ESTRO 2023 - Abstract Book

S1323

Digital Posters

ESTRO 2023

Table 1. Similarity metrics (median) 95% HD [mm]

95% HD [mm] sDSC sDSC

Manual

cDL 7.4 3.8 3.0 4.9 6.5

Manual cDL 0.70 0.84 0.81 0.96 0.82 0.98 0.73 0.93 0.80 0.94

L1 10.2 L2 15.8 L3 7.3 L4 11.1 IP 16.3

Conclusion This study shows that the use of deep learning auto-segmentation results in a higher interobserver agreement in the delineations of the nodal target volumes for breast cancer patients. All similarity metrics improved compared to the manual delineations. Therefore, implementation of deep learning auto-segmentation results in more consistent and reproducible segmentations, while reducing the delineation time by about one third.

PO-1628 Dosimetric Evaluation of MR-based Deep Learning Automatic Contouring in the Pelvis

J. Wyatt 1 , L. Rusko 2 , R. Pearson 3 , B. Kolozsvári 2 , B. Deák-Karancsi 2 , V. Czipczer 2 , Z. Karancsi 4 , E. Ruff 2 , B. Irmai 2 , B. Tass 2 , K. Hideghéty 2 , S. Petit 5 , M. Capala 5 , F. Wiesinger 6 , R. Maxwell 1 , H. McCallum 1 1 Newcastle University, Institute of Translational and Clinical Research, Newcastle upon Tyne, United Kingdom; 2 GE Healthcare, Computer Science, Budapest, Hungary; 3 Newcastle upon Tyne Hospitals NHS Foundation Trust, Northern Centre for Cancer Care, Newcastle upon Tyne, United Kingdom; 4 Ge Healthcare, Computer Science, Budapest, Hungary; 5 Erasmus MC Cancer Institute, Department of Radiotherapy, Rotterdam, The Netherlands; 6 GE Healthcare, Magnetic Resonance Physics, Munich, Germany

Purpose or Objective