ESTRO 2023 - Abstract Book

S1405

Digital Posters

ESTRO 2023

mm3) and slice orientation (axial vs sagittal). Volumes were segmented with an automatic software and normalized to the intracranial volume. Differences in volume estimations were calculated as 100 · (Vvariant - Vreference)/Vreference, taking as the reference the recommendations of the MPRAGE protocol: isotropic voxel 1x1x1mm3 and sagittal orientation. Hotelling statistical significance test was applied. Results The cerebellar white-matter was the structure with the greatest range of variation in volumetric quantification when modifying acquisition parameters, observing in average differences of 5.88% ± 2.4 % when performing axial vs sagittal acquisition; 7.7 ± 4.3 % volume underestimation when acquiring with 1.25x1.25x1.25mm3 voxel vs recommended 1x1x1mm3 and 0.6 ± 4.9 % when introducing anisotropy. Other structures, like the gray and white matter of the brain and the hippocampus showed maximum quantification deviations of 4%. Hotelling test showed no statistically significant differences, but the variations observed are not negligible.

Conclusion The use of isotropic and smaller voxels may be time consuming, but the inaccuracies introduced in quantitative volume estimations due to changes in acquisition parameters must be kept in mind, as they may be greater than the volumetric reduction due to radiation-induced atrophy, which is subject of study, and may lead to a misdiagnosis, resulting highly advisable not to vary the acquisition parameters from those recommended by the literature.

PO-1695 Biomechanical imaging biomarker to predict tumor progression after head and neck chemoradiotherapy

K. Fujimoto 1 , T. Shiinoki 1 , Y. Kawazoe 1 , Y. Yuasa 2 , W. Mukaidani 3 , Y. Manabe 1 , M. Kajima 1 , H. Tanaka 1

1 Yamaguchi university, Department of radiation oncology, Graduate school of medicine, Ube city, Japan; 2 Yamaguchi university hospital, Department of radiological technology, Ube city, Japan; 3 Yamagucni university hospital, Department of radiological technology, Ube city, Japan Purpose or Objective Biomechanical model (BM) based deformable image registration is a promising method for simulating morphological displacement of tumor and organs during radiotherapy (RT). Further, it is also reported as a beneficial imaging biomarker for organ function by capturing the mechanical behavior such as stress-strain relationship. The purposes are to construct