provides detailed information on neuropathy and the surrounding tissues [8,9]. Although 3 tesla (3T) MRI has been effective in observing larger nerves such as the common PN, TN, and sciatic nerve [10,11,12,13,14], it remains inadequate for smaller nerves and fascicles because of its limited spatial resolution. Our preliminary study also revealed suboptimal image quality for visualizing the SN, common PN at the fibula neck, as well as smaller branches of the peripheral nerve using proton density (PD) and double echo steady state (DESS) sequences at 3T MRI (Supplement, Supplementary Table 1, Supplementary Figs. 1-3). In contrast, 7T MRI offers a higher signal-to-noise ratio (SNR) and resolution [15], enabling more precise visualization of minute anatomical structures.

The DESS sequence, which incorporates water excitation technology for fat suppression (FS) [16], enhances the distinction between fat and other anatomical structures [17,18]. This sequence has been used to image the alveolar nerve [19], facial nerve [20], lumbosacral plexus [21], and upper extremity peripheral nerves [22] at 3T MRI. Additionally, previous studies have demonstrated its utility for visualizing the cervical nerve systems at 7T MRI [23,24]. However, the application of the three-dimensional (3D)-DESS sequence at 7T for visualizing the knee peripheral nerves remains unexplored. A comprehensive investigation into the utilization of this sequence is essential to maximize the capabilities of 7T MRI for visualizing the peripheral nerves and fascicular structures around the knee.

Peripheral nerves consist of numerous nerve bundles enveloped by connective tissue membranes, small blood vessels, and adipose tissue [25]. A previous study [26] employed the PD turbo spin-echo FS (PD-TSE FS) sequence at 7T to determine the axonal fascicle count in the median nerve. However, spin-echo imaging at 7T was compromised by ghosting artifacts from pulsatile flow, which hindered nerve visualization. Furthermore, the small size of the knee peripheral nerve bundles requires a high in-plane spatial resolution that is crucial for depicting fascicles and identifying pathological changes [27]. Thus, the objective of this study was to assess the applicability of the 7T DESS sequence for high-resolution imaging of the peripheral nerves and fascicular structures around the knee.

MATERIALS AND METHODS

Participants

This prospective study was approved by the Institutional Review Board of the local hospital (IRB No. KY2022147), and all participants provided written informed consent. From July 2022 to January 2024, healthy participants who met the inclusion and exclusion criteria were continuously enrolled based on the order of registration. The inclusion criteria were: healthy adults aged ≥20 years with a BMI of 18.5–28.0 kg/m² and no contraindications for MR examination. The exclusion criteria were: muscle disease; metabolic diseases; severe infections; heart, lung, liver, and kidney dysfunction; and life-threatening diseases or serious unstable underlying conditions posing a high risk for trial safety. The study included 38 healthy participants. Two participants who lacked PD-TSE FS images and four who showed severe motion artifacts were excluded from this study.

The final analysis included 32 participants.

MRI Acquisition

MR neurography data were prospectively acquired using a 7T scanner (MAGNETOM Terra, Siemens Healthineers, Erlangen, Germany) equipped with a 28-channel knee coil. The MRI protocols (Table 1) included acquisition of axial PD-TSE FS sequences and two three-dimensional (3D)-DESS sequences with different resolutions. Specifically, a sagittal DESS sequence with an isotropic resolution of 0.4 mm was employed for 3D structural evaluation, which was then reconstructed to the axial plane, and an axial DESS sequence with a resolution of 0.17 × 0.17 × 2 mm³ (DESS_HR) was used to assess nerve fascicles. To standardize imaging protocols, all participants underwent MRI of the right knee.

Table 1. MRI protocols at 7T.

Parameter	PD-TSE FS	3D-DESS	DESSHR
Image plane	Axial scan	Isotropic sagittal scan, with reconstruction into axial plane	Axial scan
Repetition time, msec	3200	15	15
Echo time, msec	32	4.48	4.48
Flip angle, °	180	28	20
Bandwidth, Hz/Px	139	250	249
FOV, mm2	160 × 160	158 × 158	158 × 158
Averages	2	1	1
Matrix	800 × 800	400 × 400	912 × 912
Resolution, mm3	0.2 × 0.2 × 2.5	0.4 × 0.4 × 0.4	0.17 × 0.17 × 2
Fat suppression	FS	WE	WE
Tubor factor	5	-	-
Acceleration factor	2	3	3
Partial Fourier factor	Off	Off	Off
Scan percentage, %	100	100	100
Number of slices	30	208	60
Time	8 min 40s	8 min 14s	8 min 2s

MRI = magnetic resonance imaging, 7T = 7 tesla, PD = proton density, TSE = turbo spin-echo, FS = fat suppression, 3D = three-dimensional, DESS = double echo steady state, DESS_HR = means DESS sequence with a high in-plane resolution, FOV = field of view, WE = water excitation

During the testing phase, we fine-tuned the sampling perfection with application-optimized contrasts using different flip angle evolution (SPACE) and fast low-angle shot (FLASH) alongside DESS with an isotropic resolution of 0.4 mm³ at 7T for imaging the peripheral nerves of the knee. Detailed images of these sequences are shown in Supplementary Figure 4. We found that the DESS sequence provided superior delineation of the PN and TN branches and effectively distinguished nerves and adjacent vessels when compared with the other sequences. Consequently, we selected DESS sequences for this study.

Imaging Analysis

The SNR of the nerves and contrast-to-noise ratios (CNRs) of nerve-to-fat, -vessel, and -muscle were calculated using region of interest (ROI) placed within the TN, PN, adjacent fat, popliteal vein, and gastrocnemius muscle on axial source images at the level of the popliteal fossa, as shown in Figure 1. These calculations were based on the mean signal intensity (SI) and standard deviation (SD), as per Equations (1) and (2):

$$\text{SNR}=\frac{\text{Mean}\left({\text{SI}}_{\text{nerve}}\right)}{\text{Mean SD}\left({\text{SI}}_{\text{air from four ROIs at the corners of the image}}\right)}$$ (1)

$$\text{CNR}=\frac{\text{Mean}\left({\text{SI}}_{\text{nerve}}\right)-\text{Mean}\left({\text{SI}}_{\text{surrounding tissue}}\right)}{\text{SD}\left({\text{SI}}_{\text{surrounding tissue}}\right)}$$ (2)

Fig. 1. Definition of ROIs for signal-to-noise ratio and contrast-to-noise ratio evaluation at the popliteal fossa level. The inner area of each ROI was evaluated, excluding the edges. Specifically, ROIs 1 to 4 measure noise levels. ROIs 5 and 6 measure tibial and peroneal nerves. ROIs 7 and 8 measure adjacent fat tissue, and ROI 9 measures the popliteal vein. ROIs 10 and 11 measure the gastrocnemius muscle signals. ROI = region of interest.

In addition, because some nerve-to-muscle CNR values were negative during the calculation, their absolute values were used to accurately represent the contrast magnitude ensuring consistent nerve-muscle differentiation assessment. The SNR and CNR values for the SN were not measured due to its small size, which resulted in insufficient visualization of certain 7T sequences. All ROIs of all participants were independently drawn by two radiologists (C.P.Z. and Z.Z.M., with 5 and 8 years of experience in radiology, respectively), and results from both measurements were used for further analysis.

A group of five radiologists (J.L., T.T.Y., W.J.H., M.Z., and M.S.D.), with 5–14 years of musculoskeletal MRI experience independently assessed the images without knowledge of the acquisition methodologies. Before evaluation, all patients received detailed instructions and training from a senior musculoskeletal radiologist (W.C., with 30 years of experience). The radiologists evaluated the overall image quality, pulsatile flow artifacts, and nerve segment visualization on axial source images spanning from the popliteal fossa to the PN branching level using a five-point Likert-type scale [28,29], as shown in Figure 2 and Table 2.

Fig. 2. A comparison of axial PD-TSE FS, 3D-DESS, and DESS_HR images at 7T MR in a 29-year-old healthy female, focusing on peripheral nerves at different knee levels. PN (label 1), tibial nerve (label 2), saphenous nerve (label 3), lateral sural cutaneous nerve (label 4), and tibial nerve muscular branches (label 5). PD = proton density, TSE = turbo spin-echo, FS = fat suppression, 3D = three-dimensional, DESS = double echo steady state, DESS_HR = means DESS sequence with a high in-plane resolution, 7T = 7 tesla, MR = magnetic resonance, PN = peroneal nerve.

Table 2. Criteria for the 5-point Likert scale gradings.

Variables	5-point Likert scale
Overall image quality	1. Non-diagnostic quality 2. Poor image quality 3. Acceptable image quality 4. Good image quality 5. Excellent image quality enables precise differentiation between muscle tissue and adjacent fascia/septum. The muscle signal is consistently homogeneous across the visual field
Pulsatile flow artifacts	1. Severe pulsatile flow artifacts: these artifacts severely hinder the evaluation of all nerve structures 2. Visible pulsatile flow artifacts with high signal intensity: these artifacts moderately interfere with the assessment of two adjacent nerve types 3. Moderate pulsatile flow artifacts: these artifacts slightly obstruct the assessment of adjacent nerves 4. Visible pulsatile flow artifacts with mild signal intensity: these artifacts do not impede the evaluation of adjacent nerves 5. None
Visualization of the tibial nerve and the common peroneal nerve	1. Non-diagnostic: characterized by extensive blurring that precludes accurate interpretation 2. Poor: nerve borders and anatomical details are poorly visualized, making clear delineation difficult 3. Moderate: nerve borders are visible with some blurring into surrounding tissues; individual nerve fascicles appear blurred and indistinguishable from one another 4. Good: nerve borders are well visualized with minimal blurring; anatomical depiction includes clear visualization of several individual nerve fascicles 5. Excellent: nerve borders and fascicular structures are sharply delineated, allowing for clear differentiation between individual fascicles
Visualization of the branches of the peroneal nerve	1. Non-diagnostic: characterized by extensive blurring that precludes accurate interpretation 2. Poor: the deep peroneal nerve, superficial peroneal nerve, articular branch, and muscular branch nerve borders are poorly visualized 3. Moderate: the deep peroneal and superficial peroneal nerve borders are moderately visible with some blurring into surrounding tissues; the articular branch and muscular branch nerves are not distinguishable 4. Good: the deep peroneal and superficial peroneal nerve borders are well visualized; however, the articular branch and muscular branch nerves exhibit some blurring 5. Excellent: all branches of the peroneal nerve, including the deep peroneal, superficial peroneal, articular branch, and muscular branch nerves, are clearly and perfectly delineated
Visualization of the saphenous nerve	1. None: no visualization of the saphenous nerve 2. Poor: visualization is hindered by extensive blurring, making the saphenous nerve indistinct 3. Moderate: the borders of the saphenous nerve are moderately visible, but the distal saphenous nerve cannot be identified 4. Good: both proximal and distal saphenous nerve borders are well visualized with minor blurring 5. Excellent: the saphenous nerve borders and fascicular structures are clearly and perfectly delineated

For evaluation, the PN was anatomically separated into a proximal segment (common PN) and a distal segment (branches of the PN, including the deep and superficial PNs and articular and muscular branches). The results from the five readers were averaged for further statistical analyses to compare the different sequences.

Statistical Analysis

The normality of the data distribution was verified using the Kolmogorov-Smirnov test. The summary of the data was presented as mean ± SD or median with interquartile range, accordingly. The SNR values of the nerves and the CNR values of the nerves-to-fat and -vessels exhibited a normal distribution. Therefore, a one-way repeated-measures ANOVA test was conducted to compare these measurements across the three sequences, with a post-hoc Bonferroni test used for multiple comparisons. The CNR values of the nerves-to-muscle and the qualitative measurements resulting from the five-point Likert scale did not conform to a normal distribution. Therefore, the differences in these measurements across the three sequences were evaluated using the Friedman test, followed by the pairwise post-hoc Wilcoxon test with Bonferroni-Holm correction to correct for multiple comparisons.

The interrater agreement between the two readers in measuring the SNR and CNR values was calculated using intraclass correlation coefficients (ICCs) employing a two-way mixed absolute agreement model. The interrater agreement among the five readers for qualitative measurements was calculated using Fleiss Kappa [30,31]. Kappa values of 0.41–0.60, 0.61–0.80, and 0.81–1.0 indicated moderate, substantial, and almost perfect interrater agreement, respectively.

Statistical significance was set a tP < 0.05, and computations were conducted using IBM SPSS Statistics for Windows, version 26.0. (IBM Corp., Armonk, NY, USA). Graphs were generated using GraphPad Prism, version 10 (GraphPad Software Inc., La Jolla, CA, USA).

RESULTS

Participant Characteristics

This study ultimately included 32 healthy individuals (20 females, 12 males) with a mean age of 39 ± 14 years and a BMI of 22.2 ± 2.6 kg/m².

Interrater Reliability

The ICCs for the SNR and CNR values ranged from 0.897–0.994. The interrater agreement for assessing qualitative measurements across all five readers ranged from 0.644 to 0.813 (Table 3).

Table 3. The interobserver agreement results of SNR, CNR, and qualitative data.

Quantitative data	ICC	95% CI	P
SNR	0.994	0.991–0.996	<0.001
CNR nerve to fat	0.954	0.904–0.974	<0.001
CNR nerve to vessel	0.897	0.824–0.935	<0.001
CNR nerve to muscle	0.937	0.828–0.970	<0.001
Qualitative data	Fleiss kappa	95% CI	P
Overall image quality	0.751	0.749–0.753	<0.001
Pulsatile flow artifacts	0.813	0.811–0.814	<0.001
CPN visualization	0.744	0.742–0.745	<0.001
BPN visualization	0.644	0.643–0.646	<0.001
TN visualization	0.743	0.742–0.744	<0.001
SN visualization	0.788	0.787–0.789	<0.001

SNR = signal-to-noise ratio, CNR = contrast-to-noise ratio, ICC = intraclass correlation coefficient, CI = confidence interval, CPN = common peroneal nerve, BPN = branches of the peroneal nerve, TN = tibial nerve, SN = saphenous nerve

SNR and CNR Values

The SNR and CNR values for the TN and PN were compared across the different sequences (Fig. 2, Table 4). The SNR for the TN in the DESS_HR sequence was lower than that in the PD-TSE FS (P < 0.001) and 3D-DESS (P = 0.024) sequences. We detected no significant differences in the SNR of the PN across all sequences (P = 0.387).

Table 4. The SNR value of peripheral nerves and the CNR values of peripheral nerves to tissue in different sequences at 7T MRI.

		PD-TSE FS	3D-DESS	DESSHR	3D-DESS vs. PD-TSE FS	DESSHR vs. PD-TSE FS	DESSHR vs. 3D-DESS
SNR
	TN	64.7 ± 17.0	57.1 ± 15.2	48.2 ± 15.1	0.036*	<0.001*	0.024*
	PN	71.9 ± 19.3	74.7 ± 18.0	69.9 ± 21.1	NS*	NS*	NS*
CNR nerve-fat
	TN	5.6 ± 2.6	4.8 ± 2.4	13.0 ± 5.1	0.720*	<0.001*	<0.001*
	PN	7.7 ± 2.6	8.2 ± 3.9	17.8 ± 6.4	1*	<0.001*	<0.001*
CNR nerve-vessel
	TN	15.4 ± 6.9	6.1 ± 4.7	11.3 ± 4.7	<0.001*	0.014*	0.001*
	PN	17.5 ± 8.6	9.7 ± 5.7	18.0 ± 6.7	<0.001*	1*	<0.001*
CNR nerve-muscle
	TN	1.9 (1.3, 3.3)	2.5 (1.0, 4.1)	2.2 (1.1, 3.4)	NS†	NS†	NS†
	PN	1.8 (1.1, 3.3)	2.3 (1.7, 2.91)	4.0 (2.2, 6.4)	1†	0.010†	0.016†

Data is mean ± standard deviation or median (interquartile range).

^*Means of the repeated-measures ANOVA test with post hoc analysis using the Bonferroni test were employed, ^†Means of the Friedman test with pairwise post hoc Wilcoxon test with Bonferroni-Holm correction were employed. P < 0.05 means a significant difference. With the P-values of 0.387 for SNR (from repeated-measures ANOVA) and 0.552 for CNR nerve-muscle (from the Friedman test).

SNR = signal-to-noise ratio, CNR = contrast-to-noise ratio, 7T = 7 tesla, MRI = magnetic resonance imaging, PD = proton density, TSE = turbo spin-echo, FS = fat suppression, 3D = three-dimensional, DESS = double echo steady state, DESS_HR = means DESS sequence with a high in-plane resolution, TN = tibial nerve, PN = peroneal nerve, NS = not significant

Regarding the CNR of nerves-to-fat, the CNRs of TN-to-fat and PN-to-fat of the DESS_HR sequence were higher than those of the PD-TSE FS and 3D-DESS sequences (P < 0.001). Regarding the CNR of nerves-to-vessels, the CNRs of TN-to-vessel and PN-to-vessel of the PD-TSE FS sequence were higher than those of the 3D-DESS (P < 0.001). Moreover, the CNR of TN-to-vessel of the PD-TSE FS sequence was higher than that of the DESS_HR sequence (P = 0.014). However, the DESS_HR and PD-TSE FS sequences showed no significant differences in the CNR of the PN-to-vessel (P > 0.99). The CNRs of the TN-to-vessel and PN-to-vessel of the DESS_HR sequence were higher than those of the 3D-DESS sequence (P < 0.001). And with respect to the CNR of nerves-to-muscle, the DESS_HR sequence provided higher CNR of PN-to-muscle than the other sequences (P = 0.010 vs. PD-TSE FS, P = 0.016 vs. 3D-DESS). No significant differences were found between the other nerve-to-muscle CNR values (P > 0.05).

Qualitative Analysis

Qualitative analysis compared the three sequences at 7T MR (Fig. 3, Table 5). The 3D-DESS and DESS_HR sequences demonstrated superior image quality when compared with the PD-TSE FS sequence (P < 0.001), with no significant difference observed between the 3D-DESS and DESS_HR sequences (P > 0.99). The 3D-DESS sequence exhibited fewer pulsatile flow artifacts than the PD-TSE FS and DESS_HR sequences (P < 0.001), and the DESS_HR sequence presented fewer artifacts than the PD-TSE FS sequence (P = 0.035). Regarding nerve segment visualization (Fig. 4), the DESS_HR sequence showed superior visualization of the CPN, TN, and SN when compared with the PD-TSE FS and 3D-DESS sequences (P < 0.001). The PD-TSE FS sequence showed superior TN and SN visualization when compared with the 3D-DESS sequence (P < 0.001), without a significant difference in CPN visualization (P = 0.280). However, branches of the peroneal nerve (BPN) were better visualized in the 3D-DESS sequence than in the PD-TSE FS (P < 0.001) and DESS_HR (P = 0.042) sequences (Fig. 5).

Fig. 3. Box and whisker plots summarize the SNR and CNR values across PD-TSE FS, 3D-DESS, and DESS_HR sequences. Whiskers indicate minimum and maximum values. The repeated-measures ANOVA test compared the SNR and the CNR of nerves-to-fat and nerves-to-vessel across three sequences. The Friedman test compared the CNR of nerves-to-muscle. Statistical significance is indicated on a level of ^*P < 0.05, ^†P ≤ 0.001. All given P-values are adjusted for multiple comparisons. SNR = signal-to-noise ratio, CNR = contrast-to-noise ratio, PD = proton density, TSE = turbo spin-echo, FS = fat suppression, 3D = three-dimensional, DESS = double echo steady state, DESS_HR = means DESS sequence with a high in-plane resolution, TN = tibial nerve, ns = not significant, PN = peroneal nerve.

Table 5. Qualitative analysis for PD-TSE FS, 3D-DESS, DESS_HR sequences at 7T MRI as assessed by five readers.

Sequence	PD-TSE FS	3D-DESS	DESSHR	P-values
				3D-DESS vs. PD-TSE FS	DESSHR vs. PD-TSE FS	DESSHR vs. 3D-DESS
Overall image quality	4.2 (4.0–4.8)	5 (5.0–5.0)	5 (5.0–5.0)	<0.001	<0.001	1.000
Pulsatile flow artifacts	4 (3.2–4.0)	5 (5.0–5.0)	4 (4.0–4.0)	<0.001	0.035	<0.001
CPN visualization	4 (4.0–4.2)	4 (3.8–4.0)	5 (4.8–5.0)	0.280	<0.001	<0.001
BPN visualization	3.2 (3.0–3.4)	5 (4.2–5.0)	4.2 (4.0–4.6)	<0.001	<0.001	0.042
TN visualization	4 (4.0–4.0)	3.6 (3.2–4.0)	5 (5.0–5.0)	0.017	<0.001	<0.001
SN visualization	3.2 (3.0–4.0)	2 (2.0–2.0)	5 (4.8–5.0)	<0.001	<0.001	<0.001

Data were presented as the median (IQR) of the average values derived from the five grades assigned by the five readers. The Friedman test with pairwise post-hoc Wilcoxon test with Bonferroni-Holm correction was employed. P < 0.05 means a significant difference. PD = proton density, TSE = turbo spin-echo, FS = fat suppression, 3D = three-dimensional, DESS = double echo steady state, DESS_HR = means DESS sequence with a high in-plane resolution, 7T = 7 tesla, MRI = magnetic resonance imaging, IQR = interquartile range, CPN = common peroneal nerve, BPN = branches of the peroneal nerve, TN = tibial nerve, SN = saphenous nerve

Fig. 4. Violin plots show the comparison of five Likert-type scale scores across PD-TSE FS, 3D-DESS, and DESS_HR sequences. The solid lines represent the median and the dashed lines indicate the interquartile range. The Friedman test compared these measurements across three sequences. Statistical significance is indicated on a level of ^*P < 0.05, ^†P ≤ 0.001. All given P-values are adjusted for multiple comparisons. PD = proton density, TSE = turbo spin-echo, FS = fat suppression, 3D = three-dimensional, DESS = double echo steady state, DESS_HR = means DESS sequence with a high in-plane resolution, ns = not significant, CPN = common peroneal nerve, BPN = branches of the peroneal nerve, TN = tibial nerve, SN = saphenous nerve.

Fig. 5. Image of a 30-year-old male. Red boxes indicate a close-up of BPN region. The reconstructed 3D-DESS images in the axial, coronal, and sagittal planes show that the peroneal nerve is divided into the AB, DPN, and SPN. Furthermore, the visualization of the BPN in the 3D-DESS images (A-C) is superior to that in the axial PD-TSE FS (D) and DESS_HR (E) images. BPN = branches of the peroneal nerve, 3D = three-dimensional, DESS = double echo steady state, AB = articular branch, DPN = deep peroneal nerve, SPN = superficial peroneal nerve, PD = proton density, TSE = turbo spin-echo, FS = fat suppression, DESS_HR = means DESS sequence with a high in-plane resolution.

DISCUSSION

Peripheral nerve imaging around the knee has long posed challenges at 3T MRI due to suboptimal image quality. Advanced 7T MRI holds promise for enhanced nerve visualization; however, studies investigating its applicability remain insufficient. This study demonstrates the potential of 7T MRI for imaging the peripheral nerves and fascicular structures around the knee. Specifically, the DESS_HR sequence at 7T, with an in-plane resolution of 0.17 × 0.17 mm², improved the visualization of the TN, common PN, SN, and intricate fascicular structures. Additionally, the 3D-DESS sequence with 0.4 mm³ isotropic resolution excelled in delineating and reconstructing smaller PN branches.

Our findings showed that while the SNR for the TN in the DESS_HR sequence was lower than that in the PD-TSE FS and 3D-DESS sequences, the SNR differences for other nerves among the sequences were statistically insignificant. Three sequence protocols were optimized for comparable SNR and scan times, thus maximizing the image quality. An enhanced CNR plays a crucial role in nerve visualization [32]. The DESS_HR sequence achieved higher nerve-to-fat contrast ratios, likely due to a decrease in the flip angle compared to the 3D-DESS sequence, which reduced T1 weighting, lowered fat signal brightness, and enhanced nerve signals. This led to a clearer delineation of the nerves, particularly the distal branch nerves. Notably, the CNR of the TN-to-fat was significantly lower than that of the PN-to-fat in both DESS sequences, suggesting heterogeneous FS. This discrepancy may arise from the DESS sequence’s selective excitation of water protons FS, which is sensitive to B0 inhomogeneities at 7T [15].

The separation of nerves from adjacent vessels is challenging in MR neurography [33]. In this study, PD and DESS_HR images demonstrated high CNR values from the nerves to the vessels. Vascular suppression and the higher spatial resolution of the PD and DESS_HR images at 7T enhanced the distinction between the vessels and peripheral nerves. Nerves typically exhibit a uniform intermediate SI, whereas the accompanying vessel walls exhibit an intermediate SI. The popliteal, anterior tibial, posterior tibial, and peroneal arteries and veins exhibited low or intermediate SI.

Additionally, we found that DESS images were less affected by ghosting pulsatile flow artifacts than PD-TSE FS images. This is likely due to the longer relaxation times and stronger phase-coding gradients in the PD sequence [15]. Conversely, the DESS sequence is insensitive to motion, and flow-induced displacement can attenuate the signal from flowing fluids, such as cerebrospinal fluid and blood vessels [21,34]. Furthermore, 3D scanning has the potential to reduce pulsatile motion artifacts when compared with 2D scanning. We also observed that the PD images were frequently accompanied by tissue signal inhomogeneity and instability in some slices, whereas these phenomena were absent in the DESS images. This may have occurred because B1 inhomogeneity is most prominent for current 2D spin echo, followed by 3D spin echo, and then 3D gradient echo-based imaging [15]. The DESS sequence exhibited diminished susceptibility to SI variation and pulsatile flow artifacts, thereby enhancing the overall image quality.

Notably, a high spatial resolution is crucial for visualizing nerve fascicles. The DESS_HR sequence at 7T provided exceptional detail in visualizing individual nerve fascicles of the PN, TN, and SN, surpassing the PD-TSE FS sequence. This level of resolution is essential for diagnosing subtle neuropathic lesions affecting individual fascicles. Moreover, the DESS_HR sequence achieved this detail with a shorter acquisition time than that reported in previous 7T MRI studies with similar resolution [35,36]. Remarkably, while the 3D-DESS sequence did not outperform the PD-TSE FS and DESS_HR sequences for SN and TN visualization, it excelled in depicting smaller PN branches such as articular and muscular branches [37]. Previous cadaver studies focused on the anatomical relationship between the PN and its branches with the proximal fibula to aid in surgical preparation [3,38]. However, visualizing these fine structures in vivo using the 3D-DESS sequence at 7T is clinically significant, especially for presurgical planning and reducing nerve injury during knee procedures, given the variability in individual nerve paths [3].

This study had several limitations. First, it lacked data on diagnostic performance across various nerve pathologies. Investigating the utility of the DESS sequence for different forms of peripheral neuropathy is crucial for future research. Second, we did not evaluate the individual SNR and CNR values for different slices of the peripheral nerves. Noise in MR images may not be uniformly distributed because of parallel imaging and B1 inhomogeneity, potentially affecting the accuracy of the SNR and CNR calculations. Third, while the DESS sequences at 7T offer superior visualization of nerve segments, the increased acquisition time may limit their clinical feasibility. Future research could explore deep-learning reconstruction techniques that enhance spatial resolution, leverage knowledge of local noise distribution, and locally adapt the strength to address noise variations in 7T MR while avoiding long scan times.

In conclusion, the DESS_HR sequence at 7T demonstrated superior contrast for visualizing peripheral nerves around the knee and provided detailed anatomical information on the nerve fascicular architecture. Moreover, the 3D-DESS sequence at 7T improved the visualization of the smaller and distal PN segments.

Availability of Data and Material

The datasets generated or analyzed during the study are available from the corresponding author on reasonable request.

Supplement

The Supplement is available with this article at https://doi.org/10.3348/kjr.2024.0912.