Technical Report
Readout-segmented Echo-planar Imaging for Diffusion-weighted Imaging in the Pelvis at 3T—A Feasibility Study Yee Liang Thian, MBBS, FRCR, MMed, Wanying Xie, MBBS, FRCR, MMed, David A. Porter, PhD, Bertrand Weileng Ang, MBBS, FRCR, MMed Rationale and Objectives: Diffusion-weighted imaging (DWI) of the pelvis at 3T is prone to artifacts that diminish the image quality. Readout-segmented echo-planar imaging (RS-EPI) is a new DWI technique that can reduce the artifacts associated with standard single-shot echo-planar imaging (SS-EPI) DWI. The purpose of this study was to evaluate the feasibility and image quality of RS-EPI in pelvic DWI compared to SS-EPI on a 3T imaging system. Materials and Methods: Thirty patients underwent pelvic DWI on a 3T scanner with SS-EPI and RS-EPI techniques. Two blinded readers independently assessed each set of images for geometric distortion, image blurring, ghosting artifacts, lesion conspicuity, and overall image quality on a 7-point scale. Qualitative image scores were compared using paired Wilcoxon signed rank test. Interreader correlation was assessed by Spearman rank correlation. Results: Geometric distortion, imaging blurring, ghosting artifacts, lesion conspicuity, and overall image quality were rated significantly better by both readers for RS-EPI technique (P < .01 for all parameters). There was moderate–high correlation between the readers (r = 0.649–0.752) for all parameters apart from lesion conspicuity (r = 0.351). Both readers preferred the RS-EPI set of DWI images in most of the cases (reader 1: 0.87, 95% CI 0.74–0.99; reader 2: 0.77, 95% CI 0.61–0.93). Mean difference and limits of agreement between apparent diffusion coefficient (ADC) values obtained from the two methods were 0.01 (0.08, 0.10) 103 mm2/s. Conclusions: RS-EPI DWI images showed improved image quality compared to SS-EPI technique at 3T. RS-EPI is a feasible technique in the pelvis for producing high-resolution DWI. Key Words: Readout-segmented; diffusion magnetic resonance imaging; echo-planar imaging; pelvis; artifacts. ªAUR, 2014
D
iffusion-weighted imaging (DWI) has emerged in the past decade as an important functional imaging technique in extracranial oncologic imaging. In the pelvis, DWI has been applied to imaging of rectal, prostate, endometrial, and cervical cancers (1–6). DWI provides an excellent contrast mechanism that is useful for tumor detection and delineating disease extent. Currently the most widely used sequence for clinical DWI in the pelvis is single-shot echo-planar imaging (SS-EPI), whereby all the lines in k-space are filled by multiple gradient reversals in a single acquisition after a single radiofrequency pulse. However, the SS-EPI technique suffers from significant artifacts that Acad Radiol 2014; 21:531–537 From the Department of Diagnostic Imaging, National University Hospital Singapore, 5 Lower Kent Ridge Rd, Singapore 119074 Singapore (Y.L.T., B.W.A.); Department of Nuclear Medicine, Singapore General Hospital, Singapore, Singapore (W.X.); and Imaging and IT Division, Healthcare Sector, Siemens AG, Erlangen, Germany (D.A.P.). Received November 19, 2013; accepted January 8, 2014. Address correspondence to: Y.L.T. e-mail: [email protected] ªAUR, 2014 http://dx.doi.org/10.1016/j.acra.2014.01.005
reduce the image quality of DWI. It is vulnerable to T2*induced blurring and geometric distortions due to magnetic field inhomogeneities that cause accumulation of phase errors (7). In the pelvis, the presence of gas-containing viscera, such as the rectum, small and large bowel, and vagina, frequently exacerbates these artifacts. Distortion and blurring artifacts worsen with higher field strength and higher resolution, limiting the achievable resolution with SS-EPI before these effects become prohibitive (8). The functional image quality and resolution achieved with DWI thus presently lags considerably behind the high-resolution T2-weighted sequences used for morphologic assessment and anatomic correlation in the pelvis. One way to overcome the limitations of the SS-EPI technique is to use a readout-segmented echo-planar imaging (RS-EPI) approach (9). With this technique, k-space is acquired using multiple ‘‘shots’’ or segments and a twodimensional (2D) navigator phase correction is used to minimize the effect of motion artifact without the requirement for ECG or pulse triggering. RS-EPI allows a shorter echo spacing in the EPI echo train compared to SS-EPI which in turn theoretically allows for high-resolution diffusion531
THIAN ET AL
weighted images with low susceptibility based image distortion and T2* blurring. The technique has previously been applied mainly to intracranial pathology (10–12). In these studies, the higher spatial resolution and reduced blurring and distortion were found to be particularly useful for evaluation of areas at tissue–air interfaces, such as the skull base, sinonasal cavities, and orbits. We hypothesize that similar improvements might be seen in the pelvis. The purpose of this study was to demonstrate the feasibility of the RS-EPI technique in DWI of the pelvis and to compare the image quality of RS-EPI DWI of the pelvis with SS-EPI diffusion images on a 3T clinical scanner.
MATERIALS AND METHODS This study was approved by the institutional review board with informed consent obtained. A retrospective review of the images of 30 consecutive patients who underwent pelvic magnetic resonance (MR) imaging with both SS-EPI and RS-EPI DWI at our institution between June 1, 2012 and December 31, 2012 was performed. MR Imaging
SS-EPI and RS-EPI diffusion-weighted images were acquired on 30 patients using a 3.0T whole body MR imager (MAGNETOM Skyra; Siemens Healthcare, Erlangen, Germany) using a 18-channel surface coil in combination with a 32channel spine coil and a high-performance gradient system (gradient strength = 45mT/m, slew rate = 200 T/m/s). To reduce colonic motility, a 20 mg dose of scopolamine butlybromide was injected intramuscularly 15 minutes before MR imaging. Conventional localizer and T2-weighted sequences were obtained followed by diffusion-weighted images. DWI of the pelvis was obtained in the axial plane, with imaging parameters detailed in Table 1. The RS-EPI sequence was performed immediately after the SS-EPI sequence in all patients. The routine GRAPPA-accelerated SS-EPI sequence, a standard clinical protocol at our institution, was used. RS-EPI used a monopolar diffusion preparation with seven readout segments and one signal average, which were used by previous authors for the brain (10). The RS-EPI sequence parameters used in this study were established by the authors as an acceptable trade-off between image quality and scan time. Apparent diffusion coefficient (ADC) maps from each acquisition were generated using a monoexponential fit of the acquired b-value data points using b-values of 0 and 1000 s/mm2. Siemens Healthcare provided the prototype RS-EPI DWI sequence (RESOLVE) based on the RS-EPI technique.
Academic Radiology, Vol 21, No 4, April 2014
with 11 and 7 years of experience separately. The evaluators were blinded to subject data and MR acquisition conditions. SS-EPI and RS-EPI diffusion-weighted images and ADC maps were scored in terms of geometric distortion level, image blurring, ghosting artifacts, lesion conspicuity, and overall image quality of visualized anatomy in the pelvis. For artifacts (geometric distortion, image blurring, and ghosting), the severity of the artifact was scored on a 7-point Likert scale with a score of 1 indicating extremely severe and score of 7 indicating the images were free from the particular artifact. Lesion conspicuity and overall image quality were scored on a 7-point scale as follows: 1, unacceptable; 2, poor; 3, below average; 4, standard; 5, above average; 6, good; and 7, outstanding. The evaluators were finally asked to indicate which set of diffusion-weighted images they found preferable (if any) for clinical evaluation. To compare the absolute ADCs obtained using both DWI methods, concordant regions of interest (ROIs) of various primary lesions on the ADC maps were drawn manually by one radiologist. ROI measurements were obtained only in patients with a mass lesion evident on the scan (eg, tumor). An ROI was drawn at a representative portion of the lesion which could be clearly depicted on both DWI techniques, taking care to avoid areas of necrosis or large blood vessels identified on standard T2 or contrast-enhanced images. Circular ROIs with a minimum size of 10 mm2 were obtained. Estimates of signal-to-noise ratio (SNR) were made by taking the ratio of the mean signal intensity in an ROI of at least 25 mm2 in the left or right obturator externus muscle (raw images acquired with b = 0 s/mm2) to the standard deviation of a similar ROI in a homogenous artifact-free region of background air in close proximity to the site of the signal intensity measurement, using an approach described elsewhere (13,14). However as parameters were not normalized for scan time, SNR estimates reported here for both techniques are not directly comparable for SNR efficiency. Statistical Analysis
Statistical analysis was performed using software (Stata Statistical Software: Release 12; StataCorp LP, College Station, TX). The Wilcoxon signed rank test was used to determine significant differences between the radiologist’s ratings of the two techniques. Interreader correlation of qualitative image scores was estimated using the Spearman rank correlation test. Bland–Altman plots were used to assess agreement in ADC obtained using SS-EPI and RS-EPI techniques. A P value of
Readout-segmented Echo-planar Imaging for Diffusion-weighted Imaging in the Pelvis at 3T—A Feasibility Study Yee Liang Thian, MBBS, FRCR, MMed, Wanying Xie, MBBS, FRCR, MMed, David A. Porter, PhD, Bertrand Weileng Ang, MBBS, FRCR, MMed Rationale and Objectives: Diffusion-weighted imaging (DWI) of the pelvis at 3T is prone to artifacts that diminish the image quality. Readout-segmented echo-planar imaging (RS-EPI) is a new DWI technique that can reduce the artifacts associated with standard single-shot echo-planar imaging (SS-EPI) DWI. The purpose of this study was to evaluate the feasibility and image quality of RS-EPI in pelvic DWI compared to SS-EPI on a 3T imaging system. Materials and Methods: Thirty patients underwent pelvic DWI on a 3T scanner with SS-EPI and RS-EPI techniques. Two blinded readers independently assessed each set of images for geometric distortion, image blurring, ghosting artifacts, lesion conspicuity, and overall image quality on a 7-point scale. Qualitative image scores were compared using paired Wilcoxon signed rank test. Interreader correlation was assessed by Spearman rank correlation. Results: Geometric distortion, imaging blurring, ghosting artifacts, lesion conspicuity, and overall image quality were rated significantly better by both readers for RS-EPI technique (P < .01 for all parameters). There was moderate–high correlation between the readers (r = 0.649–0.752) for all parameters apart from lesion conspicuity (r = 0.351). Both readers preferred the RS-EPI set of DWI images in most of the cases (reader 1: 0.87, 95% CI 0.74–0.99; reader 2: 0.77, 95% CI 0.61–0.93). Mean difference and limits of agreement between apparent diffusion coefficient (ADC) values obtained from the two methods were 0.01 (0.08, 0.10) 103 mm2/s. Conclusions: RS-EPI DWI images showed improved image quality compared to SS-EPI technique at 3T. RS-EPI is a feasible technique in the pelvis for producing high-resolution DWI. Key Words: Readout-segmented; diffusion magnetic resonance imaging; echo-planar imaging; pelvis; artifacts. ªAUR, 2014
D
iffusion-weighted imaging (DWI) has emerged in the past decade as an important functional imaging technique in extracranial oncologic imaging. In the pelvis, DWI has been applied to imaging of rectal, prostate, endometrial, and cervical cancers (1–6). DWI provides an excellent contrast mechanism that is useful for tumor detection and delineating disease extent. Currently the most widely used sequence for clinical DWI in the pelvis is single-shot echo-planar imaging (SS-EPI), whereby all the lines in k-space are filled by multiple gradient reversals in a single acquisition after a single radiofrequency pulse. However, the SS-EPI technique suffers from significant artifacts that Acad Radiol 2014; 21:531–537 From the Department of Diagnostic Imaging, National University Hospital Singapore, 5 Lower Kent Ridge Rd, Singapore 119074 Singapore (Y.L.T., B.W.A.); Department of Nuclear Medicine, Singapore General Hospital, Singapore, Singapore (W.X.); and Imaging and IT Division, Healthcare Sector, Siemens AG, Erlangen, Germany (D.A.P.). Received November 19, 2013; accepted January 8, 2014. Address correspondence to: Y.L.T. e-mail: [email protected] ªAUR, 2014 http://dx.doi.org/10.1016/j.acra.2014.01.005
reduce the image quality of DWI. It is vulnerable to T2*induced blurring and geometric distortions due to magnetic field inhomogeneities that cause accumulation of phase errors (7). In the pelvis, the presence of gas-containing viscera, such as the rectum, small and large bowel, and vagina, frequently exacerbates these artifacts. Distortion and blurring artifacts worsen with higher field strength and higher resolution, limiting the achievable resolution with SS-EPI before these effects become prohibitive (8). The functional image quality and resolution achieved with DWI thus presently lags considerably behind the high-resolution T2-weighted sequences used for morphologic assessment and anatomic correlation in the pelvis. One way to overcome the limitations of the SS-EPI technique is to use a readout-segmented echo-planar imaging (RS-EPI) approach (9). With this technique, k-space is acquired using multiple ‘‘shots’’ or segments and a twodimensional (2D) navigator phase correction is used to minimize the effect of motion artifact without the requirement for ECG or pulse triggering. RS-EPI allows a shorter echo spacing in the EPI echo train compared to SS-EPI which in turn theoretically allows for high-resolution diffusion531
THIAN ET AL
weighted images with low susceptibility based image distortion and T2* blurring. The technique has previously been applied mainly to intracranial pathology (10–12). In these studies, the higher spatial resolution and reduced blurring and distortion were found to be particularly useful for evaluation of areas at tissue–air interfaces, such as the skull base, sinonasal cavities, and orbits. We hypothesize that similar improvements might be seen in the pelvis. The purpose of this study was to demonstrate the feasibility of the RS-EPI technique in DWI of the pelvis and to compare the image quality of RS-EPI DWI of the pelvis with SS-EPI diffusion images on a 3T clinical scanner.
MATERIALS AND METHODS This study was approved by the institutional review board with informed consent obtained. A retrospective review of the images of 30 consecutive patients who underwent pelvic magnetic resonance (MR) imaging with both SS-EPI and RS-EPI DWI at our institution between June 1, 2012 and December 31, 2012 was performed. MR Imaging
SS-EPI and RS-EPI diffusion-weighted images were acquired on 30 patients using a 3.0T whole body MR imager (MAGNETOM Skyra; Siemens Healthcare, Erlangen, Germany) using a 18-channel surface coil in combination with a 32channel spine coil and a high-performance gradient system (gradient strength = 45mT/m, slew rate = 200 T/m/s). To reduce colonic motility, a 20 mg dose of scopolamine butlybromide was injected intramuscularly 15 minutes before MR imaging. Conventional localizer and T2-weighted sequences were obtained followed by diffusion-weighted images. DWI of the pelvis was obtained in the axial plane, with imaging parameters detailed in Table 1. The RS-EPI sequence was performed immediately after the SS-EPI sequence in all patients. The routine GRAPPA-accelerated SS-EPI sequence, a standard clinical protocol at our institution, was used. RS-EPI used a monopolar diffusion preparation with seven readout segments and one signal average, which were used by previous authors for the brain (10). The RS-EPI sequence parameters used in this study were established by the authors as an acceptable trade-off between image quality and scan time. Apparent diffusion coefficient (ADC) maps from each acquisition were generated using a monoexponential fit of the acquired b-value data points using b-values of 0 and 1000 s/mm2. Siemens Healthcare provided the prototype RS-EPI DWI sequence (RESOLVE) based on the RS-EPI technique.
Academic Radiology, Vol 21, No 4, April 2014
with 11 and 7 years of experience separately. The evaluators were blinded to subject data and MR acquisition conditions. SS-EPI and RS-EPI diffusion-weighted images and ADC maps were scored in terms of geometric distortion level, image blurring, ghosting artifacts, lesion conspicuity, and overall image quality of visualized anatomy in the pelvis. For artifacts (geometric distortion, image blurring, and ghosting), the severity of the artifact was scored on a 7-point Likert scale with a score of 1 indicating extremely severe and score of 7 indicating the images were free from the particular artifact. Lesion conspicuity and overall image quality were scored on a 7-point scale as follows: 1, unacceptable; 2, poor; 3, below average; 4, standard; 5, above average; 6, good; and 7, outstanding. The evaluators were finally asked to indicate which set of diffusion-weighted images they found preferable (if any) for clinical evaluation. To compare the absolute ADCs obtained using both DWI methods, concordant regions of interest (ROIs) of various primary lesions on the ADC maps were drawn manually by one radiologist. ROI measurements were obtained only in patients with a mass lesion evident on the scan (eg, tumor). An ROI was drawn at a representative portion of the lesion which could be clearly depicted on both DWI techniques, taking care to avoid areas of necrosis or large blood vessels identified on standard T2 or contrast-enhanced images. Circular ROIs with a minimum size of 10 mm2 were obtained. Estimates of signal-to-noise ratio (SNR) were made by taking the ratio of the mean signal intensity in an ROI of at least 25 mm2 in the left or right obturator externus muscle (raw images acquired with b = 0 s/mm2) to the standard deviation of a similar ROI in a homogenous artifact-free region of background air in close proximity to the site of the signal intensity measurement, using an approach described elsewhere (13,14). However as parameters were not normalized for scan time, SNR estimates reported here for both techniques are not directly comparable for SNR efficiency. Statistical Analysis
Statistical analysis was performed using software (Stata Statistical Software: Release 12; StataCorp LP, College Station, TX). The Wilcoxon signed rank test was used to determine significant differences between the radiologist’s ratings of the two techniques. Interreader correlation of qualitative image scores was estimated using the Spearman rank correlation test. Bland–Altman plots were used to assess agreement in ADC obtained using SS-EPI and RS-EPI techniques. A P value of
