ESTRO 2025 - Abstract Book

S3823

Physics - Radiomics, functional and biological imaging and outcome prediction

ESTRO 2025

Bland-Altman plots comparing Spearman correlation (A) and Dice similarity coefficient (DSC, B and C) between V DPRNN and V ESRD maps. The vertical axis indicates the inter-method differences of the metrics.

Conclusion: The study demonstrates the feasibility of using machine learning to analyze SRD patterns for lung function assessment, offering an alternative perspective on leveraging respiratory temporal information for ventilation mapping. The improved mapping could potentially benefit radiotherapy planning through more accurate identification of functional lung regions.

Keywords: machine learning, 4DCT, functional imaging

References: 1. J. Midroni et al., “Incorporation of Functional Lung Imaging into Radiation therapy Planning in Patients with Lung Cancer: A Systematic Review and Meta-Analysis,” International Journal of Radiation Oncology*Biology*Physics, vol. 120, no. 2, pp. 370–408, Apr. 2024, doi: 10.1016/j.ijrobp.2024.04.001. 2. Y.-H. Huang et al., “Constructing Surrogate Lung Ventilation Maps from 4DCT-derived Subregional Respiratory Dynamics,” International Journal of Radiation Oncology*Biology*Physics, Nov. 2024, doi: 10.1016/j.ijrobp.2024.11.074. 3. J. Kipritidis et al., “The VAMPIRE challenge: A multi ‐ institutional validation study of CT ventilation imaging,” Medical Physics, vol. 46, no. 3, pp. 1198–1217, Dec. 2018, doi: 10.1002/mp.13346.

3474

Digital Poster Spatial and temporal changes in functional magnetic resonance imaging parameters in cervical cancer chemoradiotherapy Mohammed Abdul-Latif 1,2 , Amani Chowdhury 1,2 , Hannah Tharmalingam 1 , Peter Hoskin 1,2 , Yatman Tsang 3 1 Department of Clinical Oncology, Mount Vernon Cancer Centre, Northwood, United Kingdom. 2 Department of Cancer Sciences, University of Manchester, Manchester, United Kingdom. 3 Radiation medicine program, Princess Margaret Cancer Centre, Toronto, Canada Purpose/Objective: Locally advanced cervical cancer is treated with chemoradiotherapy and brachytherapy. Functional magnetic resonance imaging (fMRI) can identify biological features of cervical cancer such as necrosis, hypercellularity and perfusion. Evidence is limited in how spatial and temporal changes in these features impact on treatment outcomes 1 . We investigate ways to characterise their spatial distribution and changes across treatment.

Material/Methods: Patients underwent multiparametric (Diffusion weighted (DWI), dynamic contrast enhanced (DCE) and Blood oxygen level dependant (BOLD)) fMRI before (S1) and at the end of chemoradiotherapy (S2). Data