ESTRO 2024 - Abstract Book

S5001

Physics - Radiomics, functional and biological imaging and outcome prediction

ESTRO 2024

unfiltered Test and Retest HRSs were calculated. Dice Similarity Coefficients (DSC) as well as distances between center of masses (COM) were evaluated for corresponding HRSs. Wilcoxon signed-rank test (a = 0.05) was performed using SPSS (version 28.0.0.0) to test for significant differences between Test and Retest HRSs. GTVs smaller than 1 cm3 and HRSs smaller than 0.2 cm3 were excluded from the analysis, respectively.

Results:

A total of 43 corresponding datasets were analyzed. Mean ± standard deviation GTV was 21.3 ± 15.1 cm3, while mean unfiltered HRS Test and Retest volumes resulted in 2.9 ± 5.0 cm3 and 3.0 ± 6.1 cm3 (p=0.299), respectively. Mean filtered HRS Test and Retest volumes yielded 5.0 ± 9.0 cm3 and 4.6 ± 9.3 cm3 (p=0.078), whereas mean absolute HRS differences between Test and Retest were 0.8 ± 1.4 cm3 without filtering and 1.2 ± 1.1 cm3 with application of the closing filter (cf. Figure 1).

Regarding the original unfiltered HRSs, DSC analysis revealed a mean DSC of 0.20 ± 0.14 and 0.31 ± 0.25 for the filtered HRSs. The mean distance between the COMs for unfiltered Test and Retest HRSs resulted in 4.6 ± 2.6 cm and 4.5 ± 2.6 cm for filtered Test and Retest HRSs, respectively. Figure 2 visualizes DSC and distance of COM values for all datasets depending on the delineated GTV volume.