S2087
Physics - Image acquisition and processing
ESTRO 2026
guidelines. For the RTplanbio, regions with elevated APT-weighted signal were added to the GTVclin, according Tang et al.'s methods3, to construct a biological GTV (GTVbio). No margin was added to the GTVbio to generate the CTVbio. A Wilcoxon signed- rank test compared the volumes of the GTVclin, GTVbio, CTVclin and CTVbio. For both plans, recurrences were classified in-field, marginal, or distant if >80%, 20-80%, <20% of the recurrence volume fell within the 95% isodose line4, respectively. A patient had a change in pattern-of-failure if the recurrence was classified in-field for the RTplanclin and marginal or distant for the RTplanbio. An exact binomial test evaluated whether the probability of changes in pattern-of-failure with the RTplanbio was below 0.20; a significant result suggests similar pattern-of-failure as the RTplanclin. Maximum doses (Dmax) and mean doses (Dmean) to normal brain, brainstem, chiasm and optic nerves were compared via paired t-tests (Bonferroni-corrected α =0.005). Results: From January 2024 to August 2025, 36 patients were included. Four were excluded post-hoc: three did not complete radiotherapy and for one, follow-up MRI was discontinued. As yet, twenty-two patients (16M, 6F) had recurrences and were included in this analysis. The GTVbio was larger than the GTVclin (median: 61.7 vs 54.8-mL, p=0.00036); the CTVbio was smaller than the CTVclin (median 61.7 vs 197.1-mL, p=2.4x10-7). A change in pattern-of-failure with the RTplanbio occurred in 1 of 22 patients. The exact binomial test showed that the probability of changes in pattern-of- failure was significantly lower than 0.20 (p=0.048). Apart from Dmax to normal brain (p=0.01), the Dmax and Dmean to examined OARs were significantly lower in the RTplanbio (p-value range: 4.0x10-13 - 0.00056).
Figure2.: The ratios (in %) of different types of reconstructed materials by kernels Conclusion: This investigation demonstrates that Direct Density kernel reconstruction can be used for mass-density mapping, but slightly underestimation appears in terms of mass density, which also have an effect on material decomposition. The workflow advantage is the elimination of kVp-dependent calibration curves, but deeper analysis has to be considered before application in clinical practice. Keywords: Direct density, mass density mapping Amide proton transfer-weighted MRI for improved target delineation of glioblastoma: a prospective cohort study Patrick L.Y. Tang 1,2 , Marion Smits 1,3 , Remi A. Nout 2 , Caroline van Rij 2 , Cleo Slagter 2 , Linda Stienstra 2 , Annemarie T. Swaak-Kragten 2 , Erica J. Venema 2 , Esther A.H. Warnert 1 , Alejandra Méndez Romero 2 1 Department of Radiology & Nuclear medicine, Erasmus MC, University Medical Center Rotterdam, Rotterdam, Netherlands. 2 Department of Proffered Paper 4094 Radiotherapy, Erasmus MC Cancer Institute, University Medical Center Rotterdam, Rotterdam, Netherlands. 3 Medical Delta, TU Delft, Delft, Netherlands Purpose/Objective: Glioblastoma is notorious for tumor infiltration that is invisible on conventional MRI. Consequently, a 15-mm CTV-margin is recommended1, creating large target volumes and increased risk of radiotherapy-related adverse events. Amide proton transfer (APT)-weighted MRI may better visualize tumor infiltration2, enabling improved GTV delineation and smaller CTV-margins, potentially lowering risks of radiotherapy-related adverse events. Here, we generated conceptual radiotherapy plans (RTplanbio) incorporating APT- weighted MRI, and compared pattern-of-failure and dose to OARs with the clinical radiotherapy plan (RTplanclin). Material/Methods: In a prospective cohort, APT-weighted MRI was acquired pre-radiotherapy. GTVclin and CTVclin delineation for the RTplanclin followed clinical
Made with FlippingBook - Share PDF online