Quality Comparison between 3-dimensional T2-weighted SPACE and 2-dimensional T2-weighted turbo spin echo magnetic resonance images for the brachytherapy planning evaluation of prostate and periprostatic anatomy

Abstract

Purpose:

The purpose of this study was to compare an isotropic 3D T2W sequence (SPACE) with an axial 2D T2W TSE sequence with regards to overall image quality and the delineation of normal prostate and periprostatic anatomy for low dose rate (LDR) prostate cancer brachytherapy planning evaluation.

Methods:

Patients (n = 69) with prostate cancer who had pelvic MRI for LDR brachytherapy treatment planning were included. Three radiologists independently assessed the visibility of 9 anatomic structures on each sequence by using a 5-point scale and overall image quality by using a 4-point scale. The significance of the differences in diagnostic performance was tested with a Wilcoxon signed rank test.

Results:

No significant inter-sequence differences were found for most (7/9) anatomical structures and overall image quality. The mean scores for visibility of anatomical structures on the 3D SPACE and 2D TSE sequences, respectively, were as follows zonal anatomy (3.7; 3.9, p=0.05), prostate capsule (3.9; 4.0, p=0.08), neurovascular bundle (2.9; 2.9, p=0.9), rectoprostatic angle (3.8; 3.8, p=0.35), rectum (4.2; 4.3, p=0.26), urethra (3.8; 3.9, p=0.12), urinary bladder (4.6; 4.6, p=0.61), and overall image quality (2.9; 2.9, p=0.33). 3D SPACE was superior for delineation of the genitourinary diaphragm (3.8; 3.6, p=0.003), whereas 2D TSE was superior for delineation of the seminal vesicles (3.5; 4.0, p<0.0001).

Conclusions:

Anatomic delineation of the prostatic and periprostatic anatomy provided by the 3D SPACE sequence is as robust in quality as that provided by a conventional 2D TSE sequence with superior delineation of the genitourinary diaphragm. For MRI-based brachytherapy treatment planning, the 3D SPACE sequence with subcentimeter isotropic resolution can replace the 2D TSE sequence and be incorporated into standard MR imaging protocols.

Introduction

Magnetic resonance imaging (MRI) is increasingly used as the modality of choice for brachytherapy treatment planning because of its excellent soft-tissue contrast and multiplanar imaging capability. [1, 2] MRI-based brachytherapy planning for prostate cancer can more accurately define targets compared to ultrasound or computed tomography, leading to better margin definition, smaller planning target volume (PTV), and thus decreased dose to normal tissues.[3, 4] Traditionally, two-dimensional (2D) T2-weighted (T2W) turbo spin echo (TSE) MR sequence has been the workhorse for prostate and pelvis imaging, including for radiotherapy planning.[5] However, development of the 3D fast spin echo sequence (such as SPACE, for sampling perfection with application-optimized contrasts by using flip angle evolution, on the Siemens platform) allows pelvic and prostate T2W imaging at sub-millimeter resolution. The 3D SPACE sequence has several advantages over its 2D counterpart, particularly for radiotherapy treatment planning.[6] First, the 3D SPACE sequence can be acquired at much higher spatial resolution, particularly along the slice direction and without any slice gaps. Second, the 3D SPACE sequence produces images that can be easily transferred to the treatment planning system and the high resolution images can be reconstructed into any plane without an additional acquisition, making the clinical workflow highly efficient.[7] Besides the 3D SPACE by Siemens, other vendors have similar implementation of the same sequence, such as CUBE by General Electric (GE Healthcare,Chicago, IL) and “Volumetric ISotropic TSE Acquisition” or VISTA by Philips (Philips Healthcare, Andover, MA).

The utility of 3D SPACE sequences has been established in the evaluation of various complex anatomic structures, such as the spine, intracranial structures, pelvis, and joints.[8–12] More recently, researchers showed that 3D SPACE sequences are useful in the imaging of prostate cancers.[13] In this study, an optimized 3D T2W SPACE sequence saved nearly 8 minutes over a 2D T2W TSE sequence while providing similar image quality and comparable accuracy for the diagnosis of prostate tumors and extracapsular extension. Additionally, the 3D T2W SPACE sequence performed better with regard to tumor conspicuity.[13] Another multi-reader study that compared the two sequences in terms of reader preferences and perceived interpretive quality found that the two sequences differ in image quality and artifacts but not in diagnostic ability.[14] To the best of our knowledge, no studies have specifically evaluated the utility of 3D T2W sequence for anatomic delineation of the prostate and periprostatic structures. Therefore, the purpose of our work is to evaluate pelvic MR images of the prostate obtained with the 3D T2W SPACE sequence and a conventional 2D T2W TSE sequence, and to compare how well these two sequences delineated the prostate and periprostatic anatomic structures in patients diagnosed with prostate cancer in the setting of LDR prostate brachytherapy treatment planning. We hypothesized that the 3D T2W SPACE sequence has better image quality and is equivalent or superior to the 2D T2W TSE sequence for the evaluation and delineation of prostate and periprostatic anatomy.

Methods

Patient selection

This retrospective study was compliant with the Health Insurance Portability and Accountability Act and approved by our Institutional Review Board with a waiver of written informed consent. We searched our electronic medical records for all patients that had a pre-brachytherapy prostate MRI without contrast performed on a 1.5-T Siemens Aera scanner between April 1, 2015 and December 31, 2016. Patients with biopsy-proven intermediate-risk prostate cancer who had multiparametric prostate MRI with rigid endorectal coil before prostate brachytherapy were included. All patients had diagnostic quality axial 2D T2W TSE and 3D T2W SPACE sequences as a part of the prostate multiparameter MRI protocol. The parameters for the 2D T2W TSE and 3D T2W SPACE sequences are listed in Table 1.

Table 1.

Scan parameters for 2D T2W TSE and 3D T2W SPACE sequences.

Parameter	2D T2W-TSE	3D T2W-SPACE
Orientation	Axial	Axial
Repetition time (ms)	4130	1600
Echo time (ms)	110	87
Flip angle (degrees)	152	Variable
Field of view (mm)	150	150
Matrix size	256 × 218	256 × 256
Echo train length	23	55
Slice thickness (mm)	3	1.2
No. of sections	34	80
Acquisition time (min:sec)	4:30	3:30

Anatomic evaluation, delineation and scoring

The images were reviewed independently by three subspecialty-trained abdominal radiologists (JS, AV, TB), each with more than 10 years of experience in interpreting prostate MRI. The images were reviewed on a dedicated PACS viewer (GE AW 3.2 MW Wisconsin). Each reader assessed the visibility of 9 anatomic structures (prostate capsule, zonal anatomy, seminal vesicles, neurovascular bundle, rectum, rectoprostatic angle, urinary bladder, urethra, and genitourinary diaphragm) and overall image quality.[15] Each reader scored the visibility and delineation of anatomic structures using a 5-point scale (0 = not visible, 4 = excellent visibility). Each reader also scored overall image quality using a 4-point scale (1 = poor image quality with severe artifact, 2 = fair, 3 = good, 4 = excellent image quality with no artifact). Images from each sequence were reviewed separately to exclude potential bias from direct inter-sequence comparison. To minimize recall bias, separate review sessions were conducted for the 2D T2W TSE and 3D T2W SPACE images at intervals of at least 3 weeks.

Objective image analysis:

The two-dimensional T2-weighted TSE and three-dimensional T2-weighted SPACE images were evaluated objectively by a reader (TKB) with more than 10 years of experience interpreting pelvic MRI. The reader was unblinded to T2-weighted technique. Region of interest (ROI) analysis was performed on picture archiving and communication system (Phillips, iSite) on the 2-D T2-weighted and 3-D T2-weighted SPACE images for each patient. Oval ROIs were placed within the peripheral zone, transition zone, seminal vesicles, periprostatic fat, and levator muscle at the level of prostate apex. ROI size was approximately equal to 50 sq mm in each location. A total of 690 ROIs were placed (the above five ROIs on all 69 patient’s on the 2-D TSE and 3-D SPACE images for each patient). Care was taken to size and place ROIs consistently within the patient. For example, within a given patient, similar size ROIs were placed within the same portion of the peripheral zone, transition zone, seminal vesicle, periprostatic fat and levator muscle. The 2 sequences were viewed side by side for the ROI analysis and consistency of placement was achieved by direct visual comparison of anatomical landmarks. ROIs were not placed on prostatic capsule, rectal prostatic angle, bladder, neurovascular bundle, and external urethral sphincter. When choosing ROI locations, regions with imaging artifact were avoided. For each patient, the ROI location was selected on the 2D TSE images and then reproduced on the 3-D SPACE sequence. Signal to noise ratio (SNR) was calculated as the ratio between the mean signal intensity and standard deviation of the signal intensity of the same ROI of normal peripheral zone.[16] Tissue contrast between tissue A and B was measured according to the formula (signal intensity of tissue A - signal intensity of tissue B/signal intensity of tissue A + signal intensity of tissue B). Higher values close to unity indicate better contrast.[17, 18] Comparisons were made between peripheral zone and periprostatic fat, seminal vesicles and periprostatic fat, levator muscle and periprostatic fat, peripheral zone and transition zone.

Statistical analysis

Mean values and standard deviations were calculated for the ratings of the visibility and delineation of each anatomic structure, SNR, calculated contrast ratio and of overall image quality. The mean scores for the two sequences were compared using the Wilcoxon signed rank test. This test was used because of the non-normal distribution of the data and because it can test for differences between nonparametric variables. Differences with P values less than or equal to 0.05 were considered statistically significant. Because image quality and visibility of the anatomic structures were rated subjectively by the observers, we assessed the degree of interobserver agreement. Spearman rank correlation coefficients were calculated to identify correlations between the categorical variables and observer combinations. The Fleiss kappa was used to assess the interobserver agreement of the ratings for each anatomic structure and for overall image quality.[19] Body mass index (BMI) of 30 kg/m² or more is considered obese. To study the effect of BMI on image quality, image quality scores were dichotomized into two groups, with one being poor quality and two and more being acceptable quality. A Chi-square test was used to assess the statistical difference between the two groups. All statistical calculations were performed using JMP Pro version 13.0 and SAS software version 9.4 (SAS Institute, Cary, NC).

Results

Sixty-nine men with intermediate-risk prostate cancer met our inclusion criteria. Their median age was 65 years, ranging from 48 to 78 years. All observers’ ratings of the delineation of anatomic structures and overall image quality were positively correlated (Spearman ρ ranging from 0.36 to 0.81).

There was no statistically significant difference in overall image quality scores between the 2 sequences (Table 2; Figure 1): the mean scores (on a 4-point scale) were 2.99 (SD = 0.49) for 2D T2W TSE and 2.92 (SD = 0.34) for 3D T2W SPACE (P = 0.33). We found that 2D T2W TSE images (mean score, 4.0) were rated better than 3D T2W SPACE images (mean score, 3.5) for the delineation of seminal vesicles (P < 0.0001), but 3D T2W SPACE images (mean score, 3.8) were rated better than the 2D T2W TSE images (mean score, 3.6) images for the delineation of the genitourinary diaphragm (P = 0.003). However, no significant differences were observed in the scores for the other 7 anatomic areas (Table 2; Figure 1). A few representative images of the prostate are shown in Figure 2. No significant correlation was observed between obesity and image quality. For 2D T2W TSE, the mean BMI of the group with poor and acceptable image quality was 35.6 kg/m2 (SD = 7.4) and 33.9 kg/m2 (SD=6.8) with p-value of 0.12. For 3D T2W SPACE, the mean BMI of the group with poor and acceptable image quality was 36.2 kg/m2 (SD = 8.4) and 35.4 kg/m2 (SD=7.2) with p-value of 0.42.

Table 2.

Mean scores for delineation of prostate and periprostatic anatomical structures and image quality obtained with 2D T2W TSE and 3D T2W SPACE sequences.

	MRI sequence*
Structure	2D T2W TSE	3D T2W SPACE	Δ	p-value
Prostate capsule	4.05 (0.87)	3.91 (0.69)	0.14	0.08
Zonal anatomy	3.87 (0.82)	3.74 (0.72)	0.13	0.052
Seminal vesicles	4.01 (0.84)	3.49 (0.88)	0.52	<0.0001
Neurovascular bundle	2.88 (1.16)	2.89 (1.03)	−0.01	0.88
Rectum	4.29 (0.81)	4.23 (0.62)	0.06	0.26
Rectoprostatic angle	3.83 (0.82)	3.77 (0.76)	0.06	0.35
Urinary bladder	4.59 (0.61)	4.57 (0.56)	0.03	0.61
Urethra	3.86 (0.66)	3.77 (0.56)	0.09	0.12
Genitourinary diaphragm	3.59 (0.72)	3.78 (0.63)	−0.19	0.003
Overall image quality†	2.88 (0.49)	2.92 (0.34)	−0.04	0.33

^*Data are mean scores (± standard deviation). Visibility and delineation of anatomical structures were evaluated on a 5-point scale (0 = not visible, 4 = excellent visibility).

^†Overall image quality was evaluated on a 4-point scale (1 = poor image quality with severe artifact, 2 = fair, 3 = good, 4 = excellent image quality with no artifact).

^Δ, difference of mean scores;

Figure 1.

Mean scores for visibility and delineation of the 9 evaluated anatomical sites and overall image quality.

Figure 2. Representative images of the prostate, from apex to base (right column – 3D T2W SPACE images; left column - 2D T2W TSE images).

Pc- prostate capsule, Pz-pheripheral zone, Tz- transitional zone, Cg-central gland, Sv-seminal vesicles, Pb-pubic bone, Nvb-neurovascular bundle, Rct-rectum, RpA-rectoprostatic angle, Ub-urinary bladder, Ure-urethra, Eus- External urethral sphincter, and Gu-genitourinary diaphragm.

Kappa values for interobserver agreement ranged from 0.26 to 0.60 for 2D T2W TSE and from 0.22 to 0.51 for 3D T2W SPACE images. These values are consistent with fair to moderate levels of interobserver agreement for both sequences (Table 3). The 2D T2W TSE sequence had a higher kappa (better agreement) than did 3D T2W SPACE for assessments of overall image quality and for delineation of the rectum, urinary bladder, urethra, and genitourinary diaphragm. The two sequences had similar kappa values for delineation of zonal anatomy, seminal vesicles, and the neurovascular bundle. A higher kappa value was found for 3D T2W SPACE images for delineation of the prostatic capsule and rectoprostatic angle.

Table 3.

Interobserver agreement in ratings of visibility and delineation of the prostate and periprostatic anatomical structures.

	MRI sequence*
Structure	2D T2W TSE			3D T2W SPACE
	kappa	95%LCL	95%UCL	kappa	95%LCL	95%UCL
Prostate capsule	0.36	0.27	0.45	0.44	0.35	0.54
Zonal anatomy	0.38	0.29	0.47	0.37	0.28	0.47
Seminal vesicles	0.45	0.35	0.54	0.44	0.35	0.53
Neurovascular bundle	0.26	0.08	0.23	0.24	0.05	0.22
Rectum	0.45	0.35	0.56	0.23	0.12	0.34
Rectoprostatic angle	0.33	0.24	0.42	0.47	0.38	0.57
Urinary bladder	0.36	0.24	0.48	0.27	0.15	0.40
Urethra	0.37	0.28	0.47	0.24	0.13	0.35
Genitourinary diaphragm	0.41	0.31	0.50	0.31	0.13	0.32
Overall image quality	0.60	0.50	0.71	0.51	0.39	0.63

^*Visibility and delineation of anatomical structures was evaluated using a 5-point scale (0 = not visible, 4 = excellent visibility). Overall image quality was evaluated using a 4-point scale (1 = poor image quality with severe artifact, 2 = fair, 3 = good, 4 = excellent image quality with no artifact).

Abbreviations: LCL, lower confidence limit; UCL, upper confidence limit; SV, seminal vesicles; NVB, neurovascular bundle; RP, rectoprostatic; GU, genitourinary.

SNR and relative tissue contrast ratios are provided in Tables 4 and 5, respectively. Compared with 3D T2W SPACE images, 2D T2W TSE images had slightly higher mean SNR of the normal tissues (Table 4). However, only peripheral zone showed statistically significant difference (p=0.048). There was no significant difference between the relative tissue contrast ratios with the 2D T2W TSE sequence and those with the 3D T2W SPACE (Table 5).

Table 4.

Comparison of 2D T2W TSE and 3D T2W SPACE images for signal – to- noise ration of normal tissues.

	MRI sequence*
Finding	2D T2W TSE	3D T2W SPACE	p-value
SNR of normal PZ	13.2 (4.2)	9.8 (3.6)	0.048
SNR of normal TZ	7.5 (3.8)	7.1 (3.3)	0.057
SNR of normal SV	11.3 (4.5)	10.6 (5.2)	0.083
SNR of normal levetor ani muscle	3.7 (1.6)	3.2 (1.8)	0.426
SNR of periprostatic fat	16.9 (6.4)	13.5 (5.9)	0.054

^*Data are mean SNR value (± standard deviation). Abbreviations: SNR, signal to noise ratio; PZ, peripheral zone; TZ, transition zone; SV, seminal vesicle

Table 5.

Relative contrast ratios between tissue types.

	MRI sequence*
Tissue	2D T2W TSE	3D T2W SPACE	p-value
Periprostatic fat to PZ	0.04 (0.13)	0.06 (0.15)	0.625
Periprostatic fat to SV	0.03 (0.08)	0.05 (0.12)	0.941
Periprostatic fat to muscle**	0.77 (0.05)	0.72 (0.06)	0.259
PZ to TZ	0.28 (0.10)	0.27 (0.14)	0.368

^*Data are mean calculated contrast ratio (± standard deviation).

^**levator ani muscle.

Abbreviations: SNR, signal to noise ratio; PZ, peripheral zone; TZ, transition zone; SV, seminal vesicle

Discussion

We objectively and subjectively compared the tissue delineation, SNR, contrast between tissues and overall image quality achieved using the two different T2 acquisition techniques, specifically 3D T2W SPACE and 2D T2W sequences. Our results indicate that 3D SPACE images better delineated the genitourinary diaphragm, whereas 2D TSE images better delineated the seminal vesicles. However, 7 of the 9 anatomical structures were equally well delineated on both sequences. Objective analysis showed that SNR of the 2D T2W images is better than 3D T2W SPACE images which can be explained by the differences in the scan parameters such as slice thickness, TE, TR, ETL and matrix size (Table 1). Although SNR of 2D T2W sequence was slight higher than 3D T2W SPACE sequence, no significant difference was noted between the calculated relative tissue contrast ratios. On the basis of our results, we conclude that the 3D T2W SPACE sequence could replace 2D T2W TSE imaging for the delineation of the prostate and periprostatic anatomy in LDR prostate brachytherapy treatment planning.

T2W MR images of the prostate play a key role for the evaluation of prostate cancer. Standard MRI protocols typically have 3 sets of T2W images obtained at 3 orthogonal planes perpendicular to the long axis of the prostate. These 3 sets of T2W images are essential for diagnostic purposes because they can be used to detect prostate cancer and to diagnose extraprostatic extension. However, obtaining T2W images in 3 separate planes requires a significant amount of the overall magnet time. These images are also typically used for LDR prostate brachytherapy treatment planning and anatomical contouring because they can delineate the periprostatic pelvic anatomy. It would be useful, therefore, to use 3D SPACE sequence that enables postprocessing reconstruction of images into any desired plane in the standard protocol for MRI of the prostate. Thus, the postprocessing generation of multiplanar reconstructions from the 3D SPACE dataset can decrease examination time and improve the efficiency of the MRI workflow. In a study comparing 3D and 2D T2W sequences for MRI of prostate, the acquisition time for the 3D SPACE sequence never exceeded the time needed to perform the 3 orthogonal plane acquisitions of 2D T2W sequences.[7] 3D SPACE sequence can also serve as a useful problem-solving tool for radiation oncologists and dosimetrists because it offers the option of obtaining images in any plane, even after the patient’s examination is complete. A major advantage of the 3D T2W SPACE sequence is that it can acquire isotropic imaging data at section thicknesses of a millimeter or less, a resolution that is not currently achievable by a 2D T2W TSE sequence. Therefore, the 3D T2W SPACE sequence could replace conventional 2D T2W TSE imaging in at least 2 applications in the management of prostate cancer: (1) imaging of the prostate for the diagnosis and staging of cancer,[7] and (2) ultrasonography-MRI fusion technologies that guide biopsies and focal therapy. In the latter case, imaging must not only accurately identify tumors but also precisely delineate glands to prevent misregistration.

The 3D T2W SPACE sequence has been studied for its usefulness in the evaluation of pelvic pathology and prostate cancer.[20, 13, 10, 14] However, these studies either have been focusing on pelvic pathology or have not focused on anatomical delineation of prostate and periprostatic anatomy. In one previous study, the investigators compared the diagnostic accuracy of the 2D T2W TSE and 3D T2W SPACE sequences for the evaluation of prostate cancer and extraprostatic extension, with postoperative pathologic analysis as the gold standard. In that study, the diagnostic performance of the 3D T2W SPACE sequence was equivalent to that of a conventional multiplanar 2D TSE sequence, and the sequences showed similar image quality.[13] In a more recent study, researchers studied the perceived image quality of 3-T axial T2W high-resolution 2D TSE and 3D SPACE endorectal MR images of the prostate among 6 readers. They observed no difference in the sequences’ ability to delineate the glandular anatomy and detect prostate cancer.[14] In the current study, we not only studied the prostatic anatomy, but further evaluated the utility of the sequences in the delineation of periprostatic anatomic structures. This is important because contouring errors are common at the base and apex of the prostate gland. The contours should be both precise and accurate to avoid errors and over estimation of the target volume.[4] Our data confirmed the findings of these previous studies and demonstrated the utility and adequacy of 3D T2W SPACE images in the anatomic delineation needed for LDR prostate brachytherapy treatment planning.

Panebianco et al.[21] compared 2D TSE and 3D T2W SPACE sequences acquired on 1.5 T with endorectal coil for the assessment of changes in the neurovascular bundles in 53 men who underwent nerve-sparing prostatectomy and determined whether these changes correlated with posttreatment erectile function. They found that the findings on the 3D T2W images better correlated with the International Index Erectile Function Five-Item (IIEF-5) score, suggesting that 3D T2W SPACE images allow clearer depiction of the neurovascular bundle, likely owing to the high-resolution isotropic acquisition. In our study, we found no difference between the 2 acquisition types in the anatomic delineation of the neurovascular bundles with intact prostate on images acquired using a 1.5-T magnet with endorectal coil.

There were several limitations to our study. First, our study was retrospective and has an inherent selection bias in design, and may have other unknown confounders. However, our study population was homogeneous with intermediate-risk prostate cancer without evidence of extraprostatic extension or seminal vesicle involvement and who were under consideration for LDR prostate brachytherapy. Another limitation of our study is that we analyzed images using subjective, qualitative scores and did not employ objective criteria for anatomic delineation due to lack of any such validated criteria to assess visibility of periprostatic anatomic structures. However, our results are still clinically relevant as we had three experienced readers to assess images which simulates clinical practice and objective image assessment was performed through calculation of SNR and relative tissue contrast. In addition, we did not adjust scores to account for reader experience since all our reader were fellowship trained and experienced abdominal radiologists and therefore could not compare the scores of less experienced observers with those of highly experienced observers. Further studies could use a multireader design, which would allow a more fine-grained evaluation of interobserver agreement. Finally, we did not compare the treatment plans based on each sequence, as our goal was to assess image quality and anatomical delineation before using it for contouring.

Conclusions

The overall image quality and the anatomic delineation of prostate and periprostatic structures on the images obtained with an axial 3D T2W SPACE sequence were as robust in quality and comparable to those of conventional 2D T2W TSE images for the evaluation of the prostate and periprostatic anatomy. The 3D T2W SPACE sequence, with its ability to acquire isotropic imaging data at millimeter or submillimeter section thickness, is a promising tool for brachytherapy planning because it allows reconstruction of images in any orientation and improved contouring of prostate and periprostatic structures. Further research on the utility of this sequence in LDR prostate brachytherapy treatment planning is warranted.

Abbreviations:

2D TSE

two-dimensional turbo spin echo

3D SPACE

three-dimensional sampling perfection with application-optimized contrasts by using flip angle evolution

LDR

low dose rate