Optimization of the Parameters for Diffusion Tensor Magnetic Resonance Imaging Data Acquisition for Breast Fiber Tractography at 1.5 T

Optimization of the Parameters for Diffusion Tensor Magnetic Resonance Imaging Data Acquisition for Breast Fiber Tractography at 1.5 T

Original Study Optimization of the Parameters for Diffusion Tensor Magnetic Resonance Imaging Data Acquisition for Breast Fiber Tractography at 1.5 T...

1MB Sizes 0 Downloads 44 Views

Original Study

Optimization of the Parameters for Diffusion Tensor Magnetic Resonance Imaging Data Acquisition for Breast Fiber Tractography at 1.5 T Yuan Wang, Xiao-Peng Zhang, Yan-Ling Li, Xiao-Ting Li, Yan Hu, Yong Cui, Ying-Shi Sun, Xiao-Yan Zhang Abstract Optimizing image acquisition parameters is essential for producing high-quality diffusion tensor (DT) imaging (DTI) data. This study aimed to optimize the parameters for DTI data acquisition for breast fiber tractograhy. Twenty-one healthy volunteers received breast DTI scanning using an array spatial sensitivity encoding technique (ASSET)-based echo-planar imaging (EPI) technique. The optimization of data acquisition parameters could improve the quality of breast DT magnetic resonance imaging (MRI) images and assist fiber tractography at 1.5 T. Introduction: Diffusion tensor MRI has emerged as a promising tool for the analysis of the microscopic properties of tissues. Optimizing image acquisition parameters is essential for producing high-quality DTI. This study aimed to optimize the parameters for DTI data acquisition for breast fiber tractography at 1.5 T. Patients and Methods: A total of 21 healthy volunteers received breast DTI scanning using an ASSET-based EPI technique operated under different parameters including b value, the number of diffusion gradient directions, and spatial resolution. The images were analyzed for signal-to-noise, signal intensity ratio, mean number and length of reconstructive fiber tracts, and fractional anisotropy value. Results: The optimal acquisition parameters at 1.5 T for breast DT-MRI fiber tractography were determined as follows: axial 31 direction, b ¼ 600 seconds per mm2, matrix 128  128 with slice thickness of 3 mm. Conclusion: The optimization of data acquisition parameters could improve the quality of breast DT-MRI images and assist fiber tractography at 1.5 T. Clinical Breast Cancer, Vol. 14, No. 1, 61-7 ª 2014 Elsevier Inc. All rights reserved. Keywords: Array spatial sensitivity encoding technique, B value, Diffusion gradient directions, Diffusion tensor imaging, Spatial resolution

Introduction Diffusion tensor magnetic resonance (MR) imaging (MRI) has emerged as a promising tool for the analysis of the microscopic properties of tissues. In particular, this technique can be used to infer the directionality of white matter fiber architecture within a voxel.1,2 With the further development of MRI technology, diffusion tensor imaging (DTI) can be used for the organs in the body.3,4

Key Laboratory of Carcinogenesis and Translational Research (Ministry of Education), Department of Radiology, Peking University Cancer Hospital and Institute, Beijing, China Submitted: Jun 12, 2013; Revised: Sep 17, 2013; Accepted: Sep 24, 2013; Epub: Sep 27, 2013 Address for correspondence: Xiao-Peng Zhang, MD, Peking University Cancer Hospital and Institute, No. 52 Fu Cheng Rd, Beijing 100142, China Fax: þ86-10-88133398; e-mail contact: [email protected]

1526-8209/$ - see frontmatter ª 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.clbc.2013.09.002

Currently, few studies have reported the application of DTI on breast tissue and most of them focused on the differential diagnosis between benign and malignant lesions. Parameters such as average diffusion coefficient and maximum eigen value have been demonstrated to be helpful in the differential diagnosis of breast lesions.5-7 Fractional anisotropy (FA) revealed the difference between breast carcinomas and normal breast tissue.6 Partridge et al demonstrated the anisotropy of water diffusion in normal breast tissue.8 The primary source of anisotropy measurements in the breast tissue is not clear, although it might be related to the architecture of fibrous breast stroma.8 Therefore, the low anisotropy presents a challenge to fiber-tracking using breast DTI data. Diffusion tensor imaging measures such as FA and relative anisotropy are more affected by measurement noise than diffusion weighted imaging (DWI) measures such as apparent diffusion coefficient (ADC). Thus, optimizing image acquisition parameters is

Clinical Breast Cancer February 2014

- 61

Optimization of DTI-MRI an essential step for producing high-quality DTI scans.9 To increase the accuracy of breast fiber tractography, it is also important to improve the quality of the images.10 Factors affecting image quality often include the MRI hardware configuration and software, acquisition sequences, and scanning parameters.10 With the current scanner, coils, and acquisition sequences as fixed conditions, adjustments to the scanning parameters would provide a method of improving image quality. In this study we aimed to optimize the acquisition parameters at 1.5 T for breast fiber tractography to develop a new protocol to improve the image quality of breast DTI at 1.5 T.

Patients and Methods Patients Between November 2011 and March 2012, a total of 21 healthy volunteers (age range, 25-65 years; median age, 35 years), who had no previous breast disease and no symptoms of breast disease, participated in this study. This study was approved by our institutional review boards and all volunteers provided informed consent.

the first 2 groups. The scanning times for 6, 15, and 31 directions were 4 minutes 21 seconds, 9 minites 45 seconds, and 19 minutes 21 seconds, respectively.

Imaging Postprocessing All of the DTI images were sent to a GE workstation Advantage Windows 4.4 and analyzed by a radiologist with 2 years of experience using the Functool software version 9.4.05. Fiber tracking for breast tissue was performed using an off-line software named Diffusion Toolkit (version 0.6.1, Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, MA, USA) based on the fiber assessment by continuous tracking method. Minimum FA and maximum turning angle for reconstruction were set as the default setting by Diffusion Toolkit (minimum FA ¼ 0.05, maximum turning angle ¼ 50 ). At the largest slice of the breast tissue, a sphere-shaped region of interest (ROI) with a uniform 6  6 size was used to generate fiber tracts. For fiber-tracking effect evaluation, the parameters of mean number and mean length of reconstructed fibers were used as the evaluation index according to previous studies.11,12

Magnetic Resonance Imaging Scanning Protocol All of the MRI was performed on a 1.5 T MR scanner (Optima, MR360; General Electric Healthcare, Beijing, China) using a dedicated phase-array 8-channel bilateral breast-receiving coil. For MRI, the subjects were in the prone position, MR examination included a T2-weighted fast spin-echo fat saturation (fat-sat) sequence in the sagittal direction (repetition time (TR) ¼ approximately 3000-3500 millisecond [ms], echo time (TE) ¼ 79 ms, number of excitations (Nex) ¼ 2, flip angle ¼ 90 , spatial resolution ¼ 1.4  2.1  4.0 mm3), T1-weighted fat-sat sequence in the axial direction (TR ¼ 5.3 ms, TE ¼ 2.5 ms, Nex ¼ 0.75, flip angle ¼ 10 , spatial resolution ¼ 0.8  0.7  3 mm3).

Diffusion Tensor Imaging-MRI The breast DTI was performed using a single-shot echo-planar imaging sequence with an asset factor of 2. The constant parameters were as follows: TR/TE ¼ 9000/minimum, Nex ¼ 4, field of view ¼ 36 cm, flip angle ¼ 90 . Other DTI sequence scanning parameters varied in 3 groups.

Image Analysis and ROI Arrangement The signal to noise ratio (SNR), mean number and mean length of reconstructed fibers, and FA value were analyzed to confirm the desirable b value, the suitable spatial resolution, and the appropriate NDGD. There were 2 types of ROI: the first was in the normal breast parenchyma; the second was in the background noise region beyond the breast parenchyma. To compare the image quality of different groups, SNRs for diffusion weighted image with moderate or high b-value ( 400 seconds per mm2) were calculated from the mean gray values (or signal intensity) of ROI parenchyma and ROI noise according to the following formula: SNR ¼ SIparenchyma/ Sdnoise, where SIparenchyma was signal intensity of breast tissue and Sdnoise was the average of 3 measurements of background noise. Next, the image quality from the optimal acquisition parameters was evaluated and the total 3-D tractography effects of reconstructed fibers were evaluated by calculation of mean length and mean number of the reconstructed fibers.

Statistical Analysis

62

-

Group 1: Different b Value. Seven healthy volunteers received a DTI examination with different b values of 400, 600, 800, and 1000 seconds per mm2. For all of the b values, the matrix was 128  128, and the thickness was 3 mm. Number of diffusion gradient directions (NDGD) was fixed at 6 directions.

All data were reported as mean  standard deviation (SD). Multiple comparison was analyzed using repeated measurement variance analysis with Bonferroni correction of the P value. The statistical analyses were performed using the SPSS 16.0 software. A value of P < .05 was considered significant.

Group 2: Different Spatial Resolution. Seven healthy volunteers received a DTI examination with the following parameters: the matrix size and thickness varied from matrix 128  128 and 3 mm thickness, matrix 192  192 and 5 mm thickness, to matrix 256  256 and 5 mm thickness, yielding 3 sets of diffusion tensor images for each person. NDGD was 6 directions and b value was fixed at the values determined in group 1.

Results

Group 3: Different NDGD. Seven healthy volunteers received DTI examination with different NDGD of 6, 15, and 31 directions. The b value and the spatial resolution were fixed at the value determined in

Clinical Breast Cancer February 2014

Optimization of b Value First we optimized b value. As shown in Table 1 and Figure 1, the SNR decreased with the increase of b value (from 400 to 1000 seconds per mm2) and SNR was the highest with b value of 400 seconds per mm2 compared with SNR with other b values (P < .001). The mean length of the reconstructed fibers showed a decreasing trend with the increase of b value (P < .001). The mean numbers of reconstructed fibers with b value of 600 and 800 seconds per mm2 showed no significant difference (P > .05), but generally decreased with the increase of b value (from 400 to 1000

Yuan Wang et al Table 1 Optimization of b Value for DW-MRI in 7 Healthy Volunteers b Value 400 600 800 1000 P

SNR

Mean Length (mm)

Mean Track Fibers, n

FA

38.18  14.95 28.18  11.03 20.67  7.23 14.72  6.68 < .001

9.68  3.70 7.99  2.26 6.97  2.06 6.47  2.38 < .001

597.53  156.62 510.15  84.16 509.27  115.55 467.49  96.07 < .001

0.29  0.11 0.28  0.10 0.28  0.08 0.29  0.09 .595

Abbreviations: DW ¼ diffusion weighted; FA ¼ fractional anisotropy; MRI ¼ magnetic resonance imaging; SNR ¼ signal to noise ratio.

seconds per mm2; P < .001). The FA value showed no differences when b values were varied (P ¼ .595). After comparison, a b-value of 600 seconds per mm2 was chosen for further study by balancing the diffusion weight and SNR and fiber tractography effect.

Optimization of Spatial Resolution Next we optimized the spatial resolution parameter and the results are summarized in Table 2 and Figure 2. When the spatial resolution was set at 2.8  2.8  3 mm3, the image exhibited the highest SNR and the best tractography effect of reconstructed fibers. The number of reconstructed fibers was the largest at the spatial resolution of 2.8  2.8  3 mm3, with statistical significance comparable with other resolutions (P < .001). The longest fibers

presented at the spatial resolution of 2.8  2.8  3 mm3, with significant difference between 2.8  2.8  3 mm3 and 1.9  1.9  5 mm3 or 1.4  1.4  5 mm3 (P128-192 ¼ .007, P128-256 ¼ .003), and the mean length of the reconstructed fibers showed no difference between 1.9  1.9  5 mm3 and 1.4  1.4  5 mm3. The FA value showed no differences when resolution parameters were varied (P ¼ .091). In this group, the 2.8  2.8  3 mm3 was best suitable for the breast fiber tractography at 1.5 T MR scanning.

Optimization of NDGD Finally, we optimized the NDGD parameter and the results are summarized in Table 3 and Figure 3. In this step, we kept a constant b value of 600 seconds per mm2 and a fixed spatial resolution

Figure 1 Optimization of b Factor in a Female, 38-Year-Old Healthy Volunteer. (A-D) Breast DTI map With Different b Values (b s 0 Seconds per mm2); Scale is on the Right Side of the Breast. (E-L) The Images, Postprocessed Using the Diffusion Tool Software. (E-H) Shows the Position of the Sphere ROI (Located in the Central Portion of the Left Breast). (I-L) The Reconstructive Fibers Through the ROI, With Green Presenting in an Anterior-Posterior Direction, red Presenting on the LeftRight Side, and Blue Presenting a Superior-Inferior Direction. With the b Value Increasing From 400 to 1000 Seconds per mm2, SNR Decreased and the Mean Length and Mean Number of Reconstructive Fibers Decreased, Fiber Became Disorganized, and Some Fibers run Outside of the Gland Contour

Abbreviations: DTI ¼ diffusion tensor imaging; ROI ¼ region of interest; SNR ¼ signal to noise ratio.

Clinical Breast Cancer February 2014

- 63

Optimization of DTI-MRI Table 2 Optimization of Spatial Resolution for DT-MRI in 7 Healthy Volunteers (b Value [ 600 Seconds per mm2) Spatial Resolution

SNR

Mean Length (mm)

Mean Track Fibers, n

FA

2.8 3 2.8 3 3 mm3 1.9 3 1.9 3 5 mm3 1.4 3 1.4 3 5 mm3 P

31.91  13.72 29.45  14.02 16.89  7.12 .005

4.90  0.89 3.45  0.46 2.76  0.26 .008

344.14  89.92 123.29  30.71 52.86  21.33 < .001

0.10  0.07 0.10  0.07 0.11  0.08 .091

Abbreviations: DT ¼ diffusion tensor; FA ¼ fractional anisotropy; MRI ¼ magnetic resonance imaging; SNR ¼ signal to noise ratio.

with 2.8  2.8  3 mm3 in groups with different NDGD. The 6-, 15-, and 31-direction scans showed no statistical significance with regard to the SNR (P ¼ .493), and the SNR was the highest for the 6-direction scan. The mean number and length of reconstructed fibers demonstrated significant differences when different NDGD parameters were used (P ¼ .001). The 31-direction scan showed the best fiber-tracking effect. The FA measurements also showed significant differences when different NDGD parameters were used

(P ¼ .027). The 31-direction scan showed a lower FA than the 6-direction scan (P ¼ .01), and there was no difference between 15 directions and 6 or 31 directions with respect to the FA measurements.

Discussion It is well established that mammary malignancies originate in the epithelial tissue of the ducts, and spread along ducts. Consequently,

Figure 2 Optimization of Spatial Resolution in a Female, 25-Year-Old Healthy Volunteer. (A-C) Breast DTI map With Different Spatial Resolution, Scale is on the Right Side of the Breast. (D-I) The Images Postprocessed Using Diffusion Tool Software. (D-F) The Position of the Sphere ROI (Located in the Central Area of the Left Breast). (G-I) The Reconstructive Fibers Through the ROI. With Increasing Voxel Size, SNR Increased and the Mean Length and Mean Number Reconstructive Fibers Increased Greatly; 2.8 3 2.8 3 3 mm Showed the Remarkable Reconstructive Effect

64

-

Abbreviations: DTI ¼ diffusion tensor imaging; ROI ¼ region of interest; SNR ¼ signal to noise ratio.

Clinical Breast Cancer February 2014

Yuan Wang et al Table 3 Optimization of NDGD for DT-MRI in 7 Healthy Volunteersa NDGD 6 15 31 P

SNR

Mean Length (mm)

Mean Track Fibers, n

FA

Scanning Time

45.54  21.92 45.85  20.62 40.67  14.32 .493

9.54  3.98 11.69  4.42 16.75  7.99 .001

614.43  170.85 658.57  166.26 841.29  250.15 .001

0.16  0.08 0.12  0.05 0.11  0.06 .027

4 min 21 s 9 min 45 s 19 min 21 s e

Abbreviations: DT ¼ diffusion tensor; FA ¼ fractional anisotropy; MRI ¼ magnetic resonance imaging; NDGD ¼ number of diffusion gradient directions; SNR ¼ signal to noise ratio. a b Value ¼ 600 seconds per mm2; spatial resolution was 2.8  2.8  3 mm3.

the ductal structure is an imperative area of investigation of malignant breast transformation. Disruption of the breast ductal network by invading cancer cells might cause alterations in water diffusion properties (rate and directionality) in the tissue that will be reflected by DTI parameters. Here we developed a novel method for the optimization of fiber tractography toward in vivo observation of the stereoscopic breast parenchyma (ductal and fiber support structures). In this study we optimized several parameters of DTI for the breast in healthy volunteers. We found that FA measurements had no obvious correlation with b value. These results suggest that FA and the ellipsoid shape of water molecule diffusion are not susceptible to b value, consistent with previous studies on brain. It is known that many microvessels exist in breast tumors, especially in malignancy. At lower b values, the motion of water molecules includes diffusion and microperfusion.13-15 Higher b values might overcome the microperfusion effect and approach natural diffusion.15 Therefore, it is necessary to use high b values to obtain accurate measurements. Indeed, b value is the important factor but there is more contribution from faster-moving water molecules at lower b values that affects the SNR of the DWI image, because at lower b values it is difficult to accurately demonstrate the diffusion nature of water molecules. However, if the b value reached 1000 seconds per mm2, the signal of breast tissue decreased to noise level, and if b value increased to greater than 1000 seconds per mm2,

signal attenuation of breast tissue became too severe to permit fiber tracking.13,16 Thus, in this study we chose b values within the range of 400 and 1000 seconds per mm2. We found that SNR and tractography effect of reconstructed fibers showed a decreasing tendency if b value increased. Especially, the image quality was compromised at high b values, making it difficult to observe the details. Adequate SNR is a prerequisite for fiber tracking.17,18 In this study, b values of 400 and 600 seconds per mm2 showed better fiber tracking effects than b values of 800 and 1000 seconds per mm2. To balance the diffusion weight and tractography effect, we chose 600 seconds per mm2 as the optimal b value for breast DTI tractography. In addition, typical ADC values of 1.6 to 2.0  103 mm2 per second for normal breast tissue correspond with an optimal diffusion weighted of approximately b ¼ 600 seconds per mm2.19 This is consistent with the simple rule of thumb that the optimal b-value multiplied by the ADC of the tissue under investigation should be close to 1.10 The low spatial resolution might directly induce the partial volume effect, which affects the accuracy of DTI data and causes fiber connections to jump, and faulty connections between nonrelated reconstructed fibers. The thicker slice might cause a partial volume effect along the z-axis, resulting in the reduction of reconstructed fibers and faulty connections. In this study, we used a 3-mm slice thickness to improve the resolution of slice selection direction. Although the resolution within the slice was reduced,

Figure 3 Optimization of NDGD in a Female, 38-Year-Old Healthy Volunteer. (A-C) Breast DTI map With Different NDGD; Scale is on the Right Side of the Breast. (D-F) The Images Postprocessed Using Diffusion Tool Software. The Reconstructive Fibers Through the Whole Slice, With Green Presenting an Anterior-Posterior Direction, red Presenting a Left-Right Side, and Blue Presenting a Superior-Inferior Direction. As the NDGD was Increased From 6 to 31 Directions, the Mean Length and Mean Number of Reconstructive Fibers (Green Regions) Increased. Thirty-one Directions Showed a Much Better Fiber Tractography Effect

Abbreviations: DTI ¼ diffusion tensor imaging; NDGD ¼ number of diffusion gradient directions.

Clinical Breast Cancer February 2014

- 65

Optimization of DTI-MRI Figure 4 The Effects of Spatial Resolution on DTI Images in a Female, 25-Year-Old Healthy Volunteer. (A-C) Breast DTI map With Different Spatial Resolution From 2.8 3 2.8 3 3 mm3 (Matrix, 128 3 128; Slice Thickness, 3 mm), 1.9 3 1.9 3 2 mm3 (Matrix, 192 3 192; Slice Thickness 2 mm), 1.4 3 1.4 3 1.5 mm3 (Matrix, 256 3 256; Slice Thickness, 1.5 mm), Respectively. All the Voxel Sizes Were Close to Cubic Shape. (D) The Anatomical Reference From T1WI fat Saturation Sequence at the Same Level of the DTI map. The SNR Decreased With Increasing Spatial Resolution on DTI Maps. Compared With the Normal Breast Tissue Shown in (D), the Architectural Distortion and Visible Breast Volume Reduction Became Severe at Spatial Resolutions of 1.9 3 1.9 3 2 mm3 or 1.4 3 1.4 3 1.5 mm3

Abbreviations: DTI ¼ diffusion tensor imaging; SNR ¼ signal to noise ratio; T1WI ¼ T1 weighted imaging.

66

-

the final SNR was improved. The cubic shapes facilitate 3-D reconstruction, which will bring the additional benefits to image postprocessing. Our results showed that the cubic voxels obtained the best tractography effect. We also tried to improve the spatial resolution to 1.9  1.9  2 mm or 1.4  1.4  1.5 mm, but the image quality decreased markedly, the visible breast volume was clearly reduced, and the distortion became severe (Fig. 4). Therefore, we speculate that a 3-mm slice thickness might be the limit for breast tractography in current conditions using 1.5 T MR scanners. Higher SNR and better image quality are expected from higher magnetic fields. In addition, we found that the 31-direction mode could track integral fibers that were distributed along the mammary segment. Fibers acquired from the 6- and 15-direction modes were not satisfactory because of the shorter length and lesser volume. These results are in agreement with a previous study reporting that at least 30 directions were required for a robust estimation of tensor orientation.20 It is known that more directions help obtain detailed and precise information for fiber tracking.21,22 However, more directions have the disadvantages of requiring longer time, which might increase the danger of motion artifacts ruining the DTI image quality. In this study, we found that in breast DTI, the SNR was different when the b value and spatial resolution were adjusted, but there was no difference in SNR when NDGD was adjusted. FA value is a crucial factor for tractography, and it was different when NDGD was adjusted, but not when b value and spatial resolution were adjusted. Conversely, in breast DTI, b value and spatial resolution affected

Clinical Breast Cancer February 2014

the breast tractography via SNR, and NDGD influenced the reconstructed effect via FA value. The improvement of the SNR could correct the deviation of tensor parameters such as FA.17,23 However, in our study, we did not find the relationship between SNR and FA value in breast DTI. This might be because of the small size of the sample we collected in this study. Several limitations of this study should be mentioned. First, the subjects were not rigorously selected in terms of breast glandular type, age, and menstrual cycle, which might cause statistical variability among individuals. Second, the study included 3 groups of different volunteers to investigate each scanning parameter. The results should be more comparable if the same group of volunteers is used for each optimization step. Third, every volunteer might receive at least 3 to 4 scanning sequences during the examination; the sequencing order might cause differences beyond each acquisition parameter. Fourth, validation of fiber tracking in breast tissue is difficult in living subjects because histological confirmation is unavailable.

Conclusion In this study we optimized acquisition parameters for breast DTI tractography using a 1.5 T MR system: cubic voxel facilitated the tractography effect and 3 mm was the limit slice thickness for breast DTI reconstructed tractography. To obtain continuous breast fiber bundles, the number of diffusion gradient directions should increase to 31. Our results demonstrate that breast DTI tractography is technically feasible using a 1.5 T scanner.

Yuan Wang et al Clinical Practice Points  Diffusion anisotropy of the breast parenchyma has been per-

formed and the repeatability of diffusion tensor imaging (DTI) quantitative parameters Davg (Average diffusion coefficient) and FA (fractional anisotropy ) value had been assessed in latest literatures. Currently, DTI is mainly used for the differential diagnosis of benign and malignant breast masses. In fact, DTI in other part of body, such as the brain, has helped the treatment of brain tumor and neuropsychological disorder. DTI fiber tracking is a new field for breast imaging. There is no report about breast DTI fibertractography.  In this paper we optimized acquisition parameters for breast DTI tractography and obtained continuous breast fiber bundles using a 1.5 T MR system. We shared valuable experiences of breast DTI reconstructed tractography, such as cubic voxel shape, the limit slice thickness (3 mm for 1.5T) and the proper number of diffusion gradient directions for use.  Although the validation of breast fiber tractography has not been performed in vivo by histology, our results demonstrate the anisotropy of normal breast in the way of fiber tractography. In future, breast fiber bundles may help establish the relationship with breast duct system and fibrous support structures. That may facilitate the detection and therapy of breast cancer.

Acknowledgments This study was supported by fund from the National Basic Research Program of China (973 Program) (No. 2011CB707705), Beijing Health System High Level Health Technical Personnel Training Plan (No. 2011-2-21) and the Capital Characteristic Clinical Application Research (No. Z121107001012115).

Disclosure The authors have stated that they have no conflicts of interest.

References 1. Tournier JD, Calamante F, King MD, et al. Limitations and requirements of diffusion tensor fiber tracking: an assessment using simulations. Magn Reson Med 2002; 47:701-8. 2. Peters JM, Sahin M, Vogel-Farley VK, et al. Loss of white matter microstructural integrity is associated with adverse neurological outcome in tuberous sclerosis complex. Acad Radiol 2012; 19:17-25.

3. Ellingson BM, Cloughesy TF, Lai A, et al. High order diffusion tensor imaging in human glioblastoma. Acad Radiol 2011; 18:947-54. 4. Hasan KM, Walimuni IS, Abid H, et al. Multi-modal quantitative MRI investigation of brain tissue neurodegeneration in multiple sclerosis. J Magn Reson Imaging 2012; 35:1300-11. 5. Baltzer PA, Schafer A, Dietzel M, et al. Diffusion tensor magnetic resonance imaging of the breast: a pilot study. Eur Radiol 2011; 21:1-10. 6. Partridge SC, Ziadloo A, Murthy R, et al. Diffusion tensor MRI: preliminary anisotropy measures and mapping of breast tumors. J Magn Reson Imaging 2010; 31:339-47. 7. Eyal E, Shapiro-Feinberg M, Furman-Haran E, et al. Parametric diffusion tensor imaging of the breast. Invest Radiol 2012; 47:284-91. 8. Partridge SC, Murthy RS, Ziadloo A, et al. Diffusion tensor magnetic resonance imaging of the normal breast. Magn Reson Imaging 2010; 28:320-8. 9. Liacu D, Idy-Peretti I, Ducreux D, et al. Diffusion tensor imaging tractography parameters of limbic system bundles in temporal lobe epilepsy patients. J Magn Reson Imaging 2012; 36:561-8. 10. Mukherjee P, Chung SW, Berman JI, et al. Diffusion tensor MR imaging and fiber tractography: technical considerations. AJNR Am J Neuroradiol 2008; 29: 843-52. 11. Choi SI, Kang JW, Chun EJ, et al. High-resolution diffusion tensor MR imaging for evaluating myocardial anisotropy and fiber tracking at 3T: the effect of the number of diffusion-sensitizing gradient directions. Korean J Radiol 2010; 11:54-9. 12. Lee JW, Kim JH, Kang HS, et al. Optimization of acquisition parameters of diffusion-tensor magnetic resonance imaging in the spinal cord. Invest Radiol 2006; 41:553-9. 13. Baron P, Dorrius MD, Kappert P, et al. Diffusion-weighted imaging of normal fibroglandular breast tissue: influence of microperfusion and fat suppression technique on the apparent diffusion coefficient. NMR Biomed 2010; 23:399-405. 14. Thoeny HC, De Keyzer F, Boesch C, et al. Diffusion-weighted imaging of the parotid gland: Influence of the choice of b-values on the apparent diffusion coefficient value. J Magn Reson Imaging 2004; 20:786-90. 15. Woodhams R, Matsunaga K, Iwabuchi K, et al. Diffusion-weighted imaging of malignant breast tumors: the usefulness of apparent diffusion coefficient (ADC) value and ADC map for the detection of malignant breast tumors and evaluation of cancer extension. J Comput Assist Tomogr 2005; 29:644-9. 16. Takanaga M, Hayashi N, Miyati T, et al. Influence of b value on the measurement of contrast and apparent diffusion coefficient in 3.0 Tesla breast magnetic resonance imaging [in Japanese]. Nihon Hoshasen Gijutsu Gakkai Zasshi 2012; 68:201-8. 17. Ding Z, Gore JC, Anderson AW. Reduction of noise in diffusion tensor images using anisotropic smoothing. Magn Reson Med 2005; 53:485-90. 18. Andreisek G, White LM, Kassner A, et al. Diffusion tensor imaging and fiber tractography of the median nerve at 1.5T: optimization of b value. Skeletal Radiol 2009; 38:51-9. 19. Tagliafico A, Rescinito G, Monetti F, et al. Diffusion tensor magnetic resonance imaging of the normal breast: reproducibility of DTI-derived fractional anisotropy and apparent diffusion coefficient at 3.0 T. Radiol Med 2012; 117:992-1003. 20. Jones DK. The effect of gradient sampling schemes on measures derived from diffusion tensor MRI: a Monte Carlo study. Magn Reson Med 2004; 51:807-15. 21. Kim CK, Jang SM, Park BK. Diffusion tensor imaging of normal prostate at 3 T: effect of number of diffusion-encoding directions on quantitation and image quality. Br J Radiol 2012; 85:e279-83. 22. Yoshikawa T, Aoki S, Abe O, et al. Diffusion tensor imaging of the brain: effects of distortion correction with correspondence to numbers of encoding directions. Radiat Med 2008; 26:481-7. 23. 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.

Clinical Breast Cancer February 2014

- 67