Background:The combination of Stereotactic Body Radiotherapy (SBRT) with immune checkpoint inhibitors (ICIs) has gained increasing interest due to its potential to enhance antitumor immune responses. However, the optimal timing, dose, and safety profile of this combined approach remain unclear, and available clinical evidence is highly heterogeneous. Methods:A systematic review of the literature was conducted in accordance with PRISMA guidelines. PubMed/MEDLINE, Embase, and Cochrane Library databases were searched for clinical studies evaluating the combination of SBRT and ICIs. Studies were included if they reported toxicity outcomes and involved patients with solid tumors treated with SBRT in combination with ICIs. Data on study design, patient characteristics, SBRT dose and fractionation, treatment sequencing, and treatment-related toxicities were extracted and qualitatively analyzed. Results:A total of 31 clinical studies met the inclusion criteria, encompassing 2,355 patients across multiple tumor types, including non-small cell lung cancer, melanoma, breast cancer, renal cell carcinoma, prostate cancer, hepatocellular carcinoma, and pancreatic or biliary malignancies. Fifteen studies employed concurrent SBRT-ICI administration, 13 adopted a sequential approach, and 3 included both strategies. SBRT dose and fractionation were highly variable, ranging from palliative regimens to ablative schedules, with reported BED10 values spanning approximately 43-113 Gy; detailed dosimetric data were lacking in about 38% of studies. The overall incidence of grade ≥ 3 treatment-related toxicity was comparable between concurrent and sequential approaches (approximately 12-15%). High-grade adverse events were predominantly immune-related, with pneumonitis more frequently reported in concurrent regimens, while gastrointestinal and dermatologic toxicities were slightly more common in sequential strategies. No consistent signal of increased severe toxicity attributable to the addition of SBRT was observed. Conclusions:Current clinical evidence suggests that the combination of SBRT and ICIs is generally feasible and does not appear to systematically increase the risk of severe toxicity compared with immunotherapy alone. However, substantial heterogeneity in study design, SBRT parameters, treatment sequencing, and toxicity reporting limits definitive conclusions regarding safety and efficacy. Future prospective trials with harmonized protocols, standardized toxicity attribution, and integrated translational endpoints are needed to define the optimal therapeutic window for SBRT-ICI combinations across different tumor types.

BACKGROUND:Radiomics involves extracting and analyzing quantitative imaging features to support medical decision-making, particularly in radiology and oncology. When applied to radiotherapy dose distributions, this approach, referred to as 'dosiomics', aims to identify the spatial dose patterns associated with treatment outcomes. However, software discrepancies in feature extraction may hinder reproducibility and limit the clinical adoption of radiomic/dosiomic models. PURPOSE:This study presents the first comprehensive evaluation of software agreement and feature reproducibility across tools in the field of dosiomics, assessing seven feature-extraction tools. The evaluation focused on the impact of built-in image pre-processing steps (e.g., interpolation and discretization), feature-extraction configurations (i.e., aggregation methods), and the morphological characteristics of the regions of interest (ROIs), such as the presence of holes or disconnected components. MATERIALS AND METHODS:Five open-source programs (MIRP, S-IBEX, RaCaT, SERA, and PyRadiomics) and two proprietary tools (SPAARC and RadiomiCRO) were evaluated. The Image Biomarker Standardization Initiative (IBSI) digital phantom was used to preliminarily assess software IBSI-compliance and to identify and exclude features with inconsistent implementation from subsequent analyses. Dosiomic features were then extracted from a digital dataset comprising eight Intensity Modulated Radiation Therapy (IMRT) dose distributions emulating a head and neck radiotherapy plan (available in both isotropic and anisotropic formats) and 10 ROIs, following a systematic feature extraction framework. The effects of pre-processing parameters, feature-extraction configurations, and ROI morphological characteristics were analyzed systematically. The evaluation metrics included the percentage of matching features across software to the third significant digit, the Agreement metric, and the coefficient of variation (CV) to quantify both software performance and dosiomic feature variability across them. RESULTS:The preliminary IBSI-compliance evaluation showed that MIRP, S-IBEX, RaCaT, and SERA achieved over 94% matching features with IBSI benchmark values. In contrast, SPAARC, RadiomiCRO, and PyRadiomics demonstrated lower compliance due to non-computable features. On dose distributions, all tools exhibited high match percentages (>77%) for the isotropic dataset, which did not require software-specific interpolation. However, discrepancies increased significantly with program-specific interpolation for the anisotropic dose dataset, with match rates dropping to 14%. Agreement across software was consistently high for the isotropic dataset but notably lower for the anisotropic dataset. This trend was less evident when looking at the CV, which showed only a mild increase for the anisotropic format. Fixed bin size (FBS) discretization displayed lower Agreement and higher CV values, particularly in the cumulative intensity-volume histogram (IVH) feature family. High CV values were predominantly observed for some feature family-ROI combinations, including GLRLM, GLSZM, and NGLDM computed using 2.5D/3D aggregation methods. Additionally, we observed that some binary masks were incorrectly generated (e.g., without holes) when using the DICOM format, therefore, we relied on NRRD input files whenever possible, resulting in feature reproducibility remaining unaffected by this aspect. CONCLUSION:The findings of this study indicate that, when properly configured, the tools show good overall agreement, with variability limited to specific features and pre-processing choices. While variations in program-specific resampling and FBS discretization implementation are present, their overall impact on dosiomic feature reproducibility remains minimal.

The combination of radiotherapy (RT) and immunotherapy has emerged as a central strategy in modern oncology, yet clinical outcomes remain highly variable. This variability is critically influenced by timing, including sequencing, interval, and temporal alignment with immune dynamics. We performed a systematic review to evaluate how mechanism-guided timing of RT relative to different immunotherapy classes shapes antitumor efficacy. A systematic literature search was conducted in PubMed, Embase, and Web of Science for studies published between 2009 and January 2026 evaluating RT–immunotherapy combinations with explicit timing or sequencing analysis. Preclinical and clinical studies reporting immune mechanisms, systemic immune responses, or efficacy outcomes were included. Data were extracted on immunotherapy class, RT regimen, sequencing strategy, and outcomes. Risk of bias was assessed using design-appropriate tools. A total of 84 studies met inclusion criteria: 56 preclinical and 28 clinical, including four randomized controlled trials. Timing emerged as a critical biological determinant of synergy. Radiotherapy induced a temporally dynamic immune response encompassing immunogenic cell death, innate immune activation, T‑cell priming, and adaptive immune resistance. Optimal sequencing was mechanism-dependent: Cytotoxic T‑lymphocyte-associated protein 4 (CTLA-4) blockade was most effective before or concomitantly with RT; programmed cell death protein 1 (PD-1)/programmed death-ligand 1 (PD-L1) inhibitors performed best after RT, during peak T‑cell infiltration and checkpoint upregulation; co-stimulatory agonists showed optimal activity shortly after RT; and innate immune activators required immediate post-RT administration. Emerging evidence suggests circadian timing may further modulate efficacy. Timing is not a logistical variable but a central biological component of RT–immunotherapy combinations. A mechanism-guided temporal framework substantially influences immune responses and therapeutic outcomes, representing a low-cost, high-impact optimization strategy. Circadian considerations remain compelling but require validation in prospective trials.

We evaluated the feasibility and clinical outcomes of MRIgSBRT for nodal recurrence in the oligometastatic disease (OMD) setting, focusing on per-lesion outcomes and prognostic factors. We collected clinical and dosimetric data from a retrospective single-center cohort of patients treated with a 0.35 T MRIgSBRT for nodal recurrences. Endpoints included the 1-year progression-free survival (PFS), local progression-free survival (LPFS), and 3-year overall survival (OS) rate from recurrence. Per-lesion Kaplan–Meier and Cox regression assessed clinical, dosimetric, and technical predictors. 71 patients received nodal MRIgSBRT, with a total of 115 treated metastatic lesions. Local control was high: LPFS 95.5

Upright positioning is re-emerging as a potential strategy to improve the pediatric radiotherapy experience, particularly for selected children in whom distress, limited cooperation, or repeated general anesthesia represent major barriers to treatment. This practice development report, produced by the Upright Radiotherapy Pediatric Task Group — an international, multiprofessional group bringing together radiation oncology, medical physics, radiation therapy, pediatric anesthesia and industry expertise — reviews the historical rationale, contemporary technological developments, and clinical requirements for evaluating upright pediatric radiotherapy. The considerations discussed apply to both photon and proton delivery, although the current economic and dosimetric drivers are strongest for gantry-less proton systems. The concept is supported by historical experience with seated treatment techniques, recent advances in upright imaging and gantry-less delivery systems, and broader pediatric evidence suggesting that seated positioning may improve procedural tolerability. However, pediatric-specific evidence remains limited, and upright treatment should not be regarded as a universal alternative to conventional supine workflows.Safe translation requires a deliberately cautious framework. Key considerations include age-appropriate immobilization, upright imaging and treatment-planning validation, both day-to-day reproducibility and within-fraction stability of setup, gravity-related anatomical changes, audiovisual distraction, and integration of child-centered preparation strategies. For sedated or anesthetized children, upright positioning introduces specific safety requirements related to airway visibility and access, hemodynamic monitoring, patient support, emergency release, rapid transition to a rescue position, and sufficient free space around the patient for the anesthesia team to intervene. These constraints should be treated as primary design and commissioning requirements rather than secondary workflow adaptations.Near-term research priorities include phantom and anatomical validation studies, assessment of imaging and dosimetric accuracy, emergency workflow testing, and prospective pilot studies evaluating setup reproducibility and within-fraction stability, treatment duration, anesthesia utilization, acute safety, and patient- and caregiver-reported experience. If developed within a rigorous multidisciplinary safety and evidence framework, upright pediatric radiotherapy may become a clinically valuable option for selected children, complementing rather than replacing established supine treatment approaches.

Magnetic resonance imaging (MRI)-guided radiotherapy (MRIgRT) integrates MRI with linear accelerators (MRI-linacs), enabling real-time motion management based on temporally resolved 2D MRI (cine-MRI). Current systems rely on template matching or deformable image registration for radiotherapy target (typically the gross tumor volume) localization, which allows beam gating. Further advances in localization could support more precise and efficient delivery methods. https://trackrad2025.grand-challenge.org/ was organized to provide a common dataset to benchmark algorithms for MRIgRT target tracking in 2D+t cine-MRI. Participants propagated target segmentation masks from an initialization frame across subsequent frames. The dataset comprised sagittal cine-MRI scans of 585 cancer patients undergoing radiotherapy at 0.35 T and 1.5 T MRI-linacs at six different institutions, with expert-annotated targets in 108 sequences. Target sites included the thorax (179 cases), abdomen (266 cases), and pelvis (140 cases). A total of 477 unlabeled and 50 labeled cases were provided for training purposes, 58 cases were kept private for preliminary testing (8) and final evaluation (50). The algorithms submitted by participants were executed on the challenge platform and assessed using metrics in three categories: geometric accuracy, surrogate dose accuracy and execution speed. Rankings were derived via a Rank-Then-Mean scheme. TrackRAD2025 attracted 148 registrations from 28 countries, 100 preliminary submissions and 24 final submissions from 14 teams. The top five methods achieved mean Dice similarity coefficients >0.87 and Euclidean center distances <2.1 mm, comparable to interobserver variability. Leading top five solutions featured foundation models with (4) or without (1) finetuning. Field strength had minimal effect on performance and tracking worked better for the pelvis with reduced motion amplitude compared to the thorax and abdomen cases, which achieved equivalent performance. TrackRAD2025 established a benchmark for MRIgRT tracking on multi-institutional cine-MRI data, highlighting foundation models as promising for clinical translation.

BACKGROUND:Magnetic Resonance Imaging (MRI)-only radiotherapy (RT) is increasingly adopted, but still lacks standardized commissioning procedures. This variability limits consistent clinical implementation: the Multicenter Evaluation of commercial Synthetic-Computed tomography ALgorithms (MESCAL) project aims to establish a benchmark dataset, provide commissioning guidelines, and define tolerance levels to support safe and reproducible MRI-only photon-beam RT adoption. METHODS:Data from 32 patients (16 brain, 16 pelvis) were retrospectively collected from two centers. Four sCT solutions licensed for clinical use were evaluated: MRI Planner (Spectronic), SyngoAI (Siemens), MR Box (Therapanacea), and MRCAT (Philips). For each patient, multiple sCTs were generated and compared with the planning CT. Image quality was assessed using mean absolute error (MAE) and three additional indicators. Dose accuracy was evaluated by recalculating treatment plans on sCTs and performing dose-volume histogram (DVH) analysis. Position-verification accuracy was quantified by comparing CBCT-to-sCT versus CBCT-to-CT registrations. Plans meeting acceptability criteria (DVH differences < 2%, gamma-passing rate 95% at 2%/2 mm) were used to derive tolerance intervals. RESULTS:Three to four sCTs were generated per case. Brain cases showed higher inter-software variability than pelvic cases, particularly in MAE within body (pelvis: 30-70 HU; brain: 40-130 HU). DVH differences remained within 3% (pelvis) and 4% (brain). Position-verification accuracy was higher in the brain (89% within 1 mm/1°) than in the pelvis (74%). Acceptability criteria were met by 37/43 pelvic and 30/39 brain plans: tolerance values were derived from these cases. CONCLUSION:MESCAL provides a commissioning framework, benchmark dataset, and tolerance levels to guide local sCT commissioning and promote standardized MRI-only photon-beam RT implementation.

Purpose/Objective Pancreatic ductal adenocarcinoma (PDAC) remains one of the most lethal malignancies, with limited treatment options. Magnetic Resonance-guided Radiotherapy (MRIgRT) combines real-time imaging and irradiation under a static magnetic field (SMF), potentially influencing biological responses beyond its technical advantages. This study aimed to investigate the combined effects of SMF and radiotherapy on patient-derived PDAC organoids (PDOs) to assess whether MRI guidance modulates radiation-induced cytotoxicity. Material/Methods Two PDO lines (PDAC-9 and PDAC-12) were established from treatment-naïve PDAC tissues and cultured in 3D BME matrices. Organoids were exposed to three experimental conditions: (i) untreated control (CTRL), (ii) 10 Gy irradiation using a standard linac (IR), and (iii) 10 Gy irradiation on a 0.35 T MR-Linac with real-time cine-MRI acquisition (IRrtMRI). Organoid viability was assessed by CellTiter-Glo luminescence assay. Morphometric analysis was performed using Bright-field images classifying organoids as small, medium, or large. Cell death was quantified by 7-AAD flow cytometry, while DNA damage and apoptosis were evaluated via western blotting for γH2AX and cleaved-PARP1. Statistical analysis was performed by ANOVA with multiple comparisons (GraphPad Prism 10). Results Both PDO lines exhibited reduced growth and viability after irradiation, with a significantly stronger effect under combined IRrtMRI exposure. The proportion of large organoids decreased to 0%, while small organoids markedly increased in both PDAC-9 and PDAC-12 (p < 0.01). Cell viability assays confirmed enhanced inhibition following IRrtMRI compared to IR alone (p < 0.001). Flow cytometry revealed that apoptotic cells increased from 40% (IR) to 70% (IRrtMRI) in PDAC-12, whereas PDAC-9 showed relative resistance. Western blot analysis demonstrated increased γH2AX phosphorylation in both PDOs and cleaved-PARP1 induction selectively in PDAC-12 under combined treatment, indicating augmented DNA damage and apoptosis. Conclusion Exposure to radiotherapy under a static 0.35 T magnetic field with real-time MRI enhances genotoxic and apoptotic effects in PDAC organoids compared to conventional irradiation. The observed radiosensitization may derive from magnetic field–mediated prolongation of radical lifetimes, absence of CBCT X-ray–induced adaptive responses, and DNA polarization phenomena. These findings suggest a possible biological advantage of MRIgRT beyond improved targeting accuracy, supporting further investigation into its translational and clinical implications in pancreatic cancer.

PURPOSE:To characterise the current clinical adoption of AI-driven automation and adaptive radiotherapy (ART) in Italy. MATERIALS AND METHODS:A structured, modular questionnaire was distributed to 133 Italian radiotherapy and medical physics departments. The survey captured: centre characteristics, clinical use of automation tools (synthetic-imaging, autosegmentation, autoplanning, motion management, and QA automation), ART implementation (offline/online approaches, triggers, dose summation, and patient-specific QA) and perceived benefits and barriers. Responses were analysed including the maturity index (MI), defined as a weighted sum of 14 binary implementation items spanning key workflow domains. RESULTS:Sixty-one centres responded (46.0%). Routine autosegmentation was reported by 45/61 centres (73.8%) and autoplanning by 27/61 (44.2%). Only 5 centres reported having performed a dedicated FMEA for automation-related processes; heterogeneous QA approaches were reported by 21/61 (34%). A broad definition of ART (including offline re-planning prompted by interfractional changes) was reported by 56/61 centres (91.8%); however, online inter-fraction and intra-fraction adaptation remained limited and largely confined to centres with dedicated ART-ready platforms. Mean (median) MI was 4.99 (5.00) on a theoretical maximum of 9.334, with a weak positive correlation with the number of installed LINACs (r = 0.25, p = 0.049). CONCLUSIONS:Among responding centres, automation tools are widely available and ART is frequently practised, predominantly through offline approaches. The results suggest that broad penetration of adaptive practices does not yet correspond to homogeneous end-to-end maturity. Standardisation of QA, clearer operational definitions of ART triggers, formal risk analysis, and validated dose accumulation workflows appear as key priorities to support broader and safer scaling of online adaptive strategies.

Automated medical image segmentation suffers from high inter-observer variability, particularly in tasks such as lung nodule delineation, where experts often disagree. Existing approaches either collapse this variability into a consensus mask or rely on separate model branches for each annotator. We introduce ProSona, a two-stage framework that learns a continuous latent space of annotation styles, enabling controllable personalization via natural language prompts. A probabilistic U-Net backbone captures diverse expert hypotheses, while a prompt-guided projection mechanism navigates this latent space to generate personalized segmentations. A multi-level contrastive objective aligns textual and visual representations, promoting disentangled and interpretable expert styles. Across the LIDC-IDRI lung nodule and multi-institutional prostate MRI datasets, ProSona reduces the Generalized Energy Distance by 17

Background and Purpose Validation of deformable image registration (DIR) remains predominantly contourbased; this study evaluated inverse consistency error (ICE) as an automated voxelwise metric for DIR accuracy. Materials and Methods Synthetic ground-truth DVFs were generated using geometric and head-and-neck (HN) digital phantoms undergoing controlled global and local deformations. DIR was performed with the ANACONDA algorithm in RayStation. ICE maps derived from clinical DVFs were compared with ground-truth registration error (GTRE), target registration error (TRE) from 20 anatomical landmarks, and mean distance to agreement (MDA) for 22 propagated ROIs. Results Ground-truth DVFs showed negligible ICE values, confirming mathematical invertibility. In HN phantoms, median ICE and GTRE were 0.8 ± 0.2 mm and 1.6 ± 0.4 mm, respectively. ICE correlated strongly with GTRE (R = 0.85, p < 0.001) and moderately with TRE (R = 0.68, p < 0.001). No significant correlation was found with contourbased MDA (2.47 ± 0.18 mm). Voxel-wise analysis showed that ICE captured spatial patterns of uncertainty consistent with regions of higher GTRE, while underestimating error for global homogeneous deformations >15 mm due to DIR regularisation. Across all datasets, ICE correctly identified high-uncertainty subregions that were not detected by contour-based metrics. Conclusions ICE enables automated voxel-wise quantification of DIR uncertainty directly from clinical DVFs. It complements traditional contour-based metrics and may support patient-specific QA and more reliable dose mapping in adaptive and re-irradiation radiotherapy workflows.

PURPOSE:To systematically investigate the behavior of plan complexity metrics (PCMs) in an MR-Linac online adaptive radiotherapy (oART) workflow for pancreatic cancer, and to evaluate their potential as surrogate indicators of delivery accuracy. METHODS:Thirty-seven patients with locally advanced pancreatic cancer were retrospectively analyzed, yielding 222 MR-Linac plans (37 reference and 185 delivered fractions). Fifteen PCMs were extracted from plans generated with three optimizers: Penalty, Objectives and Constraints, and A3i (current clinical practice). Plan specific quality assurance (PSQA) has been performed through an independent dose calculation algorithm. Statistical analyses included: (i) inter-optimizer comparisons (ANOVA and mixed-effects models), (ii) variance decomposition of adapted-plan complexity metrics using linear mixed-effects models (LMEMs), and (iii) evaluation of PSQA stability using statistical process control (SPC) and leave-one-patient-out (LOPO) cross-validation. RESULTS:Optimizer choice strongly influenced plan complexity. The Penalty optimizer generated higher-complexity plans, whereas Objectives and Constraints and A3i produced more modulation-efficient configurations with fewer small, low-MU segments. Variance decomposition identified a subset of metrics that exhibited consistent behavior across all optimizers, serving as robust descriptors independent of the algorithm. Metrics dominated by between-patient variance (σ2 between) emerged as reliable surrogates for patient-specific complexity Tongue & Groove Index, Average Leaf Gap and Number of Active Leaves consistently showed high between-patient contributions (σ2 between > 74%) among others. In contrast, metrics related to low-MU segments (SegMU < 5 and SegMU < 5 [%]) were dominated by within-patient variance (σ2 within), with A3i showing the most pronounced fluctuations (85.8% and 84.8%, respectively). These descriptors are therefore more sensitive to plan-specific or optimizer-related stochasticity than to stable patient factors. SPC analyses demonstrated that the current adaptive workflow is robustly stable for most patients: 11 of 13 never experienced a fraction below the tolerance level (TL), and 12 of 13 never exceeded the action level (AL), even when thresholds were dynamically recalculated within the LOPO-CV. CONCLUSION:This study provides the first systematic assessment of PCMs in MR-guided oART, demonstrating optimizer-specific complexity signatures, predominant inter-patient variability, and the predictive value of selected metrics for delivery accuracy. Although limited to a single tumor site and workflow, the methodology supports the development of institution-specific, complexity-aware scorecards to enhance adaptive planning and quality assurance.

Non-small cell lung cancer is the most common malignancy of the lung, with over 40

Background and purpose:Magnetic Resonance Imaging-only (MRI-only) workflows are an emerging strategy in radiotherapy, with artificial intelligence (AI) playing a central role in generating synthetic computed tomography (sCT) images. The thorax remains a particularly difficult region due to marked electron density (ED) heterogeneity and respiratory motion. This study investigates the impact of key factors on AI-based thoracic sCT generation. Materials and methods:A total of 122 thoracic patients treated with MRI-guided radiotherapy (MRIgRT) were retrospectively included. Both 0.35 Tesla (T) MR and CT simulation images were acquired under consistent breath-hold conditions. Three aspects were analyzed: (i) training set size (34, 68, and 102 cases), (ii) pre-processing of MR images (filtered versus raw), and (iii) generator architecture, comparing U-Net and ResNet with a novel model integrating Fourier space information, the Adaptive Fourier Neural Operator (AFNO). Models were tested on 20 independent patients using image similarity metrics. The best configuration was also evaluated through dose recalculations. Results:Expanding the training set improved accuracy, reducing Mean Absolute Error (MAE) from 42.0 ± 9 Hounsfield Units (HU) to 35.9 ± 6 HU. Pre-processing had limited effect, while generator architecture had a strong impact, with AFNO outperforming others (MAE = 32.4 ± 6 HU). The optimal setup, AFNO trained on raw MR images from 102 patients, yielded dosimetric deviations below 3 % for target dose-volume metrics and within 50 cGy for organs at risk (OARs). Conclusions:These findings highlight the importance of training dataset size and advanced network architectures for thoracic sCT generation. AFNO demonstrated superior performance, reinforcing the feasibility of MRI-only workflows in thoracic radiotherapy.

Background: Knowledge-based (KB) planning is a promising approach to model prior planning experience and optimize radiotherapy. To enable the sharing of models across institutions, their transferability must be evaluated. This study aimed to validate KB prediction models developed by a national consortium using data from another multi-institutional consortium in a different country. Methods: Ten right whole breast tangential field (RWB-TF) models were built within the national consortium. A cohort of 20 patients from the external consortium was used for testing. Transferability was defined when the ipsilateral (IPSI) lung first principal component (PC1) was within the 10th–90th percentile of the training set. Predicted dose–volume parameters were compared with clinical dose–volume histograms (cDVHs). Results: Planning target volume (PTV) coverage strategies were comparable between the two consortia, even though significant volume differences were observed for the PTV and contralateral breast (p = 0.002 and p = 0.02, respectively). For the IPSI lung, the standard deviation of predicted mean dose/V20 Gy was 1.13 Gy/2.9% in the external consortium versus 0.55 Gy/1.6% in the training consortium. Differences between the cDVH and the predicted IPSI lung mean dose and the volume receiving more than 20 Gy (V20 Gy) were <2 Gy and <5% in 88.7% and 92.3% of cases, respectively. PC1 values fell within the 10th–90th percentile for ≥90% of patients in 6/10 models and 65–85% for the remaining 4. Conclusions: This study demonstrates the feasibility of applying RWB-TF KB models beyond the consortium in which they were developed, supporting broader clinical implementation. This retrospective study was supported by AIRC (Associazione Italiana per la Ricerca sul Cancro) and registered on ClinicalTrials.gov (NCT06317948, 12 March 2024).

Background and purpose:In order to optimize the radiotherapy treatment and minimize toxicities, organs-at-risk (OARs) and clinical target volume (CTV) must be segmented. Deep Learning (DL) techniques show significant potential for performing this task effectively. The availability of a large single-institute data sample, combined with additional numerous multi-centric data, makes it possible to develop and validate a reliable CTV segmentation model. Materials and methods:Planning CT data of 1822 patients were available (861 from a single center for training and 961 from 8 centers for validation). A preprocessing step, aimed at standardizing all the images, followed by a 3D-Unet capable of segmenting both right and left CTVs was implemented. The metrics used to evaluate the performance were the Dice similarity coefficient (DSC), the Hausdorff distance (HD), and its 95th percentile variant (HD_95) and the Average Surface Distance (ASD). Results:The segmentation model achieved high performance on the validation set (DSC: 0.90; HD: 20.5 mm; HD_95: 10.0 mm; ASD: 2.1 mm; epoch 298). Furthermore, the model predicted smoother contours than the clinical ones along the cranial-caudal axis in both directions. When applied to internal and external data the same metrics demonstrated an overall agreement and model transferability for all but one (Inst 9) center. Conclusion:. A 3D-Unet for CTV segmentation trained on a large single institute cohort consisting of planning CTs and manual segmentations was built and externally validated, reaching high performance.

PURPOSE:Magnetic resonance imaging (MRI) to visualize anatomical motion is becoming increasingly important when treating cancer patients with radiotherapy. Hybrid MRI-linear accelerator (MRI-linac) systems allow real-time motion management during irradiation. This paper presents a multi-institutional real-time MRI time series dataset from different MRI-linac vendors. The dataset is designed to support developing and evaluating real-time tumor localization (tracking) algorithms for MRI-guided radiotherapy within the TrackRAD2025 challenge ( https://trackrad2025.grand-challenge.org/). ACQUISITION AND VALIDATION METHODS:The dataset consists of sagittal 2D cine MRIs (20-20543 frames per scan) in 585 patients from six centers (3 Dutch, 1 German, 1 Australian, and 1 Chinese). Tumors in the thorax, abdomen, and pelvis acquired on two commercially available MRI-linacs (0.35 T and 1.5 T) were included. For 108 cases, irradiation targets or tracking surrogates were manually segmented on each temporal frame. The dataset was randomly split into a public training set of 527 cases (477 unlabeled and 50 labeled) and a private testing set of 58 cases (all labeled). DATA FORMAT AND USAGE NOTES:The data is publicly available under the TrackRAD2025 collection: https://doi.org/10.57967/hf/4539. Both the images and segmentations for each patient are available in metadata format. POTENTIAL APPLICATIONS:This novel clinical dataset will enable the development and evaluation of real-time tumor localization algorithms for MRI-guided radiotherapy. By enabling more accurate motion management and adaptive treatment strategies, this dataset has the potential to advance the field of radiotherapy significantly.