Click here to


Are you sure ?

Yes, do it No, cancel

Pulse Sequence Optimization for Non-EPI Diffusion-Weighted Imaging Sequences for Head & Neck On a 1.5T MR-Linac

BA McDonald1*, LL Zhu1, S Mulder1, M Alex Dresner2, ASR Mohamed1, S Ahmed1, R He1, Y Ding1, CD Fuller1, (1)UT MD Anderson Cancer Center, Houston, TX, (2)Philips Healthcare, Philadelphia, PA


(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: To optimize two geometrically stable diffusion-weighted imaging (DWI) pulse sequences, turbo spin echo (TSE) and split acquisition of fast spin-echo signals (SPLICE), for biologically-directed adaptive radiotherapy for head & neck cancers on a 1.5T MR-linac.

Methods: A series of images were acquired on the Elekta Unity MR-linac on a volunteer: TSE with half-scan factors 0.6-0.9, SPLICE with refocusing flip angles 60°-100°, and TSE and SPLICE with different fat saturation (FS) methods (SPIR, SPAIR, STIR, no FS). Organs with high diffusion were contoured to calculate signal-to-noise ratio (SNR). Images were scored qualitatively by three expert observers (radiologist, radiation oncologist, MRI physicist) blinded to acquisition parameters. Categories included SNR, contrast-to-noise ratio (CNR), blurring, and quality of FS. Observers also ranked images based on overall impression. Univariate linear regression was performed to determine the categories that contributed most to overall rank. The images were also acquired on a QIBA diffusion phantom to calculate percent ADC bias (%bias).

Results: From the qualitative analysis, the preferred SPLICE flip angles were 60° and 100°, TSE half-scan factor was 0.9, and FS was SPIR for both TSE and SPLICE. SNR and CNR were the most important factors for the b=0 images, and blurring was most important for b=500. In vivo SNR results were surprising, with STIR performing highest of the FS methods on both TSE and SPLICE despite performing poorly in the qualitative analysis. SNR also decreased with increasing TSE half scan factor, which contradicts the qualitative rankings. %bias values grew larger with decreasing ADC values but did not show any consistent trends with flip angle, half-scan factor, or FS.

Conclusion: Optimal acquisition parameters for both SPLICE and TSE could be selected based on both qualitative and quantitative (%bias) analyses. Quantitative results and quantitative SNR values were dissimilar, warranting further investigation of the SNR metric used.

Funding Support, Disclosures, and Conflict of Interest: Brigid McDonald receives research funding from NIH (F31DE029093) and an MDA/UTHealth GSBS Kopchick Fellowship. Clifton Fuller receives research funding from NIH (R01DE028290, 1R01DE025248/R56DE025248, 1R01CA218148, P30CA016672, P50CA097007, 1R01CA214825) and NSF (NSF1557679, NSF1933369) and direct industry grants from Elekta AB. McDonald and Fuller have received travel support and honoraria from Elekta AB.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email