T2 Turbo Spin Echo With Compressed Sensing and Propeller Acquisition (Sampling k-Space by Utilizing Rotating Blades) for Fast and Motion Robust Prostate MRI: Comparison With Conventional Acquisition

Prostate cancer has the second-highest incidence and fifth-highest mortality of all cancer types in men.1 Early diagnosis of clinically significant cancer is crucial for patient survival. In recent years, multiparametric magnetic resonance imaging (mpMRI) of the prostate has emerged as an important noninvasive diagnostic tool, as it can reliably detect clinically significant cancer and MRI data can be used for targeted prostate biopsy.2 Therefore, the demand for mpMRI of the prostate will further increase in the future, amplified by an increasingly aging population. To meet this increasing demand for imaging, it is important to further accelerate mpMRI protocols while ensuring high diagnostic quality.1,2

The standard scan protocol of prostate mpMRI contains T2-weighted sequences for the assessment of anatomical properties and diffusion-weighted sequences and/or dynamic contrast-enhanced sequences for the assessment of functional properties.3 Different k-space sampling methods have been implemented to reduce motion artifacts, especially for T2-weighted images.4 One approach is to repeatedly sample the k-space by using rotating blades around the center, while the patient can breathe freely (MULTIVANE; Philips Healthcare, Best, the Netherlands; BLADE; Siemens Healthineers, Erlangen, Germany; PROPELLER; GE Healthcare, Chicago, IL). This so-called propeller acquisition technique allows for an oversampling of the center of the k-space. The in-plane information of minor patient movements during a specific repetition can be corrected due to the nonanomalous data from other repetitions. As the central k-space information is responsible for image contrast, the overall contrast resolution and especially the signal-to-noise ratio (SNR) are improved with this method.5–7

These techniques were further developed by combining them with parallel imaging, such as sensitivity encoding (SENSE; Philips Healthcare, Best, the Netherlands) and more advanced motion correction algorithms and iterative reconstructions (MultiVane XD; Philips Healthcare, Best, the Netherlands).8,9 Independently of medical imaging methods, the mathematical principle of compressed sensing (CS) was developed.10–13 During the last years, CS was systematically improved and introduced to MRI, leading to its widespread use in different fields such as cardiovascular imaging, neuroimaging, or body imaging.14,15 Compressed sensing markedly speeds up the acquisition times of MRI sequences, but could only be combined with Cartesian sampling methods, contrary to the non-Cartesian sampling methods used in propeller acquisitions.

In this study, we evaluated an enhanced CS algorithm with improved motion correction and contrast weighting (SmartSpeed MotionFree; Philips Healthcare, Best, the Netherlands) for prostate MRI that was combined with a propeller acquisition technique for the first time. This study aimed to compare this new technique to conventional T2-weighted sequences with sensitivity encoding propeller acquisition in terms of motion reduction, image sharpness, lesion conspicuity, capsule delineation, and overall image quality. Further, we aimed to evaluate whether the new sequence can reduce acquisition time. Both the increased image quality and reduced acquisition time are key properties for ensuring a fast and robust workup of patients with prostate cancer.

MATERIALS AND METHODS Study Population

This prospective study was approved by the institutional ethics committee of the University Hospital Bonn. All individuals gave written informed consent before study participation. Male participants were consecutively recruited from September 2021 to March 2022. Participants with an indication for mpMRI of the prostate based on clinical suspicion of prostate cancer were included. This suspicion was either due to an abnormal digital rectal examination, transrectal ultrasonography, or elevated prostate-specific antigen (PSA) levels defined as >4 ng/mL. Exclusion criteria were general contraindications for MRI (ie, cardiac pacemakers, neurostimulators, or long metallic implants) and severe claustrophobia.

Imaging Protocols

All scans were performed on the same clinical 3 T MRI system (Ingenia 3.0 T Elition X; Philips Healthcare, Best, the Netherlands). The scan protocol included axial T1-weighted imaging before and after administration of contrast media, axial diffusion-weighted imaging, axial dynamic contrast-enhanced imaging, and an axial and sagittal conventional parallel imaging (SENSE; Philips Healthcare, Best, the Netherlands) accelerated T2-weighted sequence with propeller acquisition (MultiVane XD, Philips Healthcare), henceforth named T2conv sequence. In addition, a CS accelerated T2-weighted sequence with propeller acquisition (SmartSpeed MotionFree; Philips Healthcare, Best, the Netherlands) was included in the scan protocol, henceforth named T2CS sequence. The used CS technique was based on a combination of the established methods of CS and parallel imaging using SENSE (Compressed SENSE; Philips Healthcare, Best, the Netherlands). In comparison to the conventional CS, the new algorithm of the T2CS sequence has improved motion correction and contrast weighting and can use the non-Cartesian sampling method of propeller acquisition. Both T2-weighted sequences were performed in axial and sagittal orientation. Axial images were planned perpendicular to the proximal urethra. The scan parameters of T2conv sequence and T2CS sequence are given in Table 1.

TABLE 1 - Imaging Parameters of the T2 Sequences Sequence T2conv Axial T2CS Axial T2conv Sagittal T2CS Sagittal Echo time, ms 133 133 140 140 Repetition time, ms 4290 4293 4526 4526 Flip angle, degrees 90 90 90 90 Matrix 512 × 512 512 × 512 480 × 480 480 × 480 Field of view, mm2 180 × 180 180 × 180 250 × 250 250 × 250 Slice thickness, mm 3 3 3 3 Echo train length 29 29 29 29 Bandwidth, Hz/pixel 204 194 174 174 Scan time of regular sequence, s 231 171 217 162

T2conv, conventional T2-weighted propeller sequence; T2CS, T2-weighted propeller sequence with compressed sensing.


Image Analysis

All T2-weighted images were qualitatively and quantitatively analyzed in a blinded manner. For qualitative assessment, 3 radiologists with 1, 5, and 9 years of experience in mpMRI of the prostate independently rated the image quality of the T2conv and the T2CS sequences regarding 4 different categories: artifacts, image sharpness, lesion conspicuity, and overall image quality. Five-point Likert items were used for each category. The rating of artifacts was defined as follows: (1) nondiagnostic; (2) severe artifacts with insufficient diagnostic confidence; (3) moderate artifacts, not interfering with diagnosis; (4) good, minimal artifacts; and (5) excellent, no artifacts. Image sharpness, lesion conspicuity, capsule delineation, and overall image quality were defined as follows: (1) nondiagnostic, blurry; (2) poor, structures can be defined but insufficient diagnostic confidence; (3) moderate, sufficient for diagnosis but low diagnostic confidence; (4) good, diagnostic with high confidence; and (5) excellent, sharp images with exceptional diagnostic confidence. For quantitative comparisons, the apparent signal-to-noise ratio (aSNR: signal intensity [SI]peripheral zone/standard deviation [SD]muscle) and apparent contrast-to-noise ratio (aCNR: SIperipheral zone − SImuscle/SDmuscle) were calculated for both the axial and sagittal plane. SImuscle was measured in the internal obturator muscle, whereas SIperipheral zone was measured in the lesion-free, normal apparent peripheral zone of the prostate. PI-RADS scores were assessed separately for both sequences.16

Statistical Analysis

Statistical analysis was performed using SPSS (IBM, Version 27; Armonk, NY) and Prism (GraphPad Software, Version 9.3.1; San Diego, CA). Continuous variables (quantitative assessment) are given as mean ± standard deviation, whereas discrete variables (qualitative assessment) are expressed as median with interquartile range (IQR) or as percentages. For group comparisons, paired-samples t test or Wilcoxon signed-rank test was used, as appropriate. Intrarater and interrater reliability was assessed by calculating the intraclass correlation coefficient (ICC) with their corresponding 95% confidence intervals (CIs). A P value of <0.05 was considered indicative of a significant difference.

RESULTS Participants Characteristics

A total of 29 male participants were included in this study, with a mean age of 66 ± 8 years, ranging from 51 to 81 years. Thirty-eight percent (11/29) of participants had previously undergone biopsy. Mean PSA was 8.1 ± 5.9 ng/mL, ranging from 0.67 to 31.4 ng/mL. Suspicious findings were noted in 22% (5/23) of participants on digital rectal examination and in 17% (3/18) on suspect transrectal ultrasonography.

Qualitative Comparisons

Representative images are shown in Figure 1. In the following, only the results for rater 1 are given. The ratings from rater 2 and rater 3 also favored the T2CS sequences over the T2conv sequences (see Fig. 2 for the rating distribution of all raters).

F1FIGURE 1:

Representative images of a 69-year-old patient with PSA elevation of 5.9 ng/mL (A–D and M–P), a 61-year-old patient with PSA elevation of 4.24 ng/mL (E–H), and an 81-year-old patient with PSA elevation of 31.37 ng/mL (I–L). Zoomed images of prostate nodules (C and D) and the prostate capsule (G and H) show increased image contrast and less artifacts (O and P) in the T2-weighted propeller sequence with compressed sensing (T2CS) compared with the conventional T2-weighted propeller sequence (T2conv). A probable infiltration of the neurovascular bundle (I–L) in the peripheral zone is better visible in the T2CS sequence.

F2FIGURE 2:

Bar plots show distributions of artifact, image sharpness, lesion conspicuity, capsule delineation, and overall image quality ratings of the T2-weighted propeller sequence with compressed sensing (T2CS) compared with the conventional T2-weighted propeller sequence (T2conv). No images were rated as nondiagnostic.

Artifacts were noticed more frequently on the axial T2conv sequence compared with the axial T2CS sequence (4 [4–4.5] vs 4 [3–4]; P < 0.001), results were similar for sagittal images (4 [4–4] vs 3 [3–4]; P < 0.001) (Fig. 1).

Image sharpness was rated higher for the axial T2CS sequence compared with the T2conv sequence (4 [4–4.5] vs 3 [3–3.5]; P < 0.001) (Fig. 1). Similar results were found for sagittal images (4 [4–5] vs 4 [3–4]; P < 0.001).

Lesion conspicuity was rated higher for the T2CS sequence on sagittal images (4 [4–4] vs 4 [3–4]; P = 0.002). However, there was no difference between the axial T2CS sequence and the T2conv sequence (4 [4–4] vs 4 [3–4]; P = 0.166).

Capsule delineation was significantly sharper for the axial T2CS sequence (4 [3–4] vs 3 [3–3.5]; P < 0.001) (Fig. 1), whereas there was no significant difference in the sagittal plane (3 [3–4] vs 3 [3–3]; P = 0.07).

Overall image quality was rated higher in the T2CS sequence compared with the T2conv sequence (axial plane: 4 [4–4] vs 4 [3–4]; P < 0.001; sagittal plane: 4 [4–4] vs 4 [3–4]; P = 0.002).

The intrarater reliability was good with an overall ICC of 0.92 (95% CI, 0.91–0.93), an ICC for the T2conv sequence of 0.92 (95% CI, 0.89–0.93), and an ICC for the T2CS sequence of 0.88 (95% CI, 0.85–0.91).

The interrater reliability was good with an overall ICC of 0.80 (95% CI: 0.77–0.83); the respective ICC for the T2conv sequence was 0.77 (95% CI, 0.72–0.81), whereas it was 0.69 (95% CI, 0.62–0.75) for the T2CS sequence.

Quantitative Comparisons

Compared with the T2conv sequence, the T2CS sequence had a significantly higher aCNR in both the axial (44.0 ± 9.6 vs 18.6 ± 3.7; P < 0.001) and the sagittal plane (23.7 ± 6.1 vs 20.4 ± 5.7; P < 0.001). Also, the aSNR was significantly higher in the T2CS sequence compared with the T2CS sequence in the axial (52.2 ± 9.7 vs 22.8 ± 3.6, P < 0.001) and the sagittal plane (27.7 ± 6.3 vs 24.2 ± 5.9; P < 0.001). Acquisition time was lower in the T2CS sequence as compared with the T2conv sequence (axial plane: 171.4 ± 2.2 seconds vs 231.6 ± 3.2 seconds; P < 0.001, sagittal plane: 164.7 ± 6.4 seconds vs 218.1 ± 8.1 seconds) (see Fig. 3).

F3FIGURE 3:

Column graphs show distribution of continuous data between the T2-weighted propeller sequence with compressed sensing (T2CS) and the conventional T2-weighted propeller sequence (T2conv). The apparent signal-to-noise ratio (aSNR, A) and the apparent contrast-to-noise ratio (aCNR, B) are significantly increased, whereas the acquisition time is significantly reduced (C). Data are presented as mean with standard deviation error bars. Estimation plots show the quantitative analysis for every participant for the aSNR (D and E) and aCNR (F–G).

Comparison of PI-RADS Scores

PI-RADS scores were the same for both sequences for each participant. PI-RADS scores were as follows: 21% (6/29) participants had a PI-RADS-1 score, 24% (7/29) participants had a PI-RADS-2 score, 24% (7/29) participants had a PI-RADS-3 score, 17% (5/29) participants had a PI-RADS-4 score, and 14% (4/29) participants had a PI-RADS-5 score. Sixty-two percent (18/29) of participants subsequently underwent MR fusion biopsy, and prostate cancer was confirmed in 53% of patients who underwent biopsy (10/18) (see Table 2).

TABLE 2 - Results of Biopsies Depending Upon the PI-RADS Score PI-RADS Score Performed MRI Fusion Biopsy Confirmed Malignancy ISUP Grade Prostate Zone 1 50% (3/6) 33% (1/3) 1 × ISUP 1 1 × peripheral zone 2 29% (2/7) 0% (0/2) — — 3 71% (5/7) 40% (2/5) 2 × ISUP 1 2 × peripheral zone 4 80% (4/5) 75% (3/4) 1 × ISUP 1
1 × ISUP 2
1 × ISUP 3 2 × peripheral zone
1 × transitional zone/anterior fibromuscular stroma 5 100% (4/4) 100% (4/4) 2 × ISUP 2
1 × ISUP 3
1 × ISUP 5 2 × peripheral zone
2 × transitional zone

PI-RADS, Prostate Imaging Reporting and Data System; ISUP, International Society of Urological Pathology.


DISCUSSION

This study compared a new investigational T2-weighted CS accelerated propeller sequence to a conventional T2-weighted SENSE accelerated propeller sequence regarding qualitative (artifacts, image sharpness, lesion conspicuity, capsule delineation, and overall image quality) and quantitative imaging differences (aSNR and aCNR). The new CS accelerated propeller sequence had a superior image quality compared with the conventional propeller sequence and led to a significant reduction in acquisition time.

The continuing demographic development of an aging society necessitates efficient medical diagnostics to secure satisfactory medical care for all individuals. Especially the field of radiology has an important key role, as it provides crucial information leading to cancer diagnoses, which after cardiovascular disease is the second highest cause of death in the world. With the overall second-highest incidence and the fifth-highest mortality of all cancer types in men, prostate cancer causes 375,000 deaths per year worldwide.1,17 Multiparametric magnetic resonance imaging of the prostate has been an important diagnostic tool for the detection of prostate cancer in recent years, as it can visualize the prostate and its different zones noninvasively, estimate the probability of clinically significant cancer, and provide the basis for fusion prostate biopsies. The exact characterization and delineation not only of the lesions themselves but also of possible infiltration of the prostatic capsule or neurovascular bundle are particularly important for grading in the PI-RADS system. This information may be crucial in determining whether a biopsy is performed and in which zone. Although the sensitivity of prostatic mpMRI for cancer detection with an estimated 89% is relatively high, the relatively long examination protocols limit the more widespread use of this modality.2,3

Prostate MRI was recently significantly improved by the introduction of multidimensional diffusion MRI and deep learning–based diagnosis systems.18,19 A different approach even proposed a 5-minute screening protocol for prostate MRI using a simultaneous multislice technique for diffusion weighted MRI.20 The newly introduced acquisition method, combining compressed sensitivity encoding with propeller acquisition techniques, could solve this problem even further as it reduces the acquisition time of T2-weighted sequences by 26% in the axial plane and 24% in the sagittal plane.7–10,15 As shown in our study, this can be achieved while at the same time enhancing image quality. With exception of lesion conspicuity in the axial plane and capsule delineation in the sagittal plane, this new technique further reduces motion artifacts and improves the image sharpness and overall image quality in both axial and sagittal orientations, while further improving lesion conspicuity in the sagittal plane and capsule delineation in the axial plane. Interestingly, the lesion conspicuity was not significantly enhanced in the axial plane, which could be partly explained by heterogeneity of lesions in the transitional and peripheral zone. Furthermore, image sharpness rating was done on the whole image, whereas for the lesion conspicuity assessment only the prostate itself was evaluated. It must be noted that rater 3 was especially critical in his ratings for the sagittal plane and slightly differed from rater 1 and 2, as he considered only the capsule delineation for T2CS as significantly better, although there was a trend favoring the T2CS sequence. Corresponding to the qualitative analysis, quantitative analysis confirmed these results by showing a significantly increased aSNR and aCNR, confirming the superiority of the new CS technique. Especially for radiotherapy, a sharp delineation of the prostate capsule and a high contrast is crucial for a targeted radiation planning and the protection of surrounding organs such as the bowel.21

The calculated PI-RADS scores of the new T2CS sequence and the T2conv sequence were identical. In this context, it should be noted that the T2 sequence is only 1 criterion for the determination of the PI-RADS score, as the diffusion-weighted, and the dynamic contrast-enhanced T1 sequences also need to be considered, especially in the peripheral zone of the prostate.15 However, it is important to show that the diagnostic performance of new and faster imaging techniques is consistent and works in all patients. With a reduction of the acquisition time in the axial plane by 26% and in the sagittal plane by 24%, we could show that the faster acquisition time was not associated with a degradation of image quality. The total reduction in acquisition time was 114 seconds, which could be even higher when the technique is applied to further T2 imaging planes or when CS is applied to diffusion-weighted and T1-weighted imaging.

Our study has limitations. The sequence was only used on a high-end clinical MRI system with state-of-the-art reconstruction hardware. Therefore, reconstruction times of CS images were low in our study (approx. 30 seconds). However, image quality and reconstruction times could differ depending on the MRI system used. It should also be noted that this study only represents an initial evaluation of a rather small patient collective. Continuous use of the sequences in the routine clinical workflow needs to show their full effectiveness in a larger patient collective.

In conclusion, we present for the first time the combination of a T2-weighted propeller sequence with CS for mpMRI of the prostate. The evaluated sequence offers a superior image quality compared with the conventional T2-weighted propeller sequence. Most importantly, the new technique reduced the acquisition times of these sequences by 24% to 26%. Based on these characteristics, the new T2-weighted propeller sequence with CS has the potential to replace the conventional T2-weighted propeller sequence in mpMRI, improving and accelerating the imaging protocol for evaluation of prostate cancer patients. Furthermore, T2-weighted propeller sequences with CS are not necessarily limited to mpMRI of the prostate. This new technique may be implemented in the scan protocol for various body regions, for example, for the brain or the abdomen/liver.6,7,22

REFERENCES 1. Sung H, Ferlay J, Siegel RL, et al. Global Cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71:209–249. 2. Eklund M, Jäderling F, Discacciati A, et al. MRI-targeted or standard biopsy in prostate cancer screening. N Engl J Med. 2021;385:908–920. 3. Ghai S, Haider MA. Multiparametric-MRI in diagnosis of prostate cancer. Indian J Urol. 2015;31:194–201. 4. Hennig J. K-space sampling strategies. Eur Radiol. 1999;9:1020–1031. 5. Forbes KP, Pipe JG, Bird CR, et al. PROPELLER MRI: clinical testing of a novel technique for quantification and compensation of head motion. J Magn Reson Imaging. 2001;14:215–222. 6. Pipe JG. Motion correction with PROPELLER MRI: application to head motion and free-breathing cardiac imaging. Magn Reson Med. 1999;42:963–969. 7. Hirokawa Y, Isoda H, Maetani YS, et al. MRI artifact reduction and quality improvement in the upper abdomen with PROPELLER and prospective acquisition correction (PACE) technique. AJR Am J Roentgenol. 2008;191:1154–1158. 8. Pipe JG, Gibbs WN, Li Z, et al. Revised motion estimation algorithm for PROPELLER MRI. Magn Reson Med. 2014;72:430–437. 9. Chang Y, Pipe JG, Karis JP, et al. The effects of SENSE on PROPELLER imaging. Magn Reson Med. 2015;74:1598–1608. 10. Candes EJ, Romberg J, Tao T. Robust uncertainty principles: exact signal reconstruction from highly incomplete frequency information. IEEE Trans Inf Theory. 2006;52:489–509. 11. Candès EJ, Romberg JK, Tao T. Stable signal recovery from incomplete and inaccurate measurements. Comm Pure Appl Math. 2006;59:1207–1223. 12. Candes EJ, Tao T. Near-optimal signal recovery from random projections: universal encoding strategies?IEEE Trans Inf Theory. 2006;52:5406–5425. 13. Donoho DL. Compressed sensing. IEEE Trans Inf Theory. 2006;52:1289–1306. 14. Lustig M, Donoho D, Pauly JM. Sparse MRI: The application of compressed sensing for rapid MR imaging. Magn Reson Med. 2007;58:1182–1195. 15. Jaspan ON, Fleysher R, Lipton ML. Compressed sensing MRI: a review of the clinical literature. Br J Radiol. 2015;88:20150487. 16. ACR, ESUR, and AdMeTech Foundation. Prostate Imaging Reporting & Data System (PI-RADS). 2019. Version 2.1. 17. Siegel RL, Miller KD, Fuchs HE, et al. Cancer statistics, 2022. CA Cancer J Clin. 2022;72:7–33. 18. Langbein BJ, Szczepankiewicz F, Westin CF, et al. A pilot study of multidimensional diffusion MRI for assessment of tissue heterogeneity in prostate cancer. Invest Radiol. 2021;56:845–853. 19. Winkel DJ, Tong A, Lou B, et al. A novel deep learning based computer-aided diagnosis system improves the accuracy and efficiency of radiologists in reading biparametric magnetic resonance images of the prostate: results of a multireader, multicase study. Invest Radiol. 2021;56:605–613. 20. Weiss J, Martirosian P, Notohamiprodjo M, et al. Implementation of a 5-minute magnetic resonance imaging screening protocol for prostate cancer in men with elevated prostate-specific antigen before biopsy. Invest Radiol. 2018;53:186–190. 21. Wong OL, Poon DMC, Kam MKM, et al. 3D-T2W-TSE radiotherapy treatment planning MRI using compressed sensing acceleration for prostate cancer: image quality and delineation value. Asia Pac J Clin Oncol. 2022. doi:10.1111/ajco.13752. 22. Kang KA, Kim YK, Kim E, et al. T2-weighted liver MRI using the MultiVane technique at 3T: comparison with conventional T2-weighted MRI. Korean J Radiol. 2015;16:1038–1046.

留言 (0)

沒有登入
gif