ESTRO 2024 - Abstract Book

S4567

Physics - Machine learning models and clinical applications

ESTRO 2024

these critical principal components and the respective original radiomics shape features, we pinpointed ‘Maximum3DDiameter’, ‘ShapeFlatness’, ‘ShapeElongation’, and ‘ShapeSphericity’ as the most relevant shape features critical for liposarcoma classification.

Conclusion:

Both the deep learning approaches and the machine learning models demonstrated proficient capability in discerning between ALTs and high-grade liposarcomas. Notably, these approaches exhibit promising performance levels when solely considering tumor morphology as the discriminative feature.

Keywords: Sarcoma, Morphology, Classification

References:

[1] Navarro et al. “Development and External Validation of Deep-Learning-Based Tumor Grading Models in Soft-Tissue Sarcoma Patients Using MR Imaging”, Cancers, 2021

[2] van Griehuysen et al. “Computational Radiomics System to Decode the Radiographic Phenotype”, Cancer Research, 2017

[3] Lorensen et al. “ Marching cubes: A high resolution 3D surface construction algorithm”, SIGGRAPH, 1987

[4] Kipf et al. “Semi-supervised classification with graph convolutional networks”, ICLR, 2017

[5] Hamilton et al. “Inductive representation learning on large graphs”, Advances in neural information

processing

systems, 2017

[6] Fisher et al. “All models are wrong, but many are useful: Learning a variable’s importance by studying an entire class of prediction models simultaneously.” - Journal of Machine Learning Research, 2019

3053

Digital Poster

Knowledge Based Planning with dosimetric scorecards for Total Marrow Lymphoid Irradiation on Halcyon

Lesley Rosa 1 , Anthony Magliari 1 , Chunhui Han 2 , An Liu 2 , Jeffrey Y.C. Wong 2 , Terence Williams 2

1 Varian, Medical Affairs, Palo Alto, USA. 2 City of Hope, Radiation Oncology, Duarte, USA

Purpose/Objective: