Image artifacts on prostate diffusion-weighted magnetic resonance imaging: trade-offs at 1.5 Tesla and 3.0 Tesla

Abstract

Rationale and objectives: To identify the presence and extent of artifacts in prostate diffusion-weighted magnetic resonance imaging (DW-MRI) and discuss tradeoffs between imaging at 1.5 T (1.5 T) and 3.0 T (3.0 T). In addition, we aim to provide quantitative estimates of signal-to-noise ratios (SNRs) at both field strengths.

Materials and methods: The institutional review board waived informed consent for this Health Insurance Portability and Accountability Act-compliant, retrospective study of 53 consecutive men who underwent 3.0 T endorectal DW-MRI and 53 consecutive men who underwent 1.5 T endorectal DW-MRI between October and December 2010. One radiologist and one physicist, blinded to patient characteristics, image acquisition parameters, and field strength, scored DW-MRI artifacts. On b = 0 images, SNR was measured as the ratio of the mean signal from a region of interest (ROI) at the level of the verumontanum (the "reference region") to the standard deviation from the mean signal in an artifact-free ROI in the rectum.

Results: Both readers found geometric distortion and signal graininess significantly more often at 3.0 T than at 1.5 T (P < .0001, all comparisons). Reader 2 (but not reader 1) found ghosting artifacts more often at 3.0 T (P = .001) and blurring more often at 1.5 T (P = .006). Mean SNR at the urethra (87.92 ± 27.76) at 3.0 T was 1.43 times higher than at 1.5 T (64.51 ± 14.96) (P < .0001).

Conclusions: At 3.0 T (as compared to 1.5 T), increased SNR on prostate DW-MRI comes at the expense of geometric distortion and can also lead to more pronounced ghosting artifacts. Therefore, to take full advantage of the benefits of 3.0 T, further improvements in acquisition techniques are needed to address DW-MRI artifacts corresponding to higher field strengths.

Keywords: Diffusion-weighted imaging; MRI; artifacts; prostate cancer.

Copyright © 2013. Published by Elsevier Inc.

Figures

Figure 1

Susceptibility-induced image distortion at 3.0T. Shown are three consecutive axial slices. The air in the rectum or within the balloon of the endorectal coil causes local magnetic field inhomogeneity and susceptibility-related artifacts. Images were obtained with the following imaging parameters: b-value = 1000 s/mm2, TR/TE = 3500/76.8 ms, 2 NEX, matrix 128×128, FOV 160×160 mm2, resolution 1.25×1.25×3 mm3.

Figure 2

3.0T DW-MR images showing the impact of field inhomogeneity. Ghosting artifacts of the bright rectal wall are seen on the right and left sides of prostate on three consecutive axial slices. Due to the location of the prostate within the field-of-view, ghosting artifacts have not wrapped back onto the prostate. Sequence parameters are identical to those used for Figure 1.

Figure 3

Motion/ghosting and susceptibility-related artifacts at 3.0T DW-MRI. Shown are three consecutive axial slices. Artifacts that occur close to or through the prostate can obscure the anatomy and limit tumor detection. Sequence parameters are identical to those used for Figure 1.

Figure 4

Box-and-whisker plot of the SNRs of the reference regions in 53 patients imaged at 1.5T and 53 patients imaged at 3.0T. The red line indicates the median, the whiskers indicate the range, and the box indicates the 25th and 75th percentiles. Comparison intervals are drawn using notches. The interval endpoints are the extremes of the notches or the centers of the triangular markers.

Figure 5

(A) Mesh surface displays of the average normalized SNRs (nSNRs) of all patients for each field strength derived from b=0 images. Both displays demonstrate a drop in nSNR as a function of distance from the coil, with a rapid drop in SNR towards the anterior portion and edges of the prostate. Average nSNR profiles in (B) three horizontal (right-to-left, RL) planes located in the left, center and right planes and (C) two vertical (superior-inferior, SI) planes located at distances of 0.55 cm and 1.65 cm from the bottom of the coil are show. Overall, the normalized profiles are similar for the two field strengths in both the RL and SI planes. However, the normalized profile of 3.0T images is consistently lower than that of 1.5T images.

Figure 5

(A) Mesh surface displays of the average normalized SNRs (nSNRs) of all patients for each field strength derived from b=0 images. Both displays demonstrate a drop in nSNR as a function of distance from the coil, with a rapid drop in SNR towards the anterior portion and edges of the prostate. Average nSNR profiles in (B) three horizontal (right-to-left, RL) planes located in the left, center and right planes and (C) two vertical (superior-inferior, SI) planes located at distances of 0.55 cm and 1.65 cm from the bottom of the coil are show. Overall, the normalized profiles are similar for the two field strengths in both the RL and SI planes. However, the normalized profile of 3.0T images is consistently lower than that of 1.5T images.

Figure 5

(A) Mesh surface displays of the average normalized SNRs (nSNRs) of all patients for each field strength derived from b=0 images. Both displays demonstrate a drop in nSNR as a function of distance from the coil, with a rapid drop in SNR towards the anterior portion and edges of the prostate. Average nSNR profiles in (B) three horizontal (right-to-left, RL) planes located in the left, center and right planes and (C) two vertical (superior-inferior, SI) planes located at distances of 0.55 cm and 1.65 cm from the bottom of the coil are show. Overall, the normalized profiles are similar for the two field strengths in both the RL and SI planes. However, the normalized profile of 3.0T images is consistently lower than that of 1.5T images.