Abstract:

Federated Learning (FL) holds the potential to advance equality in health by enabling diverse institutions to collaboratively train deep learning (DL) models, even with limited data. However, the significant resource requirements of FL often exclude centres with limited computational infrastructure, further widening existing healthcare disparities. To address this issue, we propose a Green AI-oriented adaptive layer-freezing strategy designed to reduce energy consumption and computational load while maintaining model performance. We tested our approach using different federated architectures for Magnetic Resonance Imaging (MRI)-to-Computed Tomography (CT) conversion. The proposed adaptive strategy optimises the federated training by selectively freezing the encoder weights based on the monitored relative difference of the encoder weights from round to round. A patience-based mechanism ensures that freezing only occurs when updates remain consistently minimal. The energy consumption and CO2eq emissions of the federation were tracked using the CodeCarbon library. Compared to equivalent non-frozen counterparts, our approach reduced training time, total energy consumption and CO2eq emissions by up to 23%. At the same time, the MRI-to-CT conversion performance was maintained, with only small variations in the Mean Absolute Error (MAE). Notably, for three out of the five evaluated architectures, no statistically significant differences were observed, while two architectures exhibited statistically significant improvements. Our work aligns with a research paradigm that promotes DL-based frameworks meeting clinical requirements while ensuring climatic, social, and economic sustainability. It lays the groundwork for novel FL evaluation frameworks, advancing privacy, equity and, more broadly, justice in AI-driven healthcare.

Abstract:

Cone-beam computed tomography (CBCT) has become a widely adopted modality for image-guided radiotherapy (IGRT). However, CBCT is characterized by increased noise, limited soft-tissue contrast, and artifacts. These issues result in unreliable Hounsfield unit (HU) values, which limits direct dose calculation. These issues were addressed by generating synthetic CT (sCT) from CBCT, particularly by adopting deep learning (DL) methods. However, existing DL approaches were hindered by institutional heterogeneity, scanner-dependent variations, and data privacy regulations that prevented multi-center data sharing.To overcome these challenges, we proposed a cross-silo federated learning approach for CBCT-to-sCT synthesis in the head and neck region. This approach extended our original FedSynthCT framework to a different image modality and anatomical region. A conditional generative adversarial network (cGAN) was trained using data from three European medical centers within the SynthRAD2025 public challenge dataset while maintaining data privacy at each institution. A combination of the FedAvg and FedProx aggregation strategies, alongside a standardized preprocessing pipeline, was adopted to federate the DL model.The federated model effectively generalized across participating centers, as evidenced by the mean absolute error (MAE) ranging from $64.38\pm13.63$ to $85.90\pm7.10$ HU, the structural similarity index (SSIM) ranging from $0.882\pm0.022$ to $0.922\pm0.039$, and the peak signal-to-noise ratio (PSNR) ranging from $32.86\pm0.94$ to $34.91\pm1.04$ dB. Notably, performance on an external validation dataset of 60 patients yielded comparable metrics: a MAE of $75.22\pm11.81$ HU, an SSIM of $0.904\pm0.034$ and a PSNR of $33.52\pm2.06$, confirming robust cross-center generalization despite differences in imaging protocols and scanner types, without additional training. Furthermore, a visual analysis of the results revealed that the obtained metrics were influenced by registration errors.Our findings demonstrated the technical feasibility of FL for CBCT-to-sCT synthesis task while preserving data privacy, offering a collaborative solution for developing generalizable models across institutions without requiring data sharing or center-specific models.

Abstract:

The generation of Synthetic Computed Tomography (sCT) images has become a pivotal methodology in modern clinical practice, particularly in the context of Radiotherapy (RT) treatment planning. The use of sCT enables the calculation of doses, pushing towards Magnetic Resonance Imaging (MRI) guided radiotherapy treatments. Moreover, with the introduction of MRI-Positron Emission Tomography (PET) hybrid scanners, the derivation of sCT from MRI can improve the attenuation correction of PET images.Deep learning methods for MRI-to-sCT have shown promising results, but their reliance on single-centre training dataset limits generalisation capabilities to diverse clinical settings. Moreover, creating centralised multi-centre datasets may pose privacy concerns. To address the aforementioned issues, we introduced FedSynthCT-Brain, an approach based on the Federated Learning (FL) paradigm for MRI-to-sCT in brain imaging. This is among the first applications of FL for MRI-to-sCT, employing a cross-silo horizontal FL approach that allows multiple centres to collaboratively train a U-Net-based deep learning model. We validated our method using real multicentre data from four European and American centres, simulating heterogeneous scanner types and acquisition modalities, and tested its performance on an independent dataset from a centre outside the federation.In the case of the unseen centre, the federated model achieved a median Mean Absolute Error (MAE) of 102.0 HU across 23 patients, with an interquartile range of 96.7–110.5 HU. The median (interquartile range) for the Structural Similarity Index (SSIM) and the Peak Signal to Noise Ratio (PNSR) were 0.89 (0.86–0.89) and 26.58 (25.52–27.42), respectively.The analysis of the results showed acceptable performances of the federated approach, thus highlighting the potential of FL to enhance MRI-to-sCT to improve generalisability and advancing safe and equitable clinical applications while fostering collaboration and preserving data privacy.

Abstract:

Limited medical image data hinders the training of deep learning (DL) models in the biomedical field. Image augmentation can reduce the data-scarcity problem by generating variations of existing images. However, currently implemented methods require coding, excluding non-programmer users from this opportunity.We therefore present ImageAugmenter, an easy-to-use and open-source module for 3D Slicer imaging computing platform. It offers a simple and intuitive interface for applying over 20 simultaneous MONAI Transforms (spatial, intensity, etc.) to medical image datasets, all without programming.ImageAugmenter makes accessible medical image augmentation, enabling a wider range of users to improve the performance of DL models in medical image analysis by increasing the number of samples available for training.

Abstract:

PurposeIdentifying and quantifying coronary artery calcification (CAC) is crucial for preoperative planning, as it helps to estimate both the complexity of the 2D coronary angiography (2DCA) procedure and the risk of developing intraoperative complications. Despite the relevance, the actual practice relies upon visual inspection of the 2DCA image frames by clinicians. This procedure is prone to inaccuracies due to the poor contrast and small size of the CAC; moreover, it is dependent on the physician’s experience. To address this issue, we developed a workflow to assist clinicians in identifying CAC within 2DCA using data from 44 image acquisitions across 14 patients.MethodsOur workflow consists of three stages. In the first stage, a classification backbone based on ResNet-18 is applied to guide the CAC identification by extracting relevant features from 2DCA frames. In the second stage, a U-Net decoder architecture, mirroring the encoding structure of the ResNet-18, is employed to identify the regions of interest (ROI) of the CAC. Eventually, a post-processing step refines the results to obtain the final ROI. The workflow was evaluated using a leave-out cross-validation.ResultsThe proposed method outperformed the comparative methods by achieving an F1-score for the classification step of 0.87 (0.77-0.94) (median ± quartiles), while for the CAC identification step the intersection over minimum (IoM) was 0.64 (0.46-0.86) (median ± quartiles).ConclusionThis is the first attempt to propose a clinical decision support system to assist the identification of CAC within 2DCA. The proposed workflow holds the potential to improve both the accuracy and efficiency of CAC quantification, with promising clinical applications. As future work, the concurrent use of multiple auxiliary tasks could be explored to further improve the segmentation performance.

Abstract:

In recent years, synthetic Computed Tomography (CT) images generated from Magnetic Resonance (MR) or Cone Beam Computed Tomography (CBCT) acquisitions have been shown to be comparable to real CT images in terms of dose computation for radiotherapy simulation. However, until now, there has been no independent strategy to assess the quality of each synthetic image in the absence of ground truth. In this work, we propose a Deep Learning (DL)-based framework to predict the accuracy of synthetic CT in terms of Mean Absolute Error (MAE) without the need for a ground truth (GT). The proposed algorithm generates a volumetric map as an output, informing clinicians of the predicted MAE slice-by-slice. A cascading multi-model architecture was used to deal with the complexity of the MAE prediction task. The workflow was trained and tested on two cohorts of head and neck cancer patients with different imaging modalities: 27 MR scans and 33 CBCT. The algorithm evaluation revealed an accurate HU prediction (a median absolute prediction deviation equal to 4 HU for CBCT-based synthetic CTs and 6 HU for MR-based synthetic CTs), with discrepancies that do not affect the clinical decisions made on the basis of the proposed estimation. The workflow exhibited no systematic error in MAE prediction. This work represents a proof of concept about the feasibility of synthetic CT evaluation in daily clinical practice, and it paves the way for future patient-specific quality assessment strategies.