Women's Imaging
Original Research
Effect of IV Administration of a Gadolinium-Based Contrast Agent on Breast Diffusion-Tensor Imaging
Audio Available | 
Share
OBJECTIVE. The purpose of this study was to quantify changes in diffusion-tensor imaging (DTI) parameters before and after IV administration of a gadolinium-based contrast agent (GBCA) and explore the influence of those parameters on breast cancer diagnosis.
SUBJECTS AND METHODS. A prospective cohort of 26 women with BI-RADS categories 0, 4, 5, or 6 underwent 3-T breast MRI with sequential DTI before GBCA administration and immediately after. Quantitative image analysis using dedicated DTI software yielded parametric DTI maps of each directional diffusion coefficient (DDC), mean diffusivity, and maximal anisotropy of the lesions and normal tissue. The color maps were evaluated and the lesion DTI parameters were compared before and after GBCA administration using appropriate statistical tests.
RESULTS. Of the cohort, 58% had cancer (13 infiltrating ductal carcinoma, two ductal carcinoma in situ) and 42% had benign or normal results. All breast cancers were visually detected in the DDC λ1 maps before and after GBCA administration. Mean cancer size derived from λ1 maps before GBCA administration was 15.3 mm (range, 3.3–72.3 mm), and was not statistically significantly different from the size derived after GBCA administration of 17.3 mm (range, 3.9–71.0 mm). After GBCA administration, the cancers exhibited statistically significantly lower DDCs, mean diffusivity, and b0 intensity (p < 0.05), and no change in maximal anisotropy compared with before GBCA administration, whereas these parameters in normal and benign lesions did not change significantly after GBCA administration. The mean AUC values before and after GBCA administration, ranging from 0.735 to 0.985 and from 0.867 to 0.990, respectively, were not statistically significantly different for all parameters aside from λ3.
CONCLUSION. Diagnostic accuracy using DTI was equivalent before and after GBCA administration, despite a change in the values of the DTI parameters. However, the limitations in standardization of contrast enhancement implies that unenhanced diffusion measurements should be preferred.
Keywords: breast cancer, diffusion tensor magnetic resonance imaging, DTI, gadolinium
The detection of a breast lesion using dynamic contrast-enhanced MRI (DCE-MRI) after IV administration of gadolinium-based contrast agents (GBCAs) depends on augmentation of lesion vascularity, vascular permeability, and cell density relative to normal breast tissue [1]. Quantitative analysis of the enhancement kinetics of the contrast agent and qualitative assessment of the visible morphologic features provide high diagnostic accuracy of breast cancer screening tests [2].
In recent years DWI has been increasingly performed in breast MRI as an adjunct tool to DCE-MRI [3–6]. This technique applies diffusion gradients along three orthogonal directions and quantifies a mean diffusion coefficient, termed “apparent diffusion coefficient” (ADC), over the three directions, Breast DWI meta-analysis diagnostic results have exhibited sensitivities ranging from 84% to 91% and specificities ranging from 75% to 84%. Another diffusion technique, diffusion-tensor imaging (DTI), applies diffusion gradients in multiple directions (≥ 6) and extends the averaged information from DWI to obtain symmetric tensor metrics that quantify water diffusion in restricted or hindered environments. DTI analysis is a multiparametric method that yields three directional diffusion coefficients (DDCs), anisotropy indexes, and mean diffusivity, as well as tracks microstructural elements such as brain white matter fibers [7, 8]. The diffusion of water molecules in the mammary tissue, composed of ductal and glandular elements, presents a particular example of restricted movement, yielding DTI metrics that serve to identify breast malignancy [9–11]. Indeed, in a recent meta-analysis, breast DTI studies achieved a higher accuracy than did ADC measured in DWI, although the number of studies to date is limited. DWI and DTI can be applied at any time during the menstrual cycle because they are not affected by hormonal changes [12, 13].
Very little is known about the effects of the paramagnetic GBCA applied in the standard breast DCE-MRI protocol on the measured values of the water diffusion parameters in normal and malignant breast tissue. The reciprocated effects of GBCA administration on DWI performance yielded conflicting results, showing either significant reduction of the cancer ADC and an increase in contrast-to-noise ratio on contrast-enhanced images relative to unenhanced images, or no significant difference between the cancer ADC before and after contrast enhancement and a slight increase in the normal tissue ADC contrast-enhancement [14–16].
Furthermore, the effects of GBCA on DTI performance have not yet been investigated. Consequently, whether to initially perform the DCE-MRI protocol to reach diagnosis and then apply a diffusion protocol, or to apply the diffusion protocol before DCE-MRI to avoid contrast material effects, is still an open question. To optimize and standardize breast MRI that includes diffusion, it is important to clarify the effects of IV GBCA administration on the values and the diagnostic efficiency of the diffusion parameters. Thus, the purpose of this article is to quantify the changes in the DTI parameters before and after administration of a GBCA and explore the influence of those parameters on breast cancer diagnosis.
| Subjects and Methods |   |
|---|
The institutional research board of the University of Toronto approved this pilot study in which the 26 women scheduled for a diagnostic breast MRI with BI-RADS categories 0, 4, 5, or 6 on conventional breast imaging were twice scanned using the same DTI sequence before and immediately after the breast DCE-MRI, after signing a consent form at the medical imaging department.
In cases in which a benign or malignant lesion was found, a histopathologic diagnosis was obtained using core needle biopsy before or after the MRI examination or after surgical treatment when clinically indicated.
Breast MRI was performed on a 3-T scanner (Skyra-Fit, Siemens Healthineers) equipped with a 16-channel breast coil (Sentinelle, Invivo). The MRI protocol included T1-weighted and T2-weighted imaging without and with fat saturation; DTI with a spin-echo echo-planar-imaging sequence (1.875 × 1.875 × 2.4 mm3 resolution; transverse slices; TE/TR, 86/12,600; b values, 0 and 700 s/mm2; 30 diffusion-gradient directions) and a DCE-MRI protocol (3D gradient-echo; transverse slices; TE/TR/flip angle, 1.72/3.86/18°; 0.72 × 0.72 × 1.2 mm3 or 1.1 × 0.8 × 1.1 mm3 resolution; each acquisition, ≈ 60 seconds), one before and five after IV gadobutrol administration (0.1 mmol/kg at 2.0 mL/s with 20 mL saline flush). The DTI protocol was performed twice: before and immediately after dynamic contrast enhancement (with no gap between the sequences at ≈ 6 minutes after the IV contrast agent administration). Slice thickness in the T1-weighted and T2-weighted unenhanced images and DTI protocols was identical (2.4 mm).
DTI datasets were analyzed using a proprietary software described in an earlier study [17], step by step applying the approach detailed by Basser and Jones [7]. The analysis yielded, for each pixel, three eigenvectors and their corresponding eigenvalues (λ1, λ2, λ3) termed “directional diffusion coefficients,” the mean diffusivity, and the maximal anisotropy index (λ1–λ3). Color-coded DTI parametric maps were overlaid on the corresponding images of b values of 700 to a b value of 0 (b0) or on the anatomic images of T1-weighted and T2-weighted images.
All DCE-MRI studies were postprocessed on a computerized viewing platform (MultiView Software, Hologic) and reviewed in consensus by two board-certified and breast fellowship–trained radiologists, each with more than 10 years of experience in breast MRI, to characterize the index breast lesions. The two readers that identified the lesions on DCE-MRI also detected the lesions on the corresponding DTI color-coded images before and after GBCA administration. Thereafter, three experienced breast readers trained in DTI and DWI MRI independently evaluated the qualitative assessment of the index cancers on DTI color-coded λ1 maps before and after GBCA administration. Quantitative assessment of each malignant lesion and the ipsilateral and contralateral normal breast tissue was performed by two DTI-trained readers. These latter readers delineated the ROIs on the color-coded λ1 map of a central lesion slice that included all pixel-exhibiting reduced λ1 values (λ1 < 1.7 × 10–3 mm2/s) as shown in Figure 1. This upper threshold of λ1 value was used according to previous results that applied the same DTI protocol at 3 T and the same processing tool [11, 12]. For each ROI obtained on the λ1 maps, the software calculated the median and interquartile range of all the DTI parameters and of the respective intensity values in the images recorded at b0. The lesion's size was estimated on the λ1 maps; the longest diameter in a central lesion slice was measured.
 View larger version (281K) | Fig. 1A —61-year-old woman with grade 3 invasive ductal carcinoma of no special type and ductal carcinoma in situ mass lesion. A, Axial dynamic contrast-enhanced subtracted image (A) and directional diffusion coefficient λ1 parametric maps (B and C) overlaid on T1-weighted image of central slice with unenhanced (B) and contrast-enhanced (C) administration are shown. There is substantial contrast in λ1 map between normal tissue (purple, with λ1 ≥ 1.7 × 10−3 mm2/s) and cancer (green-yellow-blue, λ1 < 1.2 × 10–3 mm2/s). Lesion is manually delineated in B and C. Units in B and C are in mm2/s × 10–3. |
 View larger version (212K) | Fig. 1B —61-year-old woman with grade 3 invasive ductal carcinoma of no special type and ductal carcinoma in situ mass lesion. B, Axial dynamic contrast-enhanced subtracted image (A) and directional diffusion coefficient λ1 parametric maps (B and C) overlaid on T1-weighted image of central slice with unenhanced (B) and contrast-enhanced (C) administration are shown. There is substantial contrast in λ1 map between normal tissue (purple, with λ1 ≥ 1.7 × 10−3 mm2/s) and cancer (green-yellow-blue, λ1 < 1.2 × 10–3 mm2/s). Lesion is manually delineated in B and C. Units in B and C are in mm2/s × 10–3. |
 View larger version (223K) | Fig. 1C —61-year-old woman with grade 3 invasive ductal carcinoma of no special type and ductal carcinoma in situ mass lesion. C, Axial dynamic contrast-enhanced subtracted image (A) and directional diffusion coefficient λ1 parametric maps (B and C) overlaid on T1-weighted image of central slice with unenhanced (B) and contrast-enhanced (C) administration are shown. There is substantial contrast in λ1 map between normal tissue (purple, with λ1 ≥ 1.7 × 10−3 mm2/s) and cancer (green-yellow-blue, λ1 < 1.2 × 10–3 mm2/s). Lesion is manually delineated in B and C. Units in B and C are in mm2/s × 10–3. |
Median values of the DTI parameters within the ROI were calculated for each patient, who was scanned consecutively with the same DTI protocol for unenhanced and contrast-enhanced imaging. A paired two-tailed t test was applied for statistical analysis comparing diffusion parameters and lesion size on unenhanced and contrast-enhanced scans. In such a repeated-measure design, each patient is used as her own control and individual differences are partialled out of the error term. ROC curve analysis was applied to investigate the ability of each DTI parameter to differentiate cancer from normal fibroglandular tissue before and after GBCA administration. An ROC AUC was generated to summarize the accuracy of the ROC curve. The DeLong [18] test was applied to compare the ROC curves before and after contrast administration and between the AUC values obtained for each of the two readers. The agreement between the three readers that evaluated the qualitative assessment of the index cancers on DTI color-coded λ1 maps before and after GBCA administration was assessed by calculating the kappa coefficients [19]. The statistical analyses were performed using MedCalc (v.19, Mariakerke) and Excel 2016 (Microsoft) software. A p value of ≤ 0.05 was considered statistically significant.
| Results | |
|---|
The study population consisted of 26 women, and none were excluded for poor quality of the diffusion-weighted scans. The mean age of all patients was 48.62 ± 8.43 (SD) years old (ranging from 37 to 69 years). Overall, a total of 20 lesions were assessed, of which 15 were biopsy-proven malignant lesions (13 infiltrating ductal carcinoma, two ductal carcinoma in situ [DCIS]) and five were benign lesions (one papilloma, two fibroadenoma, and two fat necrosis). The median lesion size on either unenhanced DTI or contrast-enhanced DTI was similar (p = 0.35): 15.3 mm (range, 3.3–72.3 mm) and 17.3 mm (range, 3.9–71.0 mm), respectively. One small DCIS lesion (histopathologic size, 2.5 mm, measured after mastectomy) was excluded from the quantitative analysis. Six patients without any pathologic MRI findings (four BI-RADS category 0 and two originally BI-RADS category 6 with pathologic complete response to presurgical chemotherapy and after surgical excision) were diagnosed as having normal MRI results.
The color-coded DTI parametric maps of the DDCs, mean diffusivity, and anisotropy indexes before and after contrast enhancement revealed the presence of malignancy by showing a reduction in the values of these parameters in malignant regions compared with normal breast tissue (Figs. 2 and 3). Close visual inspection of the DTI parametric maps of the cancers revealed reduction in the DTI parameters and b0 intensity after GBCA administration relative to before GBCA administration (Figs. 2 and 3).
 View larger version (79K) | Fig. 2A —47-year-old woman with newly diagnosed grade 3 invasive ductal carcinoma of no special type and in situ carcinoma. A, Axial image with b value of 0 (b0, A) and diffusion-tensor imaging parametric maps overlaid on b0 image (B–F) before (left) and after (right) contrast-enhancement are shown. λ1, λ2, and λ3 are directional diffusion coefficients (DDCs). There is reduction in intensity in b0 image and in DDCs in contrast-enhanced images relative to unenhanced images and retention of contrast material in dynamic contrast-enhanced (DCE) image. |
 View larger version (68K) | Fig. 2B —47-year-old woman with newly diagnosed grade 3 invasive ductal carcinoma of no special type and in situ carcinoma. B, Axial image with b value of 0 (b0, A) and diffusion-tensor imaging parametric maps overlaid on b0 image (B–F) before (left) and after (right) contrast-enhancement are shown. λ1, λ2, and λ3 are directional diffusion coefficients (DDCs). There is reduction in intensity in b0 image and in DDCs in contrast-enhanced images relative to unenhanced images and retention of contrast material in dynamic contrast-enhanced (DCE) image. |
 View larger version (70K) | Fig. 2C —47-year-old woman with newly diagnosed grade 3 invasive ductal carcinoma of no special type and in situ carcinoma. C, Axial image with b value of 0 (b0, A) and diffusion-tensor imaging parametric maps overlaid on b0 image (B–F) before (left) and after (right) contrast-enhancement are shown. λ1, λ2, and λ3 are directional diffusion coefficients (DDCs). There is reduction in intensity in b0 image and in DDCs in contrast-enhanced images relative to unenhanced images and retention of contrast material in dynamic contrast-enhanced (DCE) image. |
 View larger version (70K) | Fig. 2D —47-year-old woman with newly diagnosed grade 3 invasive ductal carcinoma of no special type and in situ carcinoma. D, Axial image with b value of 0 (b0, A) and diffusion-tensor imaging parametric maps overlaid on b0 image (B–F) before (left) and after (right) contrast-enhancement are shown. λ1, λ2, and λ3 are directional diffusion coefficients (DDCs). There is reduction in intensity in b0 image and in DDCs in contrast-enhanced images relative to unenhanced images and retention of contrast material in dynamic contrast-enhanced (DCE) image. |
 View larger version (66K) | Fig. 2E —47-year-old woman with newly diagnosed grade 3 invasive ductal carcinoma of no special type and in situ carcinoma. E, Axial image with b value of 0 (b0, A) and diffusion-tensor imaging parametric maps overlaid on b0 image (B–F) before (left) and after (right) contrast-enhancement are shown. λ1, λ2, and λ3 are directional diffusion coefficients (DDCs). There is reduction in intensity in b0 image and in DDCs in contrast-enhanced images relative to unenhanced images and retention of contrast material in dynamic contrast-enhanced (DCE) image. |
 View larger version (67K) | Fig. 2F —47-year-old woman with newly diagnosed grade 3 invasive ductal carcinoma of no special type and in situ carcinoma. F, Axial image with b value of 0 (b0, A) and diffusion-tensor imaging parametric maps overlaid on b0 image (B–F) before (left) and after (right) contrast-enhancement are shown. λ1, λ2, and λ3 are directional diffusion coefficients (DDCs). There is reduction in intensity in b0 image and in DDCs in contrast-enhanced images relative to unenhanced images and retention of contrast material in dynamic contrast-enhanced (DCE) image. |
 View larger version (212K) | Fig. 2G —47-year-old woman with newly diagnosed grade 3 invasive ductal carcinoma of no special type and in situ carcinoma. G, Six-minute DCE-MRI subtracted image of central slice is shown. Units are normalized to highest intensity in subtracted image, whereas mean percent lesion enhancement in this slice (intensity of subtracted image normalized to unenhanced intensity) is 228%. |
 View larger version (77K) | Fig. 3A —40-year-old woman with grade 2 invasive ductal carcinoma of no special type and in situ carcinoma. Inserts show enlarged area of lesion. A, Axial image with b value of 0 (b0, A) and directional diffusion coefficient parametric maps (λ1, λ2, and λ3; B–D) overlaid on b0 image before (left) and after (right) contrast enhancement are shown. |
 View larger version (60K) | Fig. 3B —40-year-old woman with grade 2 invasive ductal carcinoma of no special type and in situ carcinoma. Inserts show enlarged area of lesion. B, Axial image with b value of 0 (b0, A) and directional diffusion coefficient parametric maps (λ1, λ2, and λ3; B–D) overlaid on b0 image before (left) and after (right) contrast enhancement are shown. |
 View larger version (61K) | Fig. 3C —40-year-old woman with grade 2 invasive ductal carcinoma of no special type and in situ carcinoma. Inserts show enlarged area of lesion. C, Axial image with b value of 0 (b0, A) and directional diffusion coefficient parametric maps (λ1, λ2, and λ3; B–D) overlaid on b0 image before (left) and after (right) contrast enhancement are shown. |
 View larger version (60K) | Fig. 3D —40-year-old woman with grade 2 invasive ductal carcinoma of no special type and in situ carcinoma. Inserts show enlarged area of lesion. D, Axial image with b value of 0 (b0, A) and directional diffusion coefficient parametric maps (λ1, λ2, and λ3; B–D) overlaid on b0 image before (left) and after (right) contrast enhancement are shown. |
 View larger version (211K) | Fig. 3E —40-year-old woman with grade 2 invasive ductal carcinoma of no special type and in situ carcinoma. Inserts show enlarged area of lesion. E, Six-minute dynamic contrast-enhanced subtracted image (E) of central slice is shown. Units are normalized to highest intensity in subtracted image, whereas mean percent lesion enhancement in this slice (intensity of subtracted image normalized to unenhanced intensity) is 270%. |
Inspection of the DCE-MRI subtracted images of cancer lesions indicated retention of enhancement at the late phase of contrast administration, just before the start of the contrast-enhanced DTI protocol (Figs. 2G and 3E). The agreement between the qualitative visual inspection of the three readers was perfect (κ = 1.0), indicating that all cancers depicted in the unenhanced DTI images were also shown in the contrast-enhanced DTI images, with a high level of confidence for the radiologists' visual assessment.
Quantitative analysis of the DTI parameters yielded the medians and interquartile ranges of DDCs, mean diffusivity, anisotropy indexes, and b0 intensity of the malignant lesions and of the contralateral and ipsilateral normal breast tissue, before and after GBCA administration (Table 1). Patient-by-patient comparison of these parameters showed that in malignant tissue, GBCA administration lead to a statistically significant reduction in the values of the DDCs, mean diffusivity, and b0 intensity (p < 0.05), indicating an effect of the accumulated contrast agent on these parameters. In contrast, the maximal anisotropy index, which reflects tissue microstructure, was not significantly affected by GBCA accumulation (p = 0.74) (Table 1). In normal breast tissue, there were no statistically significant alterations (p in the range of 0.20–1.00) in all the DTI parameters and in b0 intensity, suggesting negligible breast parenchymal enhancement (Table 1). The results derived from analysis of the normal breast tissue in cancer patients were further confirmed by analyzing patients with no indication of malignancy, which showed no statistically significant changes (p ranging from 0.73 to 0.97) of the DTI parameters before and after GBCA administration (Table 2). Furthermore, no significant change was detected (p in the range of 0.08 to 0.54) in the DTI parameters before and after GBCA administration of the benign lesions (n = 5), suggesting minor accumulation of contrast agent in this small cohort of benign lesions (Table 2).
Evaluation of the clinical efficiency for detecting cancer before and after contrast enhancement according to ROC curve analysis indicated high AUC values for all DTI parameters, with λ1 exhibiting the highest AUC value (Table 3). The DeLong test between the AUC values for unenhanced and contrast-enhanced DTI parameters showed no statistically significant difference for all DTI parameters aside from λ3, which exhibited the lowest AUC (Table 3). Comparing the AUC values obtained by the two readers of each DTI parameter before and after contrast enhancement indicated no statistically significant difference between the readers (p > 0.05).
| Discussion | |
|---|
In this prospective study, we investigated the capability of the various diffusion coefficients and anisotropy indexes obtained from DTI at 3 T, before and after GBCA administration, to differentiate between malignant and normal breast tissue. Quantitative analysis of the results confirmed reduction in the DDCs and in the maximal anisotropy index in malignant lesions compared with normal breast tissue or benign lesions. Furthermore, the values of cancer DDCs were found to be statistically significantly lower in the contrast-enhanced study compared with the un-enhanced study, whereas GBCA administration had no statistically significant effect on the DTI parameters in normal tissue or benign lesions. However, the clinical cancer detection efficiency indicated similar high AUCs and hence detection efficiency of the most clinically relevant DTI parameters before and after contrast enhancement. Nevertheless, to achieve standardization, diffusion protocols should be performed without enhancement to avoid effects of GBCA administration that depend on the relaxivity and dose of the GBCA, the starting time of the contrast-enhanced diffusion protocol, and the variable vascular properties of lesions.
The origin of the reduction in the cancer DDCs and not in the normal tissue with contrast enhancement is most likely related to the changes in the T2 relaxation rate in the presence of the contrast agent (i.e., T2 relaxivity of gadobutrol in water is 3.9 s–1 mmol–1 L [20]). Generally, the spin-echo amplitude in a diffusion MRI experiment is normalized to the echo amplitude at a b value of 0, and this normalization removes the dependence on the T2 relaxation rate, assuming that all spins share the same T2 relaxation rate in the course of the pulse sequence. The accumulation of the GBCA in cancer lesions increases the effective T2 relaxation rate and causes a change in the contrast agent concentration during the contrast-enhanced diffusion image acquisition. Consequently, the normalization at the high b values (for the 30 directions of the diffusion gradients) relative to null b value (one scan with no diffusion gradient) is altered, leading to a reduction in the diffusion coefficients compared with un-enhanced conditions with a stable T2 relaxation rate during the entire DTI acquisition. Because less GBCA enters normal breast tissue and benign lesions compared with cancer tissue, the change in the T2 relaxation rate is negligible and does not alter significantly their diffusion coefficients from unenhanced to contrast-enhanced scans.
The GBCA concentration changes in cancers lead to a reduction in the three DDCs, but the maximal anisotropy index, λ1–λ3, which reflects tissue microstructure, does not change significantly, indicating no GBCA-induced structural alterations [11, 21, 22]. Although the diffusion coefficients after GBCA administration appeared to be lower in cancers, the clinical evaluation by ROC curve analysis indicated no statistically significant difference between AUCs in unenhanced and contrast-enhanced DTI for the most relevant parameters, except for λ3, which exhibited the lowest AUC for differentiating cancer from normal breast tissue (Table 3).
In a previous 1.5-T study by Yuen et al. [14], a significant reduction of cancer ADC and an increase in contrast-to-noise ratio in contrast-enhanced images relative to unenhanced images were reported. These results were attributed to the high cancer microvascularization, suggesting a suppression in the presence of GBCA of the microperfusion-induced elevation of the ADC [14]. Another explanation was proposed by Ramadan and Mulkern [23], who suggested that GBCAs induce local magnetic susceptibilities that affected the diffusion sensitization gradients and thereby altered the measured diffusion coefficients. Further studies by Janka et al. [15] also showed a significant reduction of ADC contrast enhancement in cancers and no significant change in benign lesions. However, a conflicting result was obtained by Nguyen et al. [16], in which no significant difference in unenhanced and contrast-enhanced ADC was detected and there was a slight increase in contrast-enhanced ADC in normal tissue. This discrepancy could stem from differences in the contrast agent, administration procedures, timing of the contrast-enhanced diffusion measurements, and differences in the DWI protocol. Our findings agree with those showing a reduction of the contrast agent on the calculated diffusion coefficients in breast malignant tissue, suggesting a T2 effect; however, we cannot exclude an additional source for this reduction caused by GBCA-induced field gradients.
There was full agreement among readers on the qualitative visual assessment (κ = 1.0), indicating that the visualization of the cancer on λ1 maps is independent of the contrast agent injection. Therefore, cancer lesions can be detected by the radiologist with either unenhanced images or contrast-enhanced images. This clinical impact was further shown by the comparable AUCs of DTI parameters obtained by the two readers derived from the quantitative analyses. Overall, the results indicated that equivalent clinical efficiency of DTI is achieved in the evaluation of both un-enhanced and contrast-enhanced images.
In general, diffusion MRI sequences may suffer from technical limitations caused by echo-planar-imaging (eddy current and susceptibility distortions and artifacts) and by inhomogeneous fat suppression that impair the quality of the images [24]. However, none of the cases scanned in this study were excluded for inadequate quality, and all cases were amenable for computer-aided image processing. We excluded from the analysis one 2.5-mm DCIS. The total number of lesions was small, yet it provided adequate statistical power for this study because the comparison was according to two consecutively recorded datasets with the same protocol. Other contrast agents could yield different quantitative results. However, because all small GBCAs affect the signal by the same mechanism and exhibit similar relaxivities and pharmacokinetics [20], the trend should be the same.
In conclusion, the results of this study show statistically significant reduction in the DDCs of breast cancers after dynamic contrast enhancement compared with before dynamic contrast enhancement, leading to equivalent or increased conspicuity of the cancer lesions in the visual assessment by the radiologist. However, the accuracy of cancer detection using unenhanced or contrast-enhanced DTI datasets was not significantly different. Furthermore, the use of DTI before dynamic contrast enhancement can be fully standardized, because the DTI parameters are intrinsic tissue characteristics. The contrast-enhanced DTI parameters are less amenable to standardization because their values depend on the type and dose of contrast agent and on variations in the contrast agent dynamics among the various types of breast cancers. Consequently, because of the lack of standardization of contrast-enhanced DTI, unenhanced diffusion measurements should be preferred.
Based on a presentation at the Joint International Society on Magnetic Resonance in Medicine-European Society for Magnetic Resonance in Medicine 2018 annual meeting, Paris, France.
Supported in part by peer-reviewed grant JDMI-2016 AIF from University Medical Imaging Consultants to A. Scaranelo.
| References | |
|---|
