Twenty patients with epidermoid who had undergone MRI including SS-EPI DWI and a prototype TGSE-BLADE DWI between December 2019 and October 2021 were included in this study. Two patients were excluded because the lesions were too small to evaluate.

This retrospective study was performed in accordance with the Declaration of Helsinki and was approved by Kyoto University Graduate School and Faculty of Medicine, Ethics Committee and the need to obtain informed consent was waived.

All patients underwent brain MRI using a 3-T scanner (MAGNETOM Prisma or Skyra; Siemens Healthineers, Erlangen, Germany) with a 64-channel head/neck coil or a 32-channel head coil. Two DWI sequences (SS-EPI DWI and TGSE-BLADE DWI) were acquired along with T2-weighted imaging (T2WI). SS-EPI DWI used: repetition time (TR), 5000 (5500) ms; echo time (TE), 53 (77) ms; field of view (FOV), 220 × 220 mm; matrix, 320 × 320; 35 slices; parallel imaging factor, GeneRalized Autocalibrating Partial Parallel Acquisition (GRAPPA) 3 × ; voxel size, 0.7 × 0.7 × 3 mm; diffusion time, 21.2 (27.4) ms; and acquisition time, 1 min 47 s. As a prototype sequence, TGSE-BLADE DWI used: TR, 7000 (7700) ms; TE, 46 (61) ms; flip angle (FA), 120°; FOV, 220 × 220 mm; matrix, 320 × 320; 35 slices; parallel imaging factor, N/A; voxel size, 0.7 × 0.7 × 3 mm; diffusion time, 17.6 (24.9) ms; and acquisition time, 3 min 53 s (4 m 16 s). Note that the values in parentheses are those used with the MAGNETOM Skyra (Table 1).

T2WI was acquired in the same slice position with 2 DWIs, and the details of T2WI were as follows: axial acquisition; TR, 3200 ms; TE, 79 ms; FA, 120°; FOV, 185–199 × 220 mm; matrix, 378–406 × 448; 35 slices; parallel imaging factor, 2; voxel size, 0.5 × 0.5 × 3 mm; and acquisition time, 1 min 44–50 s. T2WI was used as the reference in this study.

Other image sequences were also obtained for clinical scans but were not used for this study.

The following image analyses were performed for SS-EPI DWI and TGSE-BLADE DWI.

(1) Entire lesion analysis. Two board-certified radiologists (S.I. and Y.W., with 9 and 10 years of experience in neuroradiology, respectively) separately placed ROIs encircling the epidermoid using ImageJ version 1.53e (https://imagej.nih.gov/ij/). First, each radiologist placed ROIs over the entire epidermoid lesions in all slices on trace-weighted SS-EPI DWIs without referring to any other image sequences and saved the ROI files independently. ROIs on TGSE-BLADE DWI were placed in the same manner 3 weeks later to avoid any learning effect.

ADC of epidermoid cyst was measured with the ROIs of each rater on SS-EPI DWI and TGSE-BLADE DWI, then the intraclass correlation coefficient (ICC) between two raters was obtained for each DWI.

The areas of ROIs placed by the two raters were measured and compared using the Dice coefficient, measured using the following equation where |A| represents the area of the ROI from Rater 1, and |B| represents the area of the ROI from Rater 2 [30]. |A ∩ B| represents the overlapping area of the ROIs from Raters 1 and 2.

$$}\, = \,\frac} \cap \left. } \right| \times 2} \right.}}}| + |}|}}$$

As the equation indicates, a Dice coefficient of 1 indicates that the ROIs of the two raters are identical, and 0 indicates no overlap. Dice coefficient was obtained for SS-EPI DWI and TGSE-BLADE DWI, respectively.

(2) Selected slice analysis. To evaluate the difference in ADC between SS-EPI and TGSE-BLADE DWIs, one board-certified radiologist (S.O., with 13 years of experience in neuroradiology) selected the most appropriate slice and carefully placed the ROIs slightly smaller than the epidermoid outline to avoid artifacts in SS-EPI and TGSE-BLADE DWIs (juxta-marginal ROIs). ADC for the carefully selected slice was analyzed.

Image quality of the trace-weighted DWI of each sequence type was qualitatively assessed for geometric distortion, susceptibility artifacts, lesion conspicuity, diagnostic confidence, and overall image quality using a 4-point Likert scale [23]. The criteria for image assessment are defined in Supplementary Table 1. The assessment was conducted by three board-certified radiologists (S.I., Y.W., and S.O. with 9, 10, and 13 years of experience in neuroradiology, respectively).

The phantom study was performed on a Mini Diffusion phantom (CaliberMRI, Boulder, CO, USA) with a MAGNETOM Prisma or Skyra system using the corresponding receiver coil. The phantom contains three high-performance liquid chromatography water vials (reference), and 10 polyvinylpyrrolidone (PVP) vials (two each of 10%, 20%, 30%, 40%, and 50% PVP). SS-EPI DWI and TGSE-BLADE DWI were scanned 4 times, respectively. ROIs were placed in the trace DWI images and superposed onto ADC maps. The accuracy of SS-EPI DWI and TGSE-BLADE DWI was evaluated by comparison between measured and theoretical ADCs corresponding to the temperature of the phantom.

The relationship between ADCs of SS-EPI and TGSE-BLADE was assessed using Pearson’s correlation coefficient (r) if both values were normally distributed and Spearman’s rank correlation coefficient (ρ) if one or both values was not normally distributed. Normal distribution was evaluated using the Shapiro–Wilk test. Agreement among ADCs in the selected slice analysis from SS-EPI and TGSE-BLADE was assessed by the ICC. Dice coefficient of the area of the ROI for SS-EPI DWI and TGSE-BLADE DWI was compared using the paired t-test. The McNemar test was used by three raters to analyze image quality. The accuracy of ADC in the phantom study was evaluated by ICC.

Values of P < 0.05 were considered significant. Statistical analysis was performed using JMP Pro version 16.0 software (SAS Institute Inc., Cary, NC, USA) and MedCalc version 16.2.0 (MedCalc Software, Ostend, Belgium). Continuous variables are presented as medians and interquartile ranges unless otherwise noted.