ESTRO 2024 - Abstract Book

S898

Clinical - CNS

ESTRO 2024

Chung M, Chung WK (2021) Clinical approach of low-dose whole- brain ionizing radiation treatment in Alzheimer’s disease dementia patients. J Alzheimers Dis 80, 941 – 947.

Kim S, Chung H, Mai HN, Nam Y, Shin SJ, Park YH, Chung MJ, Lee JK, Rhee HY, Jahng G-H, Kim Y, Lim YJ, Kong M, Moon M, Chung WK (2020) Low- dose ionizing radiation modulates microglia phenotypes in the models of Alzheimer’s disease. Int J Mol Sci 21, 4532. Kim DY, Kim JS, Seo YS, Park WY, Kim BH, Hong EH, Kim JY, Cho SJ, Rhee HY, Kim A, Kim KY, Oh DJ, Chung WK. Evaluation of Efficacy and Safety Using Low Dose Radiation Therapy with Alzheimer's Disease: A Protocol for Multicenter Phase II Clinical Trial. J Alzheimers Dis. 2023;95(3):1263-1272.

2258

Digital Poster

Feasibility of using an automatic segmentation program in the treatment of metastatic spine lesions.

Thierry Gevaert, Selma Ben Mustapha, Bilal Ait Mahha, Racell Nabha, Sven Van Laere, Mark De Ridder

UZ Brussel, Vrije Universiteit Brussel, Radiotherapy, Brussels, Belgium

Purpose/Objective:

This study aims to compare the accuracy and reliability of automatic clinical target volume (CTV) delineation, using the Elements Spine SRS software by Brainlab that includes the module Elements SmartBrush Spine (ESS) (v3.0, Munich, Germany) software program. The use of a dedicated contouring tool may reduce inter- and intraobserver variability and allow for greater precision. Secondly, the research also explores the potential of extending the automatically generated contour and trying to find an answer to the question if and why this improves the quality of automatic delineation compared to that of an expert physician.

Material/Methods:

A patient group of 86 cases receiving palliative radiotherapy was used to demonstrate the feasibility of the ESS module. For each case, manual delineation by an expert radiation oncologist (SBM) was compared to automatic delineation generated using the ESS module. Spine segmentation was used for delineation of the clinical target volumes (CTV), i.e. the whole or multiple vertebrae. Several metrics were utilized to assess the similarity between the CTV contours, including the dice similarity coefficient (DSC), Hausdorff distance (HDmax), 95% Hausdorff distance (HD95), absolute volume differences (AVD), relative volume differences (RVD) and distance of center of mass (DCM). Here, averages and standard deviations for the different metrics were reported. After the delineation of the structures was made (raw delineation), we extended the delineation of the CTV by 1 mm in ESS (extended delineation). The differences in DSC between raw and extended delineation were further examined. In a linear regression model, we tested whether statistically significant differences could be observed between the different regions of the spine where the lesion(s) was/were located (cervical, lumbar or thoracic vertebrae) and between whether or not the lesion