S2796
RTT - RTT contouring, target definition, and treatment planning
ESTRO 2026
The use of magnetic resonance imaging (MRI) in radiotherapy planning has increased over the years as MRI represents a non-ionizing modality providing both an improved contrast between soft tissues compared to the computed tomography (CT) and an opportunity for hypoxia-informed therapy. A limitation toward MRI-only dose calculations is the absence of direct physical link between MR signal and electron density (ED) of the tissues. The latter is required to compute the dose distribution. One approach around this limitation relies on quantitative MRI to approximate the tissues’ ED. MRI sequences such as the ultrashort echo time (UTE) sequence can be used to obtain a signal related to the proton density of the tissues [4]. Thus, it would be possible to derive ED maps from UTE images with a tissue correction factor that would account for the bone that is difficult to measure otherwise . Material/Methods: A study using 3T MR and CT images from thirteen patients with tumors in the head-and-neck region.The UTE signal of the tissues was expressed relatively to the UTE signal of pure water. The ED value was obtained by using an equation derived from [1] and [2].A categorical mask of the main tissue types (air, bones, fat and soft tissues) was derived from theCT. The mask was used to apply a tissue correction based on mass density values from the literature [3] (Fig 1).
(95% CI: 0.665–0.783).
Conclusion: MRCDL demonstrated similar performance on FLEX images compared to CUBE images. The overall and the organ-specific accuracy was close for 7 structures. The accuracy was influenced for 3 structures (pelvis-body, prostate, seminal-vesicles) due to the differences in image acquisition (particularly reduced contrast uniformity and edge degradation in FLEX images) and patient conditions (arms included, benign prostate hyperplasia). These findings support integration of MRCDL into MR-based RT workflows, while highlighting caution with sequence variability and specific clinical conditions. Keywords: Pelvis segmentation, deep learning, MRI Proof of concept of dose calculations using MRI- based electron density maps in head and neck cancers Nils Tanneau 1 , Charlène Bouyer 2 , Benjamin Leporq 1 , Frank Pilleul 1,3 , Vincent Grégoire 4 , Olivier Beuf 1 1 CREATIS, INSA-Lyon, Universite Claude Bernard Lyon 1, CNRS, Inserm, UMR 5220, U1294, Lyon, France. 2 Département de physique médicale, Centre Léon Bérard, Lyon, France. 3 Département de radiologie, Centre Léon Bérard, Lyon, France. 4 Département de radiothérapie, Centre Léon Bérard, Lyon, France Poster Discussion 4085
Dosimetry plans used for the treatment of each patient were recomputed on the MRI scans without re- optimization using MONACO (Elekta AB). Resulting dose maps were analyzed through dose differences at 95% of tumoral volumes (TVs) and at 2% organs-at- risks volumes (OARs). Gamma passing rates at 3%/3mm and 2%/2mm were computed using PyMedPhys[3]. Results: Gamma passing rates values at 3%/3mm are 99.4±0.5% and 99.1±0.5% for the doses computed from UTEtissues and UTEwater, respectively. At 2%/2mm, mean values are 97.9± 1.5% and 97.1±1.7%. The dose differences were negative in most OARs and TVs for the UTEtissues maps while closer to zero in UTEwater maps (Fig 2).
Purpose/Objective:
Made with FlippingBook - Share PDF online