ESTRO 2025 - Abstract Book

S3824

Physics - Radiomics, functional and biological imaging and outcome prediction

ESTRO 2025

for apparent diffusion coefficient (ADC, from DWI-MRI), Ktrans (DCE-MRI) and R2* (BOLD-MRI) was extracted and identified using X,Y and Z coordinates in tumour volumes. 3D heatmaps were generated demonstrating parameter distribution at S1 and S2. Skewness and kurtosis for parameter expression was measured for whole tumour. Spatial descriptive statistics (Ripley’s K function) were performed to assess for random, regular or aggregate (cluster) parameter distributions. A subset analysis of patients with persisting residual disease at 3 months after treatment was performed. Results: Thirteen patients were imaged at S1 and S2. Three patients (EM04, EM08 and EM13) had residual disease at 3 months after completing treatment. ADC, Ktrans and R2* are higher at the end of treatment compared to before treatment for whole tumour volumes (p<0.05). 3D heatmaps (Example seen in Figure 1. using EM04) demonstrate heterogenous parameter expression across tumour volumes, with narrower ranges of expression at end of treatment. Parameter skewness and kurtosis are lower at the end of treatment compared to before (p<0.05). Spatial descriptive statistics using Ripley’s K function demonstrate parameters are distributed in a cluster fashion, rather than random or regular. For patient EM04, residual disease was in the left cervix. When dichotomised by laterality, ADC was significantly lower and R2* higher on the left cervix at end of treatment (Figure 2.). ADC and Ktrans were also lower before treatment on the left side. For patient EM13, residual disease was in the superior cervix. When dichotomised by superiority, ADC was significantly lower and R2* higher on the superior cervix at the end of treatment. ADC was also lower before treatment superiorly. EM08 had global disease so was not analysed spatially.