S2468
Physics - Radiomics, functional and biological imaging, and outcome prediction
ESTRO 2026
timepoint, varying T1 resulted in the range [-0.394, 0.306] with mean and standard deviation - 0.003±0.162. When varying hematocrit, the normalized Ktrans range was [-0.171, 0.155] (-0.009±0.163). Changing to automatic time shift resulted in a normalized Ktrans range [-0.121, 0.266] (0.004±0.068). Normalized Ktrans distributions across all patients are illustrated in Figure 1.
Conclusion: This study identified an MRI-based radiomic feature, LRHGLE, as a potential imaging biomarker of tumor radiosensitivity. Dose–response simulation demonstrated its feasibility for estimating the radiation dose required to achieve LC and for guiding individualized, biologically adaptive dose prescription in cervical cancer. Keywords: Dose–response, Radiosensitivity, Personalization Digital Poster 3600 The effect of input uncertainty on the reproducibility of perfusion parameter maps Alicia Palmér 1 , Teo Asplund 1 , Bikash Panthi 2 , Mohamad El-Jammal 2 , Andrew Miller Elliott 2 , Eleni Konstantinopoulou 2 , Stina Svensson 1 , Caroline Chung 2 1 Research, RaySearch Laboratories, Stockholm, Sweden. 2 Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA Purpose/Objective: The purpose of the investigation was to evaluate the effect of varying input parameters for calculations of the perfusion imaging parameter Ktrans from Dynamic Contrast Enhanced (DCE) MRI data, when used for monitoring gross tumor volume (GTV) changes throughout treatment of high-grade gliomas. Material/Methods: For a selection of hematocrit (0.360, 0.445, and 0.540) and blood T1 (1200, 1321, 1441, 1561, and 1682 [ms]) based on physiologically plausible values in literature1,2, as well as a choice of vascular input function (VIF) shift (no time shift, or voxel-wise automatic time shift), 30 unique combinations were created. Ktrans maps were computed using the extended Tofts model3 and all parameter combinations for 17 patients, each with one pre- and one post-treatment DCE-MRI sequence. For each Ktrans map, an average was calculated within the GTV. The midpoint of the test ranges (hematocrit = 0.445, blood T1 = 1441 ms, and no time shift of VIF) was selected as a reference for each patient. Average Ktrans was normalized with respect to this reference combination. Results: Using the averaged Ktrans values normalized per
Figure 1: Ktrans for each tested value of parameters across patients. Each box shows the interquartile range (IQR), whiskers extending up to 1.5*IQR from the box, and points representing outliers.When comparing the trend of increasing or decreasing Ktrans between the timepoints, the trend reversed in five patients (29.4%) when only the T1 value at the second timepoint was varied, in four patients (23.5%) when hematocrit was varied, and only for patient 12 (5.9%) when applying automatic time shift. Figure 2 illustrates the patient- wise distributions of relative Ktrans changes.
Made with FlippingBook - Share PDF online