In the present study, we evaluated whether MS-EPI DWI and rFOV DWI can improve image quality of the bladder and bladder tumors and increase the diagnostic performance for muscle layer invasion compared with SS-EPI DWI. To our knowledge, no study has simultaneously compared the three types of DWI, SS-EPI DWI, MS-EPI DWI and rFOV DWI, in the same patient for the evaluation of bladder cancer.

Anatomical distortion due to susceptibility artifact was the least in MS-EPI DWI, which consistent with the results of previous studies in the pelvic area [18, 21, 25, 26]. In the present study, anatomical distortion was less in both MS-EPI DWI and rFOV DWI than in SS-EPI DWI, and was less in MS-EPI DWI than rFOV DWI. MS-EPI DWI employs interleaved k-space trajectories that split the acquisition into many shots, resulting in reduced distortion compared to SS-EPI DWI [21, 27]. The reduced FOV in DWI mitigates the impact of anatomical distortion by confining the region of interest to a limited area, thereby minimizing the influence of external magnetic susceptibility variations. In imaging the bladder, gas in nearby small intestine or rectum or that was introduced into the bladder during urinary drainage was included even in the reduced FOV, which could explain the insufficient reduction of anatomical distortion compared to MS-EPI DWI.

The higher clarity of bladder wall in MS-EPI DWI is probably due to the smaller EPI factor with MS-EPI DWI than the other two DWIs, which reduces distortion and the effects of blurring, consistent with a previous study [18]. In the present study, bladder wall clarity was the lowest for rFOV DWI. In contrast to our results, Juri et al. reported that rFOV DWI delineated the wall more clearly than full-field SS-EPI DWI [17]. However, Juri et al. used a thinner slice thickness for SS-EPI DWI than for rFOV DWI, whereas we used a thinner slice thickness for rFOV DWI and the resulting reduction in SNR might have obscured the wall.

Lesion conspicuity showed no significant variation among the three DWIs except in the case of one observer. Bladder cancer is well-visualized even on SS-EPI DWI, which may have resulted in the lack of discernible difference among the three types of DWI. In fact, there was no significant difference in SNR and CNR among the three DWIs in the present study, consistent with the finding of Chen et al. that readout-segmented EPI DWI and SS-EPI DWI have comparable SNR for bladder cancer [18]. Conversely, several studies have reported higher CNR for MS-EPI DWI and rFOV DWI than for SS-EPI DWI [18, 20]. This may be because CNR is a relative value and is dependent on image parameters such as magnetic field strength, minimum echo time, FOV, matrix, signal average, number of segments, echo spacing, gradient performance, b value combination, tissue T2 value, and k-space segmentation direction.

Since VI-RADS specifies DWI as the dominant sequence, DWI imaging with high image quality and low artifacts is required [6]. MS-EPI DWI and rFOV DWI showed no improvement in diagnostic performance over SS-EPI DWI for visualizing muscle layer invasion of bladder cancer. This may be attributable to the absence of significant differences in tumor conspicuity among the three DWI sequences. In addition, diagnostic performance would be unaffected if the tumor was located in an area devoid of distortion artifacts.

Based on these results, MS-EPI DWI had the best image quality overall of the three types of DWIs. However, MS-EPI DWI necessitates twice the imaging time of SS-EPI DWI. Therefore, rather than employing MS-EPI DWI on a routine basis, supplementary MS-EPI DWI should be conducted when distortion artifacts are present on the tumor, or if bladder wall clarity is inadequate on SS-EPI DWI. Even if MS-EPI DWI does not show an overall improvement in diagnostic performance, observations of tumors that are highly affected by artifacts on SS-EPI DWI would be easier to assess on MS-EPI DWI. rFOV DWI may have a lesser role in bladder evaluation because its overall image quality is inferior to MS-EPI and the area it can evaluate is smaller. However, the application of deep learning reconstruction and compressed sensing to DWI has recently been reported, and rFOV DWI has the potential to reduce imaging time and improve image quality and diagnostic performance [28, 29].

The present study has several limitations. The major limitations of the present study were its single-institutional retrospective design. It is desirable to validate the results of this study in a multicenter prospective study. Imaging time, voxel size, and number of averages varied among the three types of DWI. If any one parameter had been unified among the three sequences, the other parameters would not be identical and would also be inappropriate. Therefore, we maintained the parameters for each sequence that were optimized for use at our hospital. The image types of the DWI were not informed to the readers, but because image types differ in appearance, the readers may have been able to identify the image type of the DWI, which may have biased their assessment. The number of cases was not large enough to enable comparison of diagnostic performance in MIBC. Future studies are needed to evaluate the diagnostic performance of MS-EPI DWI and rFOV DWI.

Among the three types of DWI evaluated, MS-EPI DWI showed the least anatomical distortion and superior bladder wall delineation. There was no significant difference among the sequences in terms of diagnostic performance for muscle layer invasion of bladder cancer. MS-EPI DWI rather than rFOV DWI may be considered as an additional sequence in the case of image distortion or obscuration of the bladder wall or tumor on SS-EPI DWI.