18 March 2024 in IEEE Journal of Biomedical and Health Informatics
Deep learning (DL) algorithms have achieved unprecedented success in low-dose CT (LDCT) imaging and are expected to be a new generation of CT reconstruction technology. However, most DL-based denoising models often lack the ability to generalize to unseen dose data. And they only learn the posterior distribution of latent normal-dose CT (NDCT) images conditioned on observed LDCT images in the traditional maximum a posteriori (MAP) framework, while ignoring the noise generation process of LDCT images. Moreover, most simulation tools for LDCT typically operate on proprietary projection data, which is generally not accessible without an established collaboration with CT manufacturers. To alleviate these issues, in this work, we propose a dose-agnostic dual-task transfer network, termed DDT-Net, for simultaneous LDCT denoising and simulation. Concretely, the dual-task learning module is constructed to integrate the LDCT denoising and simulation tasks into a unified optimization framework by learning the joint distribution of LDCT and NDCT data. We approximate the joint distribution of continuous dose level data by training DDT-Net with discrete dose data, which can be generalized to denoising and simulation of unseen dose data. In particular, the mixed-dose training strategy adopted by DDT-Net can promote the denoising performance of lower-dose data. The paired dataset simulated by DDT-Net can be used for data augmentation to further restore the tissue texture of LDCT images. Experimental results on synthetic data and clinical data show that the proposed DDT-Net outperforms competing methods in terms of denoising and generalization performance at unseen dose data, and it also provides a simulation tool that can quickly simulate realistic LDCT images at arbitrary dose levels.
28 May 2025 in IEEE Transactions on Radiation and Plasma Medical Sciences
Recent advances in deep learning-based methods have shown great potential in improving low-dose CT image quality. Meanwhile, these methods are constructed based on a large, centralized, and diverse CT dataset from multiple institutions that is difficult to collect and share due to the high-cost acquisition and data privacy regulations. Previously developed federated learning (FL)-based methods enable collaborative and decentralized training without exchanging local data to preserve data privacy. In this work, we focus on analyzing the robustness of FL-based methods against dataset shifts (i.e., the datasets among multiple institutions are from different scanners, different protocols, or different sampling conditions). The results show that the FL-based CT reconstruction methods are sensitive to domain shifts, which can be attributed to the data heterogeneity among multiple institutions. Based on these findings, we propose a unified CT reconstruction method that leverages high-quality metadata (e.g., low-dose images and their corresponding normal-dose counterparts) stored on the cloud server to address the challenge of multi-institutional domain shifts. For simplicity, we refer to the proposed method as FM-iRadonMAP, representing federated metadata learning with a personalized condition-modulated iRadonMAP. Specifically, the FM-iRadonMAP consists of two modules, i.e., Condition Modulated iRadonMap (CM-iRadonMAP) and Federated Metadata Learning (FMDL). CM-iRadonMAP introduces the knowledge of client-specific sampling conditions, i.e., imaging geometries and scan protocols, into iRadonMAP reconstruction network at each client to modulate the reconstruction effectively. FMDL trains a supervised meta model using high-quality metadata in an additional round and then adaptively unifies the network parameters of the meta model with those of the local models from all clients for broadcasting, addressing the issue of data heterogeneity. A large-scale multi-institutional CT dataset is used to validate and evaluate the reconstruction performance of the FM-iRadonMAP. The experimental results demonstrate the feasibility of the FM-iRadonMAP for multi-institutional CT reconstruction with severe data heterogeneity.
2 September 2026 in Physics in Medicine & Biology
Objective. The growing adoption of multi-center diagnostic and treatment protocols has heightened the demand for consistent image quality across Computed tomography (CT) devices, yet scanner-specific variations in image characteristics pose a significant challenge. Recently, deep learning-based low-dose CT (LDCT) reconstruction algorithms have been widely developed and successfully deployed in commercial clinical systems. However, existing methods for cross-scanner LDCT imaging suffer from inadequate alignment of data distributions across different scanners, which constrains their generalization performance. This study aims to propose a single model for cross-scanner LDCT image denoising. Approach. We propose a unified-distribution residual diffusion model (UDRDiff) for generalizable LDCT image denoising across diverse scanners. Specifically, we design a unified-distribution degeneration operator to progressively map cross-scanner LDCT data into a shared latent distribution by incorporating an input-suppression term, effectively mitigating inter-scanner discrepancies. To enhance the interpretability of the forward diffusion process, we introduce a residual diffusion mechanism that establishes a deterministic degradation process from normal-dose CT to LDCT, which can clearly guide the reverse image restoration process and significantly accelerate the sampling process. Main results. We extensively validated our method on datasets from four scanners, covering two manufacturers and two anatomical regions. Experimental results demonstrate that our method outperforms existing approaches across all test scenarios. Taking two representative test sets as examples: on the Scanner 1 (abdominal images) dataset, our method improves peak signal-to-noise ratio (PSNR) from 43.0765 dB to 43.5324 dB and structural similarity index (SSIM) from 0.9634 to 0.9660; on the Scanner 4 (head images) dataset, PSNR improves from 44.3582 dB to 45.9250 dB and SSIM from 0.9545 to 0.9744. These results sufficiently validate the effectiveness and generalization capability of our method across different scanners and anatomical regions. Significance. UDRDiff effectively addresses the data distribution alignment problem in cross-scanner LDCT denoising, thereby enhancing model generalizability under multi-center, multi-scanning conditions, and provides a feasible path for cross-scanner denoising of clinical LDCT images.