Low-dose Images Research Articles

Conceptual design of a low-dose multi-parameter imaging system: Positron annihilation interaction-transmission imaging (PAITI)

The feasibility of Positron Annihilation Interaction-Transmission Imaging (PAITI), a novel 2D imaging modality based on positron annihilation photons, has been demonstrated through theoretical analysis and GATE Monte Carlo simulations. This system consists of two pixelated detectors on opposite sides of a point-like positron source. The object under investigation is positioned between the source and one of the detectors. By analyzing singles, coincidence, and anticoincidence counts, multiple 2D maps of the medium under investigation can be produced from a single data acquisition, such as the map of the number of occurred interactions, linear attenuation coefficient map, deposited energy map, and electron density map.Performance evaluations were conducted using simulated uniform phantoms (water, carbon, cortical bone, and titanium) and nonuniform phantoms (air and bone inside water). The average relative errors for determining the number of occurred interactions and attenuation coefficient were 4.42% and 5.0%, respectively. Additionally, the average relative errors for determining deposited energy and electron density were 3.15% and 4.7%, respectively. The system's capacity to discern fluctuations in electron density was also examined, demonstrating its capability to identify changes of 12.25% with a confidence level surpassing 95.3%, even at an absorbed dose much less than 15.14 ± 0.25 μGy.The results demonstrated the system's potential for future advancements and practical applications in medical imaging, specifically for low-dose multi-parameter anatomical imaging. In addition, the system's ability to determine electron density enables it to perform bone densitometry and ion range determination, with the latter being critical for patient-specific treatment planning for ion therapy.

Semi-supervised low-dose SPECT restoration using sinogram inner-structure aware graph neural network

Objective. To mitigate the potential radiation risk, low-dose single photon emission computed tomography (SPECT) is of increasing interest. Numerous deep learning-based methods have been developed to perform low-dose imaging while maintaining image quality. However, most existing methods seldom explore the unique inner-structure inherent within sinograms. In addition, traditional supervised learning methods require large-scale labeled data, where the normal-dose data serves as annotation and is intractable to acquire in low-dose imaging. In this study, we aim to develop a novel sinogram inner-structure-aware semi-supervised framework for the task of low-dose SPECT sinogram restoration. Approach. The proposed framework retains the strengths of UNet, meanwhile introducing a sinogram-structure-based non-local neighbors graph neural network (SSN-GNN) module and a window-based K-nearest neighbors GNN (W-KNN-GNN) module to effectively exploit the inherent inner-structure within SPECT sinograms. Moreover, the proposed framework employs the mean teacher semi-supervised learning approach to leverage the information available in abundant unlabeled low-dose sinograms. Main results. The datasets exploited in this study were acquired from the (Extended Cardiac-Torso) XCAT anthropomorphic digital phantoms, which provide realistic images for imaging research of various modalities. Quantitative as well as qualitative results demonstrate that the proposed framework achieves superior performance compared to several state-of-the-art reconstruction methods. To further validate the effectiveness of the proposed framework, ablation and robustness experiments were also performed. The experimental results show that each component of the proposed framework effectively improves the model performance, and the framework exhibits superior robustness with respect to various noise levels. Besides, the proposed semi-supervised paradigm showcases the efficacy of incorporating supplementary unlabeled low-dose sinograms. Significance. The proposed framework improves the quality of low-dose SPECT reconstructed images by utilizing sinogram inner-structure and incorporating supplementary unlabeled data, which provides an important tool for dose reduction without sacrificing the image quality.

Amorphous Albumin Gadolinium-Based Nanoparticles for Ultrahigh-Resolution Magnetic Resonance Angiography.

Magnetic resonance angiography (MRA) contrast agents are extensively utilized in clinical practice due to their capability of improving the image resolution and sensitivity. However, the clinically approved MRA contrast agents have the disadvantages of a limited acquisition time window and high dose administration for effective imaging. Herein, albumin-coated gadolinium-based nanoparticles (BSA-Gd) were meticulously developed for in vivo ultrahigh-resolution MRA. Compared to Gd-DTPA, BSA-Gd exhibits a significantly higher longitudinal relaxivity (r1 = 76.7 mM-1 s-1), nearly 16-fold greater than that of Gd-DTPA, and an extended blood circulation time (t1/2 = 40 min), enabling a dramatically enhanced high-resolution imaging of microvessels (sub-200 μm) and low dose imaging (about 1/16 that of Gd-DTPA). Furthermore, the clinically significant fine vessels were successfully mapped in large mammals, including a circle of Willis, kidney and liver vascular branches, tumor vessels, and differentiated arteries from veins using dynamic contrast-enhanced MRA BSA-Gd, and have superior imaging capability and biocompatibility, and their clinical applications hold substantial promise.

The Potential of Subsampling and Inpainting for Fast Low-Dose Cryo FIB-SEM Imaging.

Traditional image acquisition for cryo focused ion-beam scanning electron microscopy (FIB-SEM) tomography often sees thousands of images being captured over a period of many hours, with immense data sets being produced. When imaging beam sensitive materials, these images are often compromised by additional constraints related to beam damage and the devitrification of the material during imaging, which renders data acquisition both costly and unreliable. Subsampling and inpainting are proposed as solutions for both of these aspects, allowing fast and low-dose imaging to take place in the Focused ion-beam scanning electron microscopy FIB-SEM without an appreciable loss in image quality. In this work, experimental data are presented which validate subsampling and inpainting as a useful tool for convenient and reliable data acquisition in a FIB-SEM, with new methods of handling three-dimensional data being employed in the context of dictionary learning and inpainting algorithms using a newly developed microscope control software and data recovery algorithm.

Dual heterogeneous interfaces enhance X-ray excited persistent luminescence for low-dose 3D imaging

Lanthanide-doped fluoride nanoparticles (NPs) showcase adjustable X-ray-excited persistent luminescence (XEPL), holding significant promise for applications in three-dimensional (3D) imaging through the creation of flexible X-ray detectors. However, a dangerous high X-ray irradiation dose rate and complicated heating procedure are required to generate efficient XEPL for high-resolution 3D imaging, which is attributed to a lack of strategies to significantly enhance the XEPL intensity. Here we report that the XEPL intensity of a series of lanthanide activators (Dy, Pr, Er, Tm, Gd, Tb) is greatly improved by constructing dual heterogeneous interfaces in a double-shell nanostructure. Mechanistic studies indicate that the employed core@shell@shell structure could not only passivate the surface quenchers to lower the non-radiative relaxation possibility, but also reduce the interfacial Frenkel defect formation energy leading to increase the trap concentration. By employing a NPs containing flexible film as the scintillation screen, the inside 3D electrical structure of a watch was clearly achieved based on the delayed XEPL imaging and 3D reconstruction procedure. We foresee that these findings will promote the development of advanced X-ray activated persistent fluoride NPs and offer opportunities for safer and more efficient X-ray imaging techniques in a number of scientific and practical areas.

EOS® is reliable to evaluate spinopelvic parameters: a validation study

BackgroundSagittal and coronal standing radiographs have been the standard imaging for assessing spinal alignment. However, their disadvantages include distortion at the image edges and low interobserver reliability in some parameters. EOS® is a low-dose biplanar digital radiographic imaging system that can avoid distortion by obtaining high-definition images.MethodsThis study aimed to evaluate spinopelvic parameters in conventional lateral C1S1 upright radiographs and EOS® images and compare them. Patients with non-deformity changes were subjected to routine clinical examinations. Plain AP and lateral X-ray radiographs were obtained along the entire spine length. Patients were also referred for full-length EOS® of the spine. Thoracic Kyphosis (TK), Lumbar Lordosis (LL), Pelvic Tilt (PT), Sacral Slope (SS), Pelvic Incidence (PI), and Sagittal Vertical Axis (SVA) were measured in the two studies by an orthopedic surgeon and a radiologist using PACS software. Also, the orthopedic surgeon evaluated the studies again after two weeks. Intra- and inter-observer reliability was then assessed using the interclass correlation coefficient (ICC). Also, the coefficient of variation was used to assess intra- and inter-observer reliability. Bland-Altman plots were drawn for each parameter.ResultsThe mean age was 48.2 ± 6.6 years. Among the 50 patients, 30 (60%) were female. The mean ICC for TK, LL, PT, SS, PI, and SVA in EOS® images are 0.95, 0.95, 0.92, 0.90, 0.94, and 0.98, respectively, and in C1S1 radiography images, it was 0.92, 0.87, 0.94, 0.88, 0.93, and 0.98, respectively which shows good to excellent results. The coefficient of variation for intraobserver reliability was relatively low (< 18.6%), while it showed higher percentages in evaluating interobserver reliability (< 54.5%). Also, the Bland-Altman plot showed good agreement for each parameter.ConclusionSpinopelvic parameters, e.g., TK, LL, SS, PI, and SS, in EOS® are reliable and comparable to those in conventional lateral upright C1S1 radiographs.

APNet: Adaptive projection network for medical image denoising.

In clinical medicine, low-dose radiographic image noise reduces the quality of the detected image features and may have a negative impact on disease diagnosis. In this study, Adaptive Projection Network (APNet) is proposed to reduce noise from low-dose medical images. APNet is developed based on an architecture of the U-shaped network to capture multi-scale data and achieve end-to-end image denoising. To adaptively calibrate important features during information transmission, a residual block of the dual attention method throughout the encoding and decoding phases is integrated. A non-local attention module to separate the noise and texture of the image details by using image adaptive projection during the feature fusion. To verify the effectiveness of APNet, experiments on lung CT images with synthetic noise are performed, and the results demonstrate that the proposed approach outperforms recent methods in both quantitative index and visual quality. In addition, the denoising experiment on the dental CT image is also carried out and it verifies that the network has a certain generalization. The proposed APNet is an effective method that can reduce image noise and preserve the required image details in low-dose radiographic images.

General Algorithm Applicability in Determining DBS Lead Orientation: Adapting 2D and 3D X-Ray Techniques for SenSightTM Leads

Introduction: With recent advancements in deep brain stimulation (DBS), directional leads featuring segmented contacts have been introduced, allowing for targeted stimulation of specific brain regions. Given that manufacturers employ diverse markers for lead orientation, our investigation focuses on the adaptability of the 2017 techniques proposed by the Cologne research group for lead orientation determination. Methods: We tailored the two separate 2D and 3D X-ray-based techniques published in 2017 and originally developed for C-shaped markers, to the dual-marker of the Medtronic SenSight™ lead. In a retrospective patient study, we evaluated their feasibility and consistency by comparing the degree of agreement between the two methods. Results: The Bland-Altman plot showed favorable concordance without any noticeable systematic errors. The mean difference was 0.79°, with limits of agreement spanning from 21.4° to −19.8°. The algorithms demonstrated high reliability, evidenced by an intraclass correlation coefficient of 0.99 (p < 0.001). Conclusion: The 2D and 3D algorithms, initially formulated for discerning the circular orientation of a C-shaped marker, were adapted to the marker of the Medtronic SenSight™ lead. Statistical analyses revealed a significant level of agreement between the two methods. Our findings highlight the adaptability of these algorithms to different markers, achievable through both low-dose intraoperative 2D X-ray imaging and standard CT imaging.

Multiple Kernel Synthesis of Head CT Using a Task-Based Loss Function

In CT imaging of the head, multiple image series are routinely reconstructed with different kernels and slice thicknesses. Reviewing the redundant information is an inefficient process for radiologists. We address this issue with a convolutional neural network (CNN)-based technique, synthesiZed Improved Resolution and Concurrent nOise reductioN (ZIRCON), that creates a single, thin, low-noise series that combines the favorable features from smooth and sharp head kernels. ZIRCON uses a CNN model with an autoencoder U-Net architecture that accepts two input channels (smooth- and sharp-kernel CT images) and combines their salient features to produce a single CT image. Image quality requirements are built into a task-based loss function with a smooth and sharp loss terms specific to anatomical regions. The model is trained using supervised learning with paired routine-dose clinical non-contrast head CT images as training targets and simulated low-dose (25%) images as training inputs. One hundred unique de-identified clinical exams were used for training, ten for validation, and ten for testing. Visual comparisons and contrast measurements of ZIRCON revealed that thinner slices and the smooth-kernel loss function improved gray-white matter contrast. Combined with lower noise, this increased visibility of small soft-tissue features that would be otherwise impaired by partial volume averaging or noise. Line profile analysis showed that ZIRCON images largely retained sharpness compared to the sharp-kernel input images. ZIRCON combined desirable image quality properties of both smooth and sharp input kernels into a single, thin, low-noise series suitable for both brain and skull imaging.

The differential diagnosis of lung precursor glandular lesions, micro-invasive adenocarcinoma, and invasive adenocarcinoma using low dose spectral computed tomography perfusion imaging.

Few studies about the association between computed tomography (CT) perfusion imaging parameters and invasiveness in lung adenocarcinoma (LUAD) have been conducted using low dose spectral CT perfusion imaging. The purpose of this study was to investigate application of spectral revolution CT low-dose perfusion imaging in the differential diagnosis of different pathological subtypes of LUAD. This was a cross-sectional study based on historical data from January 2018 to May 2019 in Peking University Cancer Hospital & Institute. A total of 62 cases were enrolled, including 2 cases of atypical adenomatous hyperplasia (AAH), 3 cases of adenocarcinoma in situ (AIS), 4 cases of minimally invasive adenocarcinoma (MIA), and 53 cases of invasive adenocarcinoma (IAC), all confirmed with pathology. The inclusion and exclusion criteria were regulated. Using Revolution low-dose CT perfusion imaging (GE, USA), the CT perfusion parameters of hemodynamics were obtained: blood flow (BF), blood volume (BV), impulse residue function time of arrival (IRF TO), maximum slope of increase (MSI), mean transit time (MTT), permeability surface area product (PS), positive enhancement integral (PEI), and maximum enhancement time (Tmax). Univariate analysis of variance (ANOVA) or Kruskal-Wallis test was used to compare the differences of CT perfusion quantitative parameters among AAH, AIS, MIA, and IAC. Mann-Whitney test was used to compare the difference of CT perfusion imaging parameters between preinvasive lesions (AAH and AIS) and invasive lung cancer (MIA and IAC). Statistically significant differences in IRF TO were observed in LUAD with different invasiveness, namely, among AIS, MIA, and IAC groups (0.56±0.74 vs. 0.54±1.08 vs. 4.39±2.19, P=0.004). Statistically significant differences in IRF TO were also observed between pre-invasive lesions group (AAH and AIS) and invasive lung cancer group (MIA and IAC) (1.12±1.27 vs. 3.75±2.79, P=0.031), and between AAH + AIS + MIA groups and IAC group (0.83±1.13 vs. 4.12±2.69, P<0.001). There were no statistically significant differences in other CT perfusion parameters of hemodynamics among different pathological subtypes of LUAD (P>0.05). The low-dose perfusion parameter IRF TO of revolution CT has the potential to be employed in the differential diagnosis of different pathological subtypes of LUAD.

Virtual clinical trial-based study for clinical evaluation of projection-reduced low-dose cone-beam CT for image guided radiotherapy.

Repeated cone-beam CT (CBCT) scans for image-guided radiotherapy (IGRT) increase the health risk of radiation-induced malignancies. Patient-enrolled studies to optimize scan protocols are inadequate. We proposed a virtual clinical trial-based approach to evaluate projection-reduced low-dose CBCT for IGRT. A total of 71 patients were virtually enrolled with 26 head, 23 thorax and 22 pelvis scans. Projection numbers of full-dose CBCT scans were reduced to 1/2, 1/4, and 1/8 of the original to simulate low-dose scans. Contrast-to-noise ratio (CNR) values in fat and muscle were measured in the full-dose and low-dose images. CBCT images were registered to planning CT to derive 6-degree-of-freedom couch shifts. Registration errors were statistically analyzed with the Wilcoxon paired signed-rank test. As projection numbers were reduced, CNR values descended and the magnitude of registration errors increased. The mean CNR values of full-dose and half-dose CBCT were >3.0. For full-dose and low-dose CBCT (i.e. 1/2, 1/4 and 1/8 full-dose), the mean registration errors were< ± 0.4 mm in translational directions (LAT, LNG, VRT) and ±0.2 degree in rotational directions (Pitch, Roll, Yaw); the mean magnitude of registration errors were< 1 mm in translation and< 0.5 degree in rotation. The couch shift differences between full-dose and low-dose CBCT were not statistically significant (p>0.05) in all the directions. The results indicate that while the impact of dose-reduction on CBCT couch shifts is not significant, the impact on CNR values is significant. Further validation on optimizing CBCT imaging dose is required.

Low-Light Phase Retrieval With Implicit Generative Priors.

Phase retrieval (PR) is fundamentally important in scientific imaging and is crucial for nanoscale techniques like coherent diffractive imaging (CDI). Low radiation dose imaging is essential for applications involving radiation-sensitive samples. However, most PR methods struggle in low-dose scenarios due to high shot noise. Recent advancements in optical data acquisition setups, such as in-situ CDI, have shown promise for low-dose imaging, but they rely on a time series of measurements, making them unsuitable for single-image applications. Similarly, data-driven phase retrieval techniques are not easily adaptable to data-scarce situations. Zero-shot deep learning methods based on pre-trained and implicit generative priors have been effective in various imaging tasks but have shown limited success in PR. In this work, we propose low-dose deep image prior (LoDIP), which combines in-situ CDI with the power of implicit generative priors to address single-image low-dose phase retrieval. Quantitative evaluations demonstrate LoDIP's superior performance in this task and its applicability to real experimental scenarios.

Randomly-Accessible Active Pixel Sensor with Logarithmic-Exponential Response Enabling Low Dose and Wide Dynamic Range Indirect-Conversion X-ray Imaging

Randomly-Accessible Active Pixel Sensor with Logarithmic-Exponential Response Enabling Low Dose and Wide Dynamic Range Indirect-Conversion X-ray Imaging

Enabling Predication of the Deep Learning Algorithms for Low-Dose CT Scan Image Denoising Models: A Systematic Literature Review

Enabling Predication of the Deep Learning Algorithms for Low-Dose CT Scan Image Denoising Models: A Systematic Literature Review

Emulating Low-Dose PCCT Image Pairs with Independent Noise for Self-Supervised Spectral Image Denoising.

Photon counting CT (PCCT) acquires spectral measurements and enables generation of material decomposition (MD) images that provide distinct advantages in various clinical situations. However, noise amplification is observed in MD images, and denoising is typically applied. Clean or high-quality references are rare in clinical scans, often making supervised learning (Noise2Clean) impractical. Noise2Noise is a self-supervised counterpart, using noisy images and corresponding noisy references with zero-mean, independent noise. PCCT counts transmitted photons separately, and raw measurements are assumed to follow a Poisson distribution in each energy bin, providing the possibility to create noise-independent pairs. The approach is to use binomial selection to split the counts into two low-dose scans with independent noise. We prove that the reconstructed spectral images inherit the noise independence from counts domain through noise propagation analysis and also validated it in numerical simulation and experimental phantom scans. The method offers the flexibility to split measurements into desired dose levels while ensuring the reconstructed images share identical underlying features, thereby strengthening the model's robustness for input dose levels and capability of preserving fine details. In both numerical simulation and experimental phantom scans, we demonstrated that Noise2Noise with binomial selection outperforms other common self-supervised learning methods based on different presumptive conditions.

Dental imaging in clinical photon-counting CT at a quarter of DVT dose

ObjectiveTo investigate the image quality of a low-dose dental imaging protocol in the first clinical photon-counting computed tomography (PCCT) system in comparison to a normal-dose acquisition in a digital volume tomography (DVT) system. Materials and methodsClinical PCCT systems offer an increased spatial resolution compared to previous generations of clinical systems. Their spatial resolution is in the order of dental DVT systems. Resolution-matched acquisitions of ten porcine jaws were performed in a PCCT (Naeotom Alpha, Siemens Healthineers) and in a DVT (Orthophos XL, Dentsply Sirona). PCCT images were acquired with 90 kV at a dose of 1 mGy CTDI16 cm. DVT used 85 kV at 4 mGy. Image reconstruction was performed using the standard algorithms of each system to a voxel size of 160 × 160 × 200 µm. The dose-normalized contrast-to-noise ratio (CNRD) was measured between dentine and enamel and dentine and bone. Two readers evaluated overall diagnostic quality of images and quality of relevant structures such as root channels and dentine. ResultsCNRD is higher in all PCCT acquisitions. CNRD is 37 % higher for the contrast dentine-enamel and 31 % higher for the dentine-bone contrast (p < 0.05). Overall diagnostic image quality was higher for PCCT over DVT (p < 0.02 and p < 0.04 for readers 1 and 2). Quality scores for anatomical structures were higher in PCCT compared to DVT (all p < 0.05). Inter- and intrareader reproducibility were acceptable (all ICC>0.64). ConclusionsPCCT provides an increased image quality over DVT even at a lower dose level and might enable complex dental imaging protocols in the future. Clinical significanceThe evolution of photon-counting technology and it's optimization will increasingly move dental imaging towards standardized 3D visualizations providing both minimal radiation exposure and high diagnostic accuracy.

Deep learning-based projection synthesis for low-dose cone-beam computed tomography imaging in image-guided radiotherapy.

The imaging dose of cone-beam computed tomography (CBCT) in image-guided radiotherapy (IGRT) poses adverse effects on patient health. To improve the quality of sparse-view low-dose CBCT images, a projection synthesis convolutional neural network (SynCNN) model is proposed. Included in this retrospective, single-center study were 223 patients diagnosed with brain tumours from Beijing Cancer Hospital. The proposed SynCNN model estimated two pairs of orthogonally direction-separable spatial kernels to synthesize the missing projection in between the input neighboring sparse-view projections via local convolution operations. The SynCNN model was trained on 150 real patients to learn patterns for inter-view projection synthesis. CBCT data from 30 real patients were used to validate the SynCNN, while data from a phantom and 43 real patients were used to test the SynCNN externally. Sparse-view projection datasets with 1/2, 1/4, and 1/8 of the original sampling rate were simulated, and the corresponding full-view projection datasets were restored using the SynCNN model. The tomographic images were then reconstructed with the Feldkamp-Davis-Kress algorithm. The root-mean-square error (RMSE), peak signal-to-noise ratio (PSNR), and structural similarity (SSIM) metrics were measured in both the projection and image domains. Five experts were invited to grade the image quality blindly for 40 randomly selected evaluation groups with a four-level rubric, where a score greater than or equal to 2 was considered acceptable image quality. The running time of the SynCNN model was recorded. The SynCNN model was directly compared with the three other methods on 1/4 sparse-view reconstructions. The phantom and patient studies showed that the missing projections were accurately synthesized. In the image domain, for the phantom study, compared with images reconstructed from sparse-view projections, images with SynCNN synthesis exhibited significantly improved qualities with decreased values in RMSE and increased values in PSNR and SSIM. For the patient study, between the results with and without the SynCNN synthesis, the averaged RMSE decreased by 3.4×10-4, 10.3×10-4, and 21.7×10-4, the averaged PSNR increased by 3.4, 6.6, and 9.4 dB, and the averaged SSIM increased by 5.2×10-2, 18.9×10-2 and 33.9×10-2, for the 1/2, 1/4, and 1/8 sparse-view reconstructions, respectively. In expert subjective evaluation, both the median scores and acceptance rates of the images with SynCNN synthesis were higher than those reconstructed from sparse-view projections. It took the model less than 0.01 s to synthesize an inter-view projection. Compared with the three other methods, the SynCNN model obtained the best scores in terms of the three metrics in both domains. The proposed SynCNN model effectively improves the quality of sparse-view CBCT images at a low time cost. With the SynCNN model, the CBCT imaging dose in IGRT could be reduced potentially.

An examination of exit-wave reconstruction algorithms for low dose imaging at atomic-scale resolution

An examination of exit-wave reconstruction algorithms for low dose imaging at atomic-scale resolution

Atomic-level imaging of beam-sensitive COFs and MOFs by low-dose electron microscopy.

Electron microscopy, an important technique that allows for the precise determination of structural information with high spatiotemporal resolution, has become indispensable in unravelling the complex relationships between material structure and properties ranging from mesoscale morphology to atomic arrangement. However, beam-sensitive materials, particularly those comprising organic components such as metal-organic frameworks (MOFs) and covalent organic frameworks (COFs), would suffer catastrophic damage from the high energy electrons, hindering the determination of atomic structures. A low-dose approach has arisen as a possible solution to this problem based on the integration of advancements in several aspects: electron optical system, detector, image processing, and specimen preservation. This article summarizes the transmission electron microscopy characterization of MOFs and COFs, including local structures, host-guest interactions, and interfaces at the atomic level. Revolutions in advanced direct electron detectors, algorithms in image acquisition and processing, and emerging methodology for high quality low-dose imaging are also reviewed. Finally, perspectives on the future development of electron microscopy methodology with the support of computer science are presented.

Unsupervised Domain Adaptation for Low-Dose CT Reconstruction via Bayesian Uncertainty Alignment.

Low-dose computed tomography (LDCT) image reconstruction techniques can reduce patient radiation exposure while maintaining acceptable imaging quality. Deep learning (DL) is widely used in this problem, but the performance of testing data (also known as target domain) is often degraded in clinical scenarios due to the variations that were not encountered in training data (also known as source domain). Unsupervised domain adaptation (UDA) of LDCT reconstruction has been proposed to solve this problem through distribution alignment. However, existing UDA methods fail to explore the usage of uncertainty quantification, which is crucial for reliable intelligent medical systems in clinical scenarios with unexpected variations. Moreover, existing direct alignment for different patients would lead to content mismatch issues. To address these issues, we propose to leverage a probabilistic reconstruction framework to conduct a joint discrepancy minimization between source and target domains in both the latent and image spaces. In the latent space, we devise a Bayesian uncertainty alignment to reduce the epistemic gap between the two domains. This approach reduces the uncertainty level of target domain data, making it more likely to render well-reconstructed results on target domains. In the image space, we propose a sharpness-aware distribution alignment (SDA) to achieve a match of second-order information, which can ensure that the reconstructed images from the target domain have similar sharpness to normal-dose CT (NDCT) images from the source domain. Experimental results on two simulated datasets and one clinical low-dose imaging dataset show that our proposed method outperforms other methods in quantitative and visualized performance.