ESTRO 2024 - Abstract Book

S5130

Physics - Radiomics, functional and biological imaging and outcome prediction

ESTRO 2024

presented as heatmaps in Figure 2 A for all tumors, primary tumors, and lymph node tumors. Strong, significant correlations were found between all eponymous parameters of TM and ETM models with PCC all > 0.79. Several significant correlations were found between different parameters from different models but only few were robust against a stratification by lesion type. Robust correlations were found for K trans (TM, ETM), which correlated strongly to moderately with K ep (TM, ETM, BM), v e (TM, ETM) and A (BM) with median PCC of 0.70, 0.63 and 0.44, respectively; and for A (BM) which correlated strongly with AUCs (median PCC = 0.76). In the T-stage stratified analysis of the primary tumors we found that one parameter, K ep (TM), significantly decreased with increasing T stage (p = 0.008; Figure 2 B). Most correlations between parameters and the association to T stage were stronger for HPV negative tumors.

Conclusion:

Individual AIF was preferred over population AIF for accurate pharmacokinetic modelling of DCE-MRI in HNC. Model parameters and their correlations were affected by the lesion type, HPV status and T stage. One Tofts parameter was significantly associated to T stage.

Figure 1: Population AIFs obtained with different approaches.