Influence of fractional anisotropy thresholds on diffusion tensor imaging tractography of the periprostatic neurovascular bundle and selected pelvic tissues: do visualized tracts really represent nerves?

Abstract

Background

Diffusion tensor imaging (DTI) tractography has recently been shown to successfully visualize periprostatic tracts allegedly representing the neurovascular bundle.

Purpose

To examine the impact of different fractional anisotropy (FA) thresholds on the results of DTI tractography in the male pelvis as well as to evaluate the resulting specificity for nerve tracts.

Material and Methods

Ten healthy male volunteers were examined at 3 Tesla. DTI tractography was performed based on seed points placed circularly around the prostate, in the rectoprostatic angle, the peripheral zone of the prostate, the sciatic nerve, and in addition the urinary bladder using FA thresholds of 0.20, 0.05, and 0.01. DTI tract number and DTI tract length measured with different FA thresholds were compared. ANOVA with repeated measures was used for statistics.

Results

DTI tract number and tract length were significantly dependent on FA thresholds. While a FA threshold of 0.20 visualized the typical distribution of DTI tracts in the sciatic nerve, a FA threshold of ≤0.05 was necessary to yield results visually mimicking the distribution of nerve tracts in the NVB. However, with such low FA thresholds even in the filled urinary bladder DTI tracts could be visualized. With FA thresholds of 0.20, the number and length of periprostatic DTI tracts did not differ from those measured within the prostate.

Conclusion

DTI tractography can be used to visualize DTI tracts periprostatically. However, one may doubt that these DTI tracts represent nerve tracts and that the periprostatic neurovascular bundle can be evaluated in a meaningful way with the current methods available.

Keywords

Magnetic resonance (MR) diffusion/perfusion prostate magnetic resonance (MR) functional imaging

Introduction

The periprostatic neurovascular bundle (NVB) is a complex meshwork of vessels and nerves located around the prostate. Since it also contains the periprostatic autonomic nerves crucial for erectile function, knowledge of the anatomy of the NVB is important for potency-sparing surgical approaches for prostate cancer. Traditionally, it has been assumed that the NVB is located dorsolaterally on both sides between the prostate and the rectum in the rectoprostatic angle (1). However, more recent pathological studies have proven that the distribution of periprostatic autonomic nerves is more complex and variable: While a majority of periprostatic nerves usually is located dorsolaterally, in some cases a high percentage of nerves can also be found ventrolaterally and dorsally of the prostate (2,3). The mean diameter of individual nerves of the neurovascular bundle is approximately 0.20 mm (4,5). Non-invasive techniques capable of visualizing the anatomy of the NVB in an individual patient would be desirable in order to further optimize nerve-sparing and thereby potency-sparing surgical approaches for prostate cancer.

Diffusion tensor imaging (DTI) tractography is a magnetic resonance imaging (MRI) technique that can be used for visualizing tracts based on anisotropic diffusion within structured tissues (6). It is a well-established technique to analyze and visualize nerve tracts in the central nervous system as well as selected parts of the peripheral nervous system (7 –10).

Several studies by different groups evaluated the feasibility of DTI tractography for the visualization of the periprostatic NVB in recent years (11 –13). These studies reported this technique to successfully visualize tracts circularly around the prostate as well as in the rectoprostatic angles potentially representing the NVB. Kitajima et al. even compared the results of patients before and after nerve-sparing and non-nerve-sparing prostatectomy and found the tract number to be decreased after non-nerve-sparing compared to nerve-sparing surgery (13). However, to our knowledge so far, a correlation of results of DTI tractography with histology as a reference standard is lacking.

Several technical parameters used for image acquisition as well as postprocessing are known to influence the number and length of tracts visualized by DTI tractography and consequently sensitivity and specificity of this technique for the visualization of nerve tracts (14). Among these parameters are spatial resolution, b-values, maximum allowed angle between two tracking steps (angle threshold), and fractional-anisotropy (FA) threshold used for tractography (14 –16). In the abovementioned studies, these details were not exhaustively reported.

The goal of the study at hand was to evaluate the impact of different FA thresholds on the results of DTI tractography of periprostatic tracts to demonstrate the impact of a single parameter on tractography results. In addition, DTI tractography with different FA thresholds was also performed in the sciatic nerve as a positive control for nerve tracts and in the prostate and the urinary bladder as negative control. In the following text, the term DTI tract will be used for any tract-like structure visualized by DTI tractography while the term nerve tract will be used exclusively for nerve fibers.

Material and Methods

Patient cohort

Ten healthy male volunteers were included in this study. The mean age of the volunteers was 27 years (age range, 22–30 years). Erectile dysfunction was excluded in all volunteers by a standardized questionnaire based on the international index of erectile function (IIEF) (17). Exclusion criteria were general contraindications for MRI due to unsafe implants or claustrophobia as well as contraindications for the application of butylscopolamine. Ethical approval for this prospective analysis was obtained from the institutions research ethics committee. All participants gave their informed consent to participate in this study.

Examination protocol

All volunteers underwent MRI of the prostate at 3 Tesla (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany). They were investigated in supine position; an 18-channel phased-array coil (Siemens Healthcare, Erlangen, Germany) was applied by an experienced MRI technician. The examination protocol included T2-weighted (T2W) imaging and diffusion tensor imaging (DTI). For DTI b-values of 0 and 800 s/mm² as well as 12 gradient directions were used. For details of the examination protocol, see Table 1. Butylscopolamine was administered intravenously in all volunteers to reduce potential movement artifacts due to bowel peristalsis.

Table 1.

Sequence parameters for T2W imaging and DTI.

Sequence	Plane	Slice thickness (mm)	FOV (mm)	Acquisition matrix	Pixel size reconstructed (measured) (mm)	PAT factor	Flip angle (°)	TR (ms)	TE (ms)	b-values (s/mm²)	Number of signal averaging	Gradient directions
T2 TSE	Axial	3	180 × 180	403 × 448	0.4 × 0.4 (0.4 × 0.4)	w/o	160	4000	108	NA	NA	NA
DTI	Axial	3	160 × 160	78 × 104	1.5 × 1.5 (2.1 × 1.5)	2	90	2900	70	0; 800	1 for b = 0; 8 for b = 800	12

DTI, diffusion tensor imaging; FOV, field of view; NA, not applicable; PAT, parallel acquisition technique; TE, echo time; TR, repetition time, TSE, turbo spin-echo.

DTI tractography and image analysis

DTI tractography was performed using a syngo Multimodality Workplace with the dedicated software Neuro3D (Siemens Healthcare, Erlangen, Germany) applying a modified fourth-order Runge-Kutta algorithm (18). High-resolution axial T2W imaging was fused with DTI datasets. Based on fused imaging with good delineation of anatomic structures, multiple seed points consisting of multiple seed voxels with a predefined volume of 7.81 mm³ were placed by a reader with experience in interpreting MRI of the prostate (AB, >4 years, >800 examinations) in standardized anatomic regions on an axial slice (Fig. 1):

Circularly around the prostate in the periprostatic tissue at the level of the midgland; a rim with a width of 2–3 seed voxels around the prostate was covered with seed voxels.

Bilaterally in the rectoprostatic angle at the level of the midgland; the area between rectum and prostate including the NVB as depictable on T2W imaging was covered with seed voxels.

Bilaterally in the sciatic nerve at the level of the ischial spine; the whole area of the sciatic nerve as depictable on T2W imaging was covered with seed voxels.

In the peripheral zone of the prostate on the left side with a minimal distance to the pseudocapsule of the prostate > 5 mm; 20 seed voxels were placed to cover a rectangular area.

In the filled urinary bladder with a minimal distance to the bladder wall of >10 mm; 20 seed voxels were placed to cover a rectangular area.

Fig. 1.

Axial fused T2W imaging and FA maps showing the placement of multiple seed voxels with a predefined size of 7.81 mm³ in different anatomic regions: (a) Circularly around the prostate at the level of the midgland. Different colors depict different seed points for the different quadrants; (b) In the rectoprostatic angles at the level of the midgland; (c) In the sciatic nerve at the level of the ischial spine; (d) In the peripheral zone of the prostate; (e) In the filled urinary bladder.

The FA values for all seed points were measured. Based on these seed points, DTI tractography was performed in four quadrants of periprostatic tissue (Q1, left anterior quadarant; Q2, left posterior quadrant; Q3, right posterior quadrant, and Q4, right anterior quadrant), bilaterally in the rectoprostatic angle, in the right and left sciatic nerve, in the peripheral zone of the prostate, and in the urinary bladder. DTI tractography was performed with FA thresholds of 0.20, 0.05, and 0.01. The angle threshold was kept constant at 30°. For each anatomic region the number as well as the length of all DTI tracts calculated with different FA thresholds was analyzed.

Statistical analysis

For comparison of DTI tract number and DTI tract length with different FA thresholds and in different anatomic regions, analysis of variance (ANOVA) with repeated measures was performed. Bonferroni correction was used to correct for multiple comparisons. A two-sided P value < 0.05 was considered statistically significant. All calculations were performed using SPSS (IBM SPSS Statistics, Version 21, IBM, Armonk, NY, USA) and Prism (Prism, Version 6.0 g, GraphPad Software, La Jolla, CA, USA).

Results

All volunteers scored 30 out of 30 achievable points answering the standardized questionnaire based on the IIEF thereby ruling out erectile dysfunction.

For the periprostatic tissue, the rectoprostatic angles, and the sciatic nerve, the total seed point volume differed between different anatomic regions as well as individual volunteers. The average total seed point volume for all anatomic regions can be seen in Table 2.

Table 2.

Average seed point volumes for different anatomic regions.

Seed point volume in mm³	Left anterior periprostatic quadrant	Left posterior periprostatic quadrant	Right posterior periprostatic quadrant	Right anterior periprostatic quadrant	Left rectoprostatic angle	Right rectoprostatic angle	Left sciatic nerve	Right sciatic nerve	Prostate	Urinary bladder
Mean	290.6	264.0	255.4	267.1	199.2	212.5	179.7	184.3	156.2	156.2
SD	54.5	24.4	15.2	48.8	59.5	46.5	54.9	48.5	0	0

SD, standard deviation.

FA values for all seed points are shown in Table 3. Significant differences in FA values were found between all quadrants of periprostatic tissue (P = 0.002) as well as both rectoprostatic angles (P = 0.004 for the left rectoprostatic angle, P = 0.001 for the right rectoprostatic angle) compared to the filled urinary bladder. While FA values in the right sciatic nerve were found to be significantly different from those of at least the posterior quadrants of periprostatic tissue as well as both rectoprostatic angles, no such difference could be demonstrated for the left sciatic nerve. In addition, no significant difference was found between the periprostatic tissues as well as both rectoprostatic angles compared to the peripheral zone of the prostate.

Table 3.

Fractional anisotropy (FA) values for different anatomical regions.

Fractional anisotropy	Left anterior periprostatic quadrant	Left posterior periprostatic quadrant	Right posterior periprostatic quadrant	Right anterior periprostatic quadrant	Left rectoprostatic angle	Right rectoprostatic angle	Left sciatic nerve	Right sciatic nerve	Prostate	Urinary bladder
Mean	0.263	0.190	0.204	0.267	0.197	0.209	0.353	0.361	0.191	0.087
SD	0.054	0.041	0.052	0.065	0.050	0.052	0.097	0.045	0.057	0.018

SD, standard deviation.

Due to the different seed point volumes, the number of DTI tracts running through these seed points was normalized by dividing it through the number of seed voxels in each individual seed point. Significant differences in DTI tract number per seed voxel were found dependent on the FA threshold in all four quadrants of periprostatic tissue, bilaterally in the rectoprostatic angles, in the peripheral zone of the prostate, and in the filled urinary bladder with P values of <0.001, respectively. While in the left sciatic nerve a significant difference was found (P = 0.046), there was no significant difference in the right sciatic nerve (P = 0.074). The number of DTI tracts per seed voxel calculated with different FA thresholds as well as P values for differences between selected FA thresholds for all anatomic regions are shown in Fig. 2. A major increase in DTI tract number was observed when lowering the FA threshold to 0.05. Using FA thresholds ≤0.05, the number of DTI tracts per seed voxel approximated 8; this maximum number of tracts per seed voxel is determined by the software and algorithm used for tractography.

Fig. 2.

Bar graph showing the average DTI tract number per seed voxel with different FA thresholds in all anatomic regions examined. An asterisk (*) marks a statistically significant difference between two FA thresholds with a P value < 0.001. Using FA thresholds ≤ 0.05, the number of tracts per seed voxel approximates 8 in all anatomic regions. Error bars represent the standard deviation.

Using a FA threshold of 0.20 no significant difference in DTI tract number per seed voxel was found between all four quadrants of periprostatic tissue, the rectoprostatic angles, and the peripheral zone of the prostate (P = 0.419).

Significant differences in DTI tract length were found dependent on the FA threshold in all anatomic regions. The overall P value was 0.002 for the left sciatic nerve and < 0.001 for all other anatomic regions examined. The average DTI tract length per seed point is shown in Fig. 3.

Fig. 3.

Bar graph showing the average DTI tract length measured with different FA thresholds in all anatomic regions examined. Significant differences in tract length were found dependent on the FA threshold in all anatomic regions. The overall P value was 0.002 for the left sciatic nerve and <0.001 for all other anatomic regions examined. Error bars represent the standard deviation.

Examples of the visualization of DTI tracts in all four quadrants of periprostatic tissues and the rectoprostatic angles as well as the sciatic nerve, the peripheral zone of the prostate, and the filled urinary bladder calculated with different FA thresholds can be seen in Figs. 4 and 5.

Fig. 4.

Three-dimensional visualization of DTI tracts circularly around the prostate (a–c) and in the rectoprostatic angle (d–f) calculated with FA thresholds of 0.20 (a, d), 0.05 (b, e), and 0.01 (c, f). All images are regarded from an oblique dorsocranial perspective; the degree of tilting is adjusted for optimal visualization of tracts. DTI tracts are represented as tubes. The orientation of tracts is color-coded. Thickness of tubes does not reflect the true thickness of tracts.

Fig. 5.

Three-dimensional visualization of DTI tracts in the sciatic nerve (a–c), in the peripheral zone of the prostate (d–f), and in the filled urinary bladder (g–i) with a FA threshold of 0.20 (a, d, g), 0.05 (b, e, h), and 0.01 (c, f, i). All images are regarded from an oblique dorsocranial perspective; the degree of tilting is adjusted for optimal visualization of tracts. DTI tracts are represented as tubes. The orientation of tracts is color-coded. Thickness of tubes does not reflect the true thickness of tracts.

Discussion

Radical prostatectomy for treatment of prostate cancer is associated with relatively high rates of complications including impotence due to injury of the NVB; even with surgical approaches aimed at preserving potency such as nerve-sparing radical retropubic prostatectomy, rates of postoperative impotence of up to 24% are seen (19,20). Non-invasive techniques capable of visualizing the complex anatomy of the NVB in individual patients would be desirable in order to further optimize surgical approaches aimed at preserving potency.

Recently, several studies have evaluated the value of DTI tractography for this purpose using seed points located circularly around the prostate and in the rectoprostatic angles (11 –13). Results show that visualization of periprostatic DTI tracts is possible using different b-values of 600 (11,13) and 1000 s/mm² (12), different numbers of diffusion sensitizing directions (6 directions (13), 12 directions (11), and 16 directions (12)), and different spatial resolutions (pixel size, 1.4 × 2.0 mm; slice thickness, 3.0 mm (12); pixel size, 2.0 × 2.3 mm; slice thickness, 2.5 mm (13)). Finley et al. did not state spatial resolution for DTI (11).

While Finley et al. conclude that correlation with histology was necessary to prove that visualized tracts really represent nerve tracts (11), Kagani et al. stated that “DTI technique may be feasible for visualization of periprostatic nerve fibers” (13). Panebianco et al. even concluded that “DTI offers optimal representation of the widely distributed periprostatic plexus” (12). However, none of the abovementioned studies used pathology as a reference standard and none of these studies reported some important details on tractography algorithms applied, e.g. the FA threshold.

In the study at hand, we wanted to examine the influence of this parameter on tractography results. We were able to reproduce the results of the abovementioned studies visualizing periprostatic DTI tracts using DTI tractography in all volunteers. However, this was only possible using very low FA thresholds and therefore the question remains if the results of tractography really represent nerve tracts or if they are rather the results of tissue microstructures other than nerve tracts causing anisotropic diffusion or even artifacts.

Several studies have described the FA values of healthy peripheral nerves including the median nerve, lumbar nerve roots, and the sciatic nerve in humans as well as animal models and found them to be significantly higher than 0.20 (8,9,21). In consequence, in these nerves DTI tractography yielded good results when performed with FA thresholds > 0.20. Our results for the sciatic nerve are in agreement with these studies: FA values in the sciatic nerve were found to be significantly higher than 0.20 and here DTI tractography successfully visualized tracts using all FA thresholds tested. However, in periprostatic tissues the FA values were found to be much lower and here only FA thresholds of 0.05 and 0.01 yielded results visually mimicking the distribution of nerve tracts of the NVB.

The FA threshold has an important influence on sensitivity and specificity of DTI tractography for nerve tracts. Using DTI tractography, tracts can theoretically be calculated based on anisotropic diffusion of protons due to any kind of physical barriers interfering with diffusion perpendicular to these barriers. This has been demonstrated for connective tissues including cardiac muscle, skeletal muscle, and tendons, where DTI has proven its capability to visualize DTI tracts representing the architecture of these tissues (22 –25). Budzik et al., for example, were able to show that DTI tractography of thigh muscles was possible with FA thresholds of as high as 0.15 (26).

In the study at hand, FA values in the region of the NVB were found to be in the same range as in the peripheral zone of the prostate. In addition, in the peripheral zone of the prostate DTI tractography using a FA threshold of 0.20 yielded similar DTI tract numbers and DTI tract lengths per seed voxel compared to periprostatic tissue in all volunteers. However, in the area inside the prostate where the seed points for tractography were placed, no relevant numbers of nerve tracts exist. Relatively high FA values in the peripheral as well as the transitional zone of the prostate have also been described in other studies and in the peripheral zone DTI tracts with a predominant craniocaudal course have been attributed to fibromuscular tissue (27,28). When the FA threshold is lowered to ≤0.05, as it was necessary to yield significant results visually mimicking the distribution of periprostatic nerve fibers in the region of the NVB, DTI tractography can even yield artificial DTI tracts in tissues or even non-vital substances without any physical barriers causing anisotropic diffusion due to image noise. This could be observed in the filled urinary bladder in all volunteers. Similar artifacts can also be assumed to occur in the rectoprostatic angle and in the periprostatic tissue when very low FA thresholds are used.

Since none of the abovementioned studies evaluating DTI tractography for visualization of the periprostatic NVB provided histology as a reference, the results have to be interpreted with much caution (11 –13), especially when focusing merely on the evaluation of DTI tract numbers as well as visualizations of these DTI tracts mimicking the distribution of nerve tracts of the NVB. The results of our study show that using low FA thresholds necessary to yield optimal results in terms of DTI tract visualization in the region of the NVB, yields similar results in the peripheral zone of the prostate and therefore is not specific for nerve tracts in the pelvic region.

Our study has some limitations. First, like in previous studies no histopathology could be used as a reference standard to evaluate sensitivity and specificity. Therefore, measurements in other tissues with or without relevant numbers of nerve tracts had to be used as a surrogate and for positive and negative controls, respectively. Second, of several technical parameters used for image acquisition as well as postprocessing that are known to influence sensitivity and specificity of DTI tractography for nerve tracts only the influence of the FA threshold was tested while the other parameters (e.g. b-values, angle threshold) were kept constant. Therefore, no conclusions about the influence of these other parameters on sensitivity and specificity for nerve tracts can be drawn. Spatial resolution also has an influence on results of DTI tractography and might be of special importance when trying to visualize nerve tracts with diameters as small as in the NVB. For example, in a study of the human forearm a higher resolution of DTI could be shown to improve the visualization of smaller peripheral nerves (15). The influence of different spatial resolutions on the results of DTI tractography of the periprostatic DTI tracts was not tested in the study at hand. However, the spatial resolution used in the used study protocol was similar or even superior to that used in previous studies evaluating DTI tractography for visualization of the NVB.

In conclusion, until now the question if the results of periprostatic DTI tractography really represent nerve tracts cannot be answered. In fact, based on the data presented above, one may doubt that periprostatic nerve tracts can be evaluated in a meaningful way with the current methods available. Further studies potentially correlating the results of DTI tractography with histology as well as functional parameters will be necessary in the future. Future studies need to report full technical details and parameters potentially influencing the results of tractography in order to allow to compare the results of different studies.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Walsh

Donker

. Impotence following radical prostatectomy: insight into etiology and prevention. J Urol 1982; 128: 492–497.

Ganzer

Blana

Gaumann

. Topographical anatomy of periprostatic and capsular nerves: quantification and computerised planimetry. Eur Urol 2008; 54: 353–360.

Ganzer

Blana

Stolzenburg

. Nerve quantification and computerized planimetry to evaluate periprostatic nerve distribution-does size matter? Urology 2009; 74: 398–403.

Sung

Lee

Park

. Neuroanatomical study of periprostatic nerve distributions using human cadaver prostate. J Korean Med Sci 2010; 25: 608–612.

Sievert

Hennenlotter

Laible

. The periprostatic autonomic nerves–bundle or layer? Eur Urol 2008; 54: 1109–1116.

Basser

Pajevic

Pierpaoli

. In vivo fiber tractography using DT-MRI data. Magn Reson Med 2000; 44: 625–632.

Nucifora

Verma

Lee

. Diffusion-tensor MR imaging and tractography: exploring brain microstructure and connectivity. Radiology 2007; 245: 367–384.

Hiltunen

Suortti

Arvela

. Diffusion tensor imaging and tractography of distal peripheral nerves at 3 T. Clin Neurophysiol 2005; 116: 2315–2323.

Kabakci

Gurses

Firat

. Diffusion tensor imaging and tractography of median nerve: normative diffusion values. Am J Roentgenol 2007; 189: 923–927.

10.

Takagi

Nakamura

Yamada

. Visualization of peripheral nerve degeneration and regeneration: monitoring with diffusion tensor tractography. Neuroimage 2009; 44: 884–892.

11.

Finley

Ellingson

Natarajan

. Diffusion tensor magnetic resonance tractography of the prostate: feasibility for mapping periprostatic fibers. Urology 2012; 80: 219–223.

12.

Panebianco

Barchetti

Sciarra

. In vivo 3D neuroanatomical evaluation of periprostatic nerve plexus with 3T-MR diffusion tensor imaging. Eur J Radiol 2013; 82: 1677–1682.

13.

Kitajima

Takahashi

Ueno

. Visualization of periprostatic nerve fibers before and after radical prostatectomy using diffusion tensor magnetic resonance imaging with tractography. Clin Imaging 2014; 38: 302–306.

14.

Seehaus

Roebroeck

Chiry

. Histological validation of DW-MRI tractography in human postmortem tissue. Cereb Cortex 2013; 23: 442–450.

15.

Zhou

Narayana

Kumaravel

. High resolution diffusion tensor imaging of human nerves in forearm. J Magn Reson Imaging 2014; 39: 1374–1383.

16.

Domin

Langner

Hosten

. Comparison of parameter threshold combinations for diffusion tensor tractography in chronic stroke patients and healthy subjects. PLoS One 2014; 9: e98211–e98211.

17.

Rosen

Riley

Wagner

. The international index of erectile function (IIEF): a multidimensional scale for assessment of erectile dysfunction. Urology 1997; 49: 822–830.

18.

Bammer

Acar

Moseley

. In vivo MR tractography using diffusion imaging. Eur J Radiol 2003; 45: 223–234.

19.

Catalona

Carvalhal

Mager

. Potency, continence and complication rates in 1,870 consecutive radical retropubic prostatectomies. J Urol 1999; 162: 433–438.

20.

Kundu

Roehl

Eggener

. Potency, continence and complications in 3,477 consecutive radical retropubic prostatectomies. J Urol 2004; 172: 2227–2231.

21.

Chuanting

Qingzheng

Wenfeng

. 3.0T MRI tractography of lumbar nerve roots in disc herniation. Acta Radiol 2014; 55: 969–975.

22.

Tseng

Wedeen

Reese

. Diffusion tensor MRI of myocardial fibers and sheets: correspondence with visible cut-face texture. J Magn Reson Imaging 2003; 17: 31–42.

23.

Froeling

Nederveen

Heijtel

. Diffusion-tensor MRI reveals the complex muscle architecture of the human forearm. J Magn Reson Imaging 2012; 36: 237–248.

24.

Sarman

Atmaca

Cakir

. Assessment of postoperative tendon quality in patients with achilles tendon rupture using diffusion tensor imaging and tendon fiber tracking. J Foot Ankle Surg 2015; 54: 782–786.

25.

Sieben

van Otten

Lataster

. In vivo reconstruction of lumbar erector spinae architecture using diffusion tensor MRI. Clin Spine Surg 2016; 29: E139–145.

26.

Budzik

Le Thuc

Demondion

. In vivo MR tractography of thigh muscles using diffusion imaging: initial results. Eur Radiol 2007; 17: 3079–3085.

27.

Manenti

Carlani

Mancino

. Diffusion tensor magnetic resonance imaging of prostate cancer. Invest Radiol 2007; 42: 412–419.

28.

Gurses

Tasdelen

Yencilek

. Diagnostic utility of DTI in prostate cancer. Eur J Radiol 2011; 79: 172–176.