We introduce a novel learning framework for accelerated Monte Carlo (MC) dose calculation termed Energy-Shifting. This approach leverages deep learning to synthesize highly complex polyenergetic dose distributions directly from simple monoenergetic inputs under identical beam configurations. Unlike conventional denoising techniques, which rely on noisy low-count dose maps that compromise beam profile integrity, our method achieves superior cross-domain generalization on unseen datasets by integrating high-fidelity anatomical textures and source-specific beam similarity into the model's input space. Furthermore, we propose a novel 3D architecture termed TransUNetSE3D, featuring Transformer blocks for global context and Residual Squeeze-and-Excitation (SE) modules for adaptive channel-wise feature recalibration. Hierarchical representations of these blocks are fused into the network's latent space alongside the primary dose-map parameters, allowing physics-aware reconstruction. This hybrid design outperforms existing UNet and Transformer-based benchmarks in both spatial precision and structural preservation, while maintaining the execution speed necessary for real-time use. Our proposed pipeline achieves a Gamma Passing Rate exceeding 98

Objective.Accurate and personalized radiation dose estimation is crucial for effective targeted radionuclide therapy (TRT). Deep learning (DL) holds promise for this purpose. However, current DL-based dosimetry methods require large-scale supervised data, which is scarce in clinical practice.Approach.To address this challenge, we propose exploring semi-supervised learning (SSL) framework that leverages readily available pre-therapy positron emission tomography (PET) data, where only a small subset requires dose labels, to predict radiation doses, thereby reducing the dependency on extensive labeled datasets. In this study, traditional classification-based SSL approaches were adapted and extended in regression task specifically designed for dose prediction. To facilitate comprehensive testing and validation, we developed a synthetic dataset that simulates PET images and dose calculation using Monte Carlo simulations.Main results.In the experiment, several regression-adapted SSL methods were compared and evaluated under varying proportions of labeled data in the training set. The overall mean absolute percentage error of dose prediction remained between 9% and 11% across different organs, which achieved comparable performance than fully supervised ones.Significance.The preliminary experimental results demonstrated that the proposed SSL methods yield promising outcomes for organ-level dose prediction, particularly in scenarios where clinical data are not available in sufficient quantities.

Over the past years, we have developed GATE version 10, a major re-implementation of the long-standing Geant4-based Monte Carlo application for particle and radiation transport simulation in medical physics. This release introduces many new features and significant improvements, most notably a Python-based user interface replacing the legacy static input files. The new functionality of GATE version 10 is described in the part 1 companion paper (Sarrut et al 2025 arXiv:2507.09842). The development brought significant challenges. In this paper, we present the solutions that we have developed to overcome these challenges. In particular, we present a modular design that robustly manages the core components of a simulation: particle sources, geometry, physics processes, and data acquisition. The architecture consists of integrated C++ and Python codes. This framework allows for the precise, time-aware generation of primary particles, a critical requirement for accurately modeling positron emission tomography, radionuclide therapies, or prompt-gamma timing systems. We present how GATE 10 handles complex Geant4 physics settings while exposing a simple interface to the user. Furthermore, we describe the methodological solutions that facilitate the seamless integration of advanced physics models and variance reduction techniques. The architecture supports sophisticated scoring of physical quantities (such as Linear Energy Transfer and Relative Biological Effectiveness) and is designed for multithreaded execution. The new user interface allows researchers to script complex simulation workflows and directly couple external tools, such as artificial intelligence models for source generation or detector response. By detailing these architectural innovations, we demonstrate how GATE 10 provides a more powerful and flexible tool for research and innovation in medical physics. This paper is not intended to be a developer guide. Its purpose is to share with the research community in-depth explanations of our development effort that made the new GATE 10 possible.

Interventional procedures expose physicians to scattered radiation, particularly to their upper extremities, posing occupational health risks. Existing extremity dosimeters such as thermoluminescents, optically stimulated luminescence rings and active personal dosimeters provide limited spatial information, exhibit angular and energy dependence and offer little or no real-time feedback. This study develops machine learning (ML) models to estimate radiation dose values at discrete upper-limb locations using Monte Carlo (MC)-derived data and procedure-specific parameters. A dataset of 10 000 MC dose maps was generated under varied clinical and geometric conditions. After log-transformation and normalisation, several ML models, including deep neural networks and tree-based regressor, were trained and assessed using five-fold cross-validation. Mean absolute error and relative error (RE) were evaluated on the original dose scale, and an ensemble of the three best-performing models was constructed to improve robustness. The ensemble consistently outperformed individual models, achieving an average RE of 3.69% and demonstrating stable performance across anatomical regions and dose levels. Highest accuracy was obtained for standard beam geometries, whereas larger discrepancies occurred in extreme configurations with steep dose gradients. Predicted dose patterns were consistent with the expected distributions across the upper-limb regions. The findings demonstrate the feasibility of ML-based extremity dose estimation in interventional environments. The proposed ensemble provides a rapid (around 10 ms), scalable alternative to full MC simulations, enabling near real-time predictions of upper-limb occupational dose and supporting optimisation of radiation protection practices in image-guided procedures.

We present GATE version 10, a major evolution of the open-source Monte Carlo simulation application for medical physics, built on Geant4. This release marks a transformative evolution, featuring a modern Python-based user interface, enhanced multithreading and multiprocessing capabilities, the ability to be embedded as a library within other software, and a streamlined framework for collaborative development. In this Part 1 paper, we outline GATE's position among other Monte Carlo codes, the core principles driving this evolution, and the robust development cycle employed. We also detail the new features and improvements. Part 2 will focus on the architectural innovations and technical challenges. By combining an open, collaborative framework with cutting-edge features, such a Monte Carlo platform supports a wide range of academic and industrial research, solidifying its role as a critical tool for innovation in medical physics.

Occupational radiation exposure in interventional radiology is spatially heterogeneous and inadequately captured by conventional point-based dosimetry. This study proposes a physics-informed deep learning framework for fast prediction of three-dimensional (3D) physician dose distributions from scattered radiation. Graphics processing unit (GPU)-accelerated Monte Carlo (MC) simulations were performed using the GPU Geant4-based Monte Carlo Simulation platform to generate 3D dose maps under varying X-ray energies, C-arm angulations, and physician configurations. These data were used to train residual and transformer-based 3D U-Net architectures. Model performance was evaluated using voxel-wise error metrics, gamma analysis, and clinically relevant personal dose equivalents. The residual 3D U-Net achieved the best performance, with mean absolute errors below 0.06 nGy and gamma passing rates exceeding 90

Objective.To develop a machine learning-based framework for accurately modeling the anode heel effect in Monte Carlo(MC) simulations of x-ray imaging systems, enabling realistic beam intensity profiles with minimal experimental calibration.Approach.Multiple regression models were trained to predict spatial intensity variations along the anode-cathode axis using experimentally acquired weights derived from beam measurements across different tube potentials. These weights captured the asymmetry introduced by the anode heel effect. A systematic fine-tuning protocol was established to minimize the number of required measurements while preserving model accuracy. The models were implemented in the OpenGATE 10 and GGEMS MC toolkits to evaluate their integration feasibility and predictive performance.Main results.Among the tested models, gradient boosting regression (GBR) delivered the highest accuracy, with prediction errors remaining below 5% across all energy levels. The optimized fine-tuning strategy required only six detector positions per energy level, reducing measurement effort by 65%. The maximum error introduced through this fine-tuning process remained below 2%. Dose actor comparisons within MC simulations demonstrated that the GBR-based model closely replicated clinical beam profiles and significantly outperformed conventional symmetric beam models.Significance.This study presents a robust and generalizable method for incorporating the anode heel effect into MC simulations using machine learning. By enabling accurate, energy-dependent beam modeling with limited calibration data, the approach enhances simulation realism for applications in clinical dosimetry, image quality assessment, and radiation protection.

Targeted Radionuclide Therapy (TRT) is a modern strategy in radiation oncology that aims to administer a potent radiation dose specifically to cancer cells using cancer-targeting radiopharmaceuticals. Accurate radiation dose estimation tailored to individual patients is crucial. Deep learning, particularly with pre-therapy imaging, holds promise for personalizing TRT doses. However, current methods require large time series of SPECT imaging, which is hardly achievable in routine clinical practice, and thus raises issues of data availability. Our objective is to develop a semi-supervised learning (SSL) solution to personalize dosimetry using pre-therapy images. The aim is to develop an approach that achieves accurate results when PET/CT images are available, but are associated with only a few post-therapy dosimetry data provided by SPECT images. In this work, we introduce an SSL method using a pseudo-label generation approach for regression tasks inspired by the FixMatch framework. The feasibility of the proposed solution was preliminarily evaluated through an in-silico study using synthetic data and Monte Carlo simulation. Experimental results for organ dose prediction yielded promising outcomes, showing that the use of pseudo-labeled data provides better accuracy compared to using only labeled data.

Objectives : Data scarcity and domain shifts lead to biased training sets that do not accurately represent deployment conditions. A related practical problem is cross-modal image segmentation, where the objective is to segment unlabelled images using previously labelled datasets from other imaging modalities. Methods : We propose a cross-modal segmentation method based on conventional image synthesis boosted by a new data augmentation technique called Generative Blending Augmentation (GBA). GBA leverages a SinGAN model to learn representative generative features from a single training image to diversify realistically tumor appearances. This way, we compensate for image synthesis errors, subsequently improving the generalization power of a downstream segmentation model. The proposed augmentation is further combined to an iterative self-training procedure leveraging pseudo labels at each pass. Results : The proposed solution ranked first for vestibular schwannoma (VS) segmentation during the validation and test phases of the MICCAI CrossMoDA 2022 challenge, with best mean Dice similarity and average symmetric surface distance measures. Conclusion and significance : Local contrast alteration of tumor appearances and iterative self-training with pseudo labels are likely to lead to performance improvements in a variety of segmentation contexts.

Over the past years, we have developed GATE version 10, a major re-implementation of the long-standing Geant4-based Monte Carlo application for particle and radiation transport simulation in medical physics. This release introduces many new features and significant improvements, most notably a Python-based user interface replacing the legacy static input files. The new functionality of GATE version 10 is described in the part 1 companion paper. The development brought significant challenges. In this paper, we present the solutions that we have developed to overcome these challenges. In particular, we present a modular design that robustly manages the core components of a simulation: particle sources, geometry, physics processes, and data acquisition. The architecture consists of parts written in C++ and Python, which needed to be coupled. We explain how this framework allows for the precise, time-aware generation of primary particles, a critical requirement for accurately modeling positron emission tomography (PET), radionuclide therapies, and prompt-gamma timing systems. We present how GATE 10 handles complex Geant4 physics settings while exposing a simple interface to the user. Furthermore, we describe the technical solutions that facilitate the seamless integration of advanced physics models and variance reduction techniques. The architecture supports sophisticated scoring of physical quantities (such as Linear Energy Transfer and Relative Biological Effectiveness) and is designed for multithreaded execution. The new user interface allows researchers to script complex simulation workflows and directly couple external tools, such as artificial intelligence models for source generation or detector response. By detailing these architectural innovations, we demonstrate how GATE 10 provides a more powerful and flexible tool for research and innovation in medical physics.

Response Evaluation Criteria in Solid Tumors (RECIST) remains the cornerstone for treatment response assessment in oncology. However, its traditional 2D approach faces challenges including measurement variability and time-intensive workflow. This study presents a comparative analysis between conventional human-performed RECIST1.1 measurements and a volumetric approach driven by artificial in-telligence (AI). Through the analysis of 23 patients with colorectal liver metastases (CRLM) treated with chemother-apy, we demonstrate that AI-based volumetric assessment achieves enhanced tumor response assessment while potentially offering improved reproducibility and efficiency. This work provides evidence supporting the clinical validity of AI-based volumetric tumor assessment as a complementary or alternative approach to the traditional RECIST evaluation.

This article presents a fluoroscopy image-based registration method along with a comprehensive protocol for robotic needle insertion in radiofrequency ablation (RFA) to treat liver cancer. The proposed method uses real-time fluoroscopic images acquired from a C-ARM system and integrates an inverse finite element (FE) simulation to compute robotic commands for accurate and adaptive needle steering. The registration procedure is fully automated and involves the injection of multiple radiopaque markers into the liver, enabling precise anatomical registration and targeted tumor localization. A key challenge addressed in this work is the integration of this image-based registration with the inverse biomechanical simulation used to guide the robot during insertion. We describe how registration constraints can be mapped onto the surface of the biomechanical model to ensure consistent alignment between image data and robotic actuation. Designed to be adaptable to varying levels of radiologist expertise and applicable across a wide range of tumor locations, this method provides a robust and versatile solution for improving the accuracy and safety of minimally invasive liver cancer treatments.

Protecting medical personnel from the harmful effects of scattered ionising radiation during x-ray-guided procedures is a critical concern. Due to the complex and invisible nature of x-rays, monitoring radiation exposure has been challenging. Existing real-time dosimeters have shown low accuracy and practical limitations. To address these challenges, this study introduces an innovative approach that combines Monte Carlo (MC) simulations and deep learning (DL) for real-time estimation of three-dimensional (3D) scattered radiation in the operating room. The neural network was trained to map patient morphology and imaging parameters to radiation maps, allowing it to adapt to various clinical scenarios. The results demonstrate that the system showcases exceptional speed by efficiently computing 3D radiation maps in 11 ms using modern GPU (NVIDIA RTX 2080). Validation experiments confirmed the reliability of the predicted scatter maps, with a mean absolute percentage error of 10.97% relative to MC simulations. When used to compute organ doses via voxelised-source simulations, the global average organ dose error was 8.2 ± 4.1%. Therefore, the combination of MC simulations and DL provides a promising solution for enhancing the safety of medical personnel during x-ray-guided procedures.

Prostate cancer (PCa) is one of the most common cancers among men, and artificial intelligence (AI) is emerging as a promising tool to enhance its diagnosis. This work proposes a classification approach for PCa cases using deep learning techniques. We conducted a comparison between unimodal models based either on biparametric magnetic resonance imaging (bpMRI) or clinical data (such as prostate-specific antigen levels, prostate volume, and age). We also introduced a bimodal model that simultaneously integrates imaging and clinical data to address the limitations of unimodal approaches. Furthermore, we propose a framework that not only detects the presence of PCa but also evaluates the uncertainty associated with the predictions. This approach makes it possible to identify highly confident predictions and distinguish them from those characterized by uncertainty, thereby enhancing the reliability and applicability of automated medical decisions in clinical practice. The results show that the bimodal model significantly improves performance, with an area under the curve (AUC) reaching 0.82±0.03, a sensitivity of 0.73±0.04, while maintaining high specificity. Uncertainty analysis revealed that the bimodal model produces more confident predictions, with an uncertainty accuracy of 0.85, surpassing the imaging-only model (which is 0.71). This increase in reliability is crucial in a clinical context, where precise and dependable diagnostic decisions are essential for patient care. The integration of clinical data with imaging data in a bimodal model not only improves diagnostic performance but also strengthens the reliability of predictions, making this approach particularly suitable for clinical use.

This work presents PI-RADS-Net, a deep-learning model inspired by the PI-RADS protocol to detect clinically significant prostate cancers (csPCa) on MRI by emulating radiologists' decision-making. The model employs a two-branch architecture, using T2-weighted images for the transition zone and a combination of diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) sequences for the peripheral zone. An initial extension, PI-RADS-BimodalNet, which incorporates clinical parameters, was also developed. Experimental results demonstrate that PI-RADS-Net significantly improves classification accuracy, while PI-RADS-BimodalNet achieves an area under the curve (AUC) of 0.83 in testing, surpassing other models. Our approach holds promise for improving the diagnosis of clinically significant prostate cancer, reducing unnecessary biopsies, and enhancing patient outcomes.

Accurate segmentation of colorectal liver metastases (CRLM) from CT scans is essential for diagnosis and treatment planning in oncology. However, accurate lesion delineation remains a significant challenge due notably to instance imbalance. In this work, we propose an instance-aware segmentation model to enhance CRLM extraction, exploiting an instance-level loss function. Through careful evaluation, we analyze performance across diverse lesion morphologies and sizes, an aspect often overlooked for model validation. By leveraging a SwinUNETR-based architecture and three distinct datasets, we demonstrate improved instance detection ability, thereby contributing to better clinical decision-making in managing CRLM. A publicly-available implementation is provided: https://github.com/marwanabb/CRLM-instance/.

Accurate detection and classification of clinically significant prostate cancer remain critical challenges in medical imaging. Despite numerous studies focusing on feature extraction and classification, none have systematically assessed the impact of harmonization techniques on multicenter imaging data. This study aimed to improve diagnostic performance by integrating harmonization via unsupervised clustering with clinical variables into machine learning models, even when the source center is unknown. We extracted features from T2-weighted magnetic resonance images using two approaches: handcrafted radiomics and deep learning-based representations obtained via a 3D convolutional autoencoder architecture. To address inter-center variability, the data were harmonized using an unsupervised clustering approach that generated 19 distinct clusters, followed by ComBat harmonization. Machine learning classifiers were then trained with and without the inclusion of clinical variables (prostate-specific antigen levels and patient age). The models were evaluated using multiple metrics. Harmonization significantly improved classification performance. In particular, models based on 3D convolutional autoencoder-derived features achieved an accuracy of 75.33% and an area under the curve (AUC) of 0.74. The incorporation of clinical variables further enhanced model performance; the best model combining radiomics features with clinical data attained an accuracy of 77.67% and an AUC of 0.85. These performance metrics are concurrent with those reported in the literature for this challenge, demonstrating that our novel approach can effectively mitigate inter-center variability and enhance diagnostic accuracy. Our findings underscore the potential of harmonization via unsupervised clustering, combined with the inclusion of clinical variables, to significantly enhance the diagnostic performance of machine learning models for prostate cancer detection. This novel strategy not only addresses a gap in the current literature but also produces performance metrics comparable to those reported in multicenter studies, thereby supporting the development of robust, clinically applicable diagnostic tools.

Despite the growing prominence of transformers in medical image segmentation, their application to clinically significant prostate cancer (csPCa) has been overlooked. Minimal attention has been paid to domain shift analysis and uncertainty assessment, critical for safely implementing computer-aided diagnosis (CAD) systems. Domain shift in medical imagery refers to differences between the data used to train a model and the data evaluated later, arising from variations in imaging equipment, protocols, patient populations, and acquisition noise. While recent models enhance in-domain performance, areas such as robustness and uncertainty estimation in out-of-domain distributions have received limited investigation, creating indecisiveness about model reliability. In contrast, our study addresses csPCa at voxel, lesion, and image levels, investigating models from traditional U-Net to cutting-edge transformers. We focus on four key points: robustness, calibration, out-of-distribution (OOD), and misclassification detection (MD). Findings show that transformer-based models exhibit enhanced robustness at image and lesion levels, both in and out of domain. However, this improvement is not fully translated to the voxel level, where Convolutional Neural Networks (CNNs) outperform in most robustness metrics. Regarding uncertainty, hybrid transformers and transformer encoders performed better, but this trend depends on misclassification or out-of-distribution tasks.

Recent advancements in medical image segmentation have been driven by deep learning's capability to extract rich features from extensive datasets. However, these improvements rely heavily on large annotated datasets, which pose significant challenges in the resource-intensive medical field. Foundational models, such as Meta's Segment Anything Model (SAM), have been developed to address these challenges. SAM has demonstrated exceptional zero-shot performance, often rivaling or surpassing fully supervised models across various tasks. Nonetheless, SAM cannot be directly applied to medical image segmentation due to domain shift, making it necessary to fine-tune the model using prompts. Reducing the annotation workload is crucial to alleviate the burden and constraints associated with extensive data annotation in the medical field. This study investigates prompt-guided strategies in SAM for medical image segmentation under few-shot and weakly supervised scenarios. We assess various strategies-bounding boxes, positive points, negative points, and their combinations-using two publicly available datasets. Optimal results are achieved using positive-negative points, demonstrating that the SAM model can perform comparably to established methods in hepatic vascular and prostate cancer segmentation, even with minimal examples. This research aims to advance medical image segmentation by decreasing reliance on extensive annotated data, providing insights into effective prompt utilization, and showcasing SAM's adaptability in specialized medical contexts.

Background: Endovascular treatment is continuously gaining ground in vascular surgery procedures. However, current patient radiation dose estimation does not take into account the exact patient morphology and organs' composition. Monte Carlo (MC) simulation can accurately estimate the dose by recreating the irradiation process generated during X -ray -guided interventions. This study aimed to validate the MC simulation models by comparing simulated and measured dose distributions in endovascular aortic aneurysm repair (EVAR) procedures. Methods: We conducted a clinical study in patients treated for EVAR. Patient dose measurements were taken with passive dosimeters using Optically Stimulated Luminescence technology in 4 specific anatomical points on the skin: xiphoid process, pubic symphysis, right and left iliac crest. Dose measurements were compared to the corresponding simulated doses with the Geant4 Application for Emission Tomography (GATE) and GPU Geant4-based Monte Carlo Simulations (GGEMS) MC simulations softwares. The MC simulation took as input the computed tomography scan of the patient and the parameters of the imaging system (orientation angles, tube voltage, and aluminum filtration) and gives as output the three-dimensional (3D) dose map for each patient and angulation. Results: A good agreement with real doses was found for doses simulated by the MC GATE method (P < 0.0001; r = 0.97; 95% confidence interval [CI] [0.96-0.98]), as well as for doses simulated by the GGEMS method (P < 0.0001; r = 0.96; 95% CI [0.94-0.97]). The mean relative error for all measurements was 5 +/- 5% in the MC GATE group and 6 +/- 5% in the GGEMS group. Process execution on GGEMS (6 sec) was faster than the GATE MC simulation (5 hr). Conclusion: Considering the current imaging settings, this study shows the potential of using the GATE and GGEMS MC simulations platforms to model the 3D dose distributions during EVAR procedures.