ESTRO 2024 - Abstract Book

S3924

Physics - Image acquisition and processing

ESTRO 2024

the repeated full acquisitions were compared: least-squares restoration and Jacobian modulation [5], referred to as topup LS and topup Jac , respectively. An EPI sequence with only b = 0 s/mm 2 images (voxel size = 3.5 x 3.5 x 4.0 mm 3 , 30 slices, b -values (averages) = 0 (2) s/mm 2 , TR = 4308 ms, TE = 75 ms, scan time = 13 s) was also acquired with reversed PE, to compare the distortion correction achieved using full and b 0-only ( topup b0 ) acquisitions. Note that for all three topup -corrected datasets, the correction is calculated from b = 0 s/mm 2 images, and is subsequently applied to all b value images. The topup b0 method using a separate b 0 sequence was included to replicate how this approach could be used in practice with the shortest acquisition time (3:40 + 0:13); for topup b0 only Jacobian modulation can be used when applying topup [5]. ADC maps for all sequences were calculated using b = 150 and 500 s/mm 2 images, and median values were extracted from regions of interest.

Results:

Figure 1 shows ADC boxplots for 13 phantom vials, for uncorrected EPI, SPLICE, and three topup corrections. All methods of topup correction gave similar values to the uncorrected EPI sequence, with all EPI-based images giving a mean absolute bias of 0.02 µm 2 /ms. SPLICE tended to overestimate ADC, with a higher mean absolute bias of 0.06 µm 2 /ms.