ESTRO 2020 Abstract book

S1025 ESTRO 2020

pseudoCT methods on rectal toxicity predictors in an MR- only RT pathway for prostate cancer. Material and Methods Twelve patients were recruited to an investigational study (*****). Patients underwent a large field of view diagnostic quality T2-weighted MRI scan on the same day as a standard-of-care (SoC) RT planning CT scan for prostate RT. All patients were planned for radical RT with 60Gy in 20 fractions. Four pseudoCTs were created per patient: “CT0” (bulk-density of zero Hounsfield Units (HU); 2) “CTpop” (bulk-density equal to population average HU); “CTstrat” (replicating a commercial system of density- stratification into 4 tissue classes); 4) “SynCT” (voxel- based HU predicted from T2tse MRI using a 2D neural network based on U-Net architecture). Treatment plans were calculated using each pseudoCT. Rectal dose volume histograms (DVH) and rectal dose surface maps (DSM) were extracted for each dose distribution and used to assess grade-2 rectal bleeding risk. Errors introduced into rectal DVH- and DSM-derived grade 2 rectal bleeding risk were evaluated compared with the gold-standard SoC CT. Results Mean (± 1 standard deviation) HU error for the SynCTs compared to the SoC CT was -5.7HU (±6.1), with mean absolute error of 40.1HU (±4.4). CT0, CTpop, CTstrat and SynCT introduced maximum absolute errors of 0.8%, 0.5%, 0.2% and 0.3% into DVH-based rectal bleeding risk prediction respectively. DSM-derived risk predictions were unaffected in all cases.

Fig 1. SoC CT, and pseudoCTs used for dose calculation

Conclusion The novel positioning setup presented here allows for high quality imaging of brain patients in SRS. The SNR is in the frontal part of the head is superior to the standard radiology setup, in the central part is equivalent and in the posterior part has a slightly lower SNR, which is still acceptable for contouring. Motion artifacts that influence contouring could also be avoided by using a mask setup. PO-1751 Effect of pseudoCT methods on dose-derived rectal toxicity prediction in MR-only prostate RT C. Thomas 1,2 , I. Dregely 2 , I. Oksuz 2 , T. Guerrero- Urbano 2,3 , A. Greener 1 , A. King 2 , S. Barrington 2 1 Guy's and St. Thomas' NHS Foundation Trust, Medical Physics, London, United Kingdom ; 2 King's College London, School of Biomedical Engineering and Imaging Sciences, London, United Kingdom ; 3 Guy's and St. Thomas' NHS Foundation Trust, Clinical Oncology, London, United Kingdom Purpose or Objective MR-planning for radiotherapy (RT) is increasingly popular and strategies for toxicity-based dose adjustment are emerging. MR-only RT requires pseudoCT generation for dose calculation. This study evaluates the effect of 4

Fig 2. Box plot of errors in DVH-derived G2 rectal bleeding toxicity. Full range of errors (whiskers), interquartile range (boxes), median (line) and mean (cross) are shown. Median for CTstrat is zero. Conclusion Tissue-stratification (CTstrat) and voxel based methods (SynCT) offer an improvement in accuracy over bulk density approaches, but all four pseudoCT methods offer acceptable resolution of rectal dose and G2 rectal bleeding risk prediction based on DVH and DSM. The voxel-based technique (SynCT) allows increased HU resolution but appears unnecessary for assessment of acute toxicity risk in prostate cancer beyond that offered by CTstrat.