Unpaired Data Research Articles

Analysis of published and clinical sample data to develop new ranges of liver and blood concentrations of vitamin A for diagnosis of vitamin A deficiency in cattle

In cattle, liver stores of vitamin A (VA) maintain blood VA concentrations until liver supplies are nearly exhausted, pro­viding VA to tissues during insufficient dietary intake. There­fore, normal blood levels do not indicate adequate VA, but low blood levels indicate a deficiency. We hypothesized that previous diagnostic criteria for VA deficiency needed updating and that blood levels of VA could be used as indicators of liver VA deficiency. We performed multiple regressions of paired blood and liver sample data from published experimental and field case reports of VA deficiency in cattle to demonstrate this. We fitted various models using data from unpaired clini­cal samples submitted to the Iowa State University Diagnostic Laboratory by comparing fractiles of liver and plasma/serum samples. Using these relationships, we propose new ranges for severely deficient, deficient, marginal and adequate, and we also propose that blood samples alone can be used to di­agnose deficiency in the presence of clinical signs of VA defi­ciency. Veterinary diagnostic laboratories can use this infor­mation to better estimate serum and liver VA concentrations for unpaired data and recommend to cattle practitioners the status of VA in their herds.

Multi-scale Transformer Network with Edge-aware Pre-training for Cross-Modality MR Image Synthesis.

Cross-modality magnetic resonance (MR) image synthesis can be used to generate missing modalities from given ones. Existing (supervised learning) methods often require a large number of paired multi-modal data to train an effective synthesis model. However, it is often challenging to obtain sufficient paired data for supervised training. In reality, we often have a small number of paired data while a large number of unpaired data. To take advantage of both paired and unpaired data, in this paper, we propose a Multi-scale Transformer Network (MT-Net) with edge-aware pre-training for cross-modality MR image synthesis. Specifically, an Edge-preserving Masked AutoEncoder (Edge-MAE) is first pre-trained in a self-supervised manner to simultaneously perform 1) image imputation for randomly masked patches in each image and 2) whole edge map estimation, which effectively learns both contextual and structural information. Besides, a novel patch-wise loss is proposed to enhance the performance of Edge-MAE by treating different masked patches differently according to the difficulties of their respective imputations. Based on this proposed pre-training, in the subsequent fine-tuning stage, a Dual-scale Selective Fusion (DSF) module is designed (in our MT-Net) to synthesize missing-modality images by integrating multi-scale features extracted from the encoder of the pre-trained Edge-MAE. Furthermore, this pre-trained encoder is also employed to extract high-level features from the synthesized image and corresponding ground-truth image, which are required to be similar (consistent) in the training. Experimental results show that our MT-Net achieves comparable performance to the competing methods even using 70% of all available paired data. Our code will be released at https://github.com/lyhkevin/MT-Net.

Cross-Domain Unpaired Learning for Low-Dose CT Imaging.

Supervised deep-learning techniques with paired training datasets have been widely studied for low-dose computed tomography (LDCT) imaging with excellent performance. However, the paired training datasets are usually difficult to obtain in clinical routine, which restricts the wide adoption of supervised deep-learning techniques in clinical practices. To address this issue, a general idea is to construct a pseudo paired training dataset based on the widely available unpaired data, after which, supervised deep-learning techniques can be adopted for improving the LDCT imaging performance by training on the pseudo paired training dataset. However, due to the complexity of noise properties in CT imaging, the LDCT data are difficult to generate in order to construct the pseudo paired training dataset. In this paper, we propose a simple yet effective cross-domain unpaired learning framework for pseudo LDCT data generation and LDCT image reconstruction, which is denoted as CrossDuL. Specifically, a dedicated pseudo LDCT sinogram generative module is constructed based on a data-dependent noise model in the sinogram domain, and then instead of in the sinogram domain, a pseudo paired dataset is constructed in the image domain to train an LDCT image restoration module. To validate the effectiveness of the proposed framework, clinical datasets are adopted. Experimental results demonstrate that the CrossDuL framework can obtain promising LDCT imaging performance in both quantitative and qualitative measurements.

GMRLNet: A graph-based manifold regularization learning framework for placental insufficiency diagnosis on incomplete multimodal ultrasound data.

Multimodal analysis of placental ultrasound (US) and microflow imaging (MFI) could greatly aid in the early diagnosis and interventional treatment of placental insufficiency (PI), ensuring a normal pregnancy. Existing multimodal analysis methods have weaknesses in multimodal feature representation and modal knowledge definitions and fail on incomplete datasets with unpaired multimodal samples. To address these challenges and efficiently leverage the incomplete multimodal dataset for accurate PI diagnosis, we propose a novel graph-based manifold regularization learning (MRL) framework named GMRLNet. It takes US and MFI images as input and exploits their modality-shared and modality-specific information for optimal multimodal feature representation. Specifically, a graph convolutional-based shared and specific transfer network (GSSTN) is designed to explore intra-modal feature associations, thus decoupling each modal input into interpretable shared and specific spaces. For unimodal knowledge definitions, graph-based manifold knowledge is introduced to describe the sample-level feature representation, local inter-sample relations, and global data distribution of each modality. Then, an MRL paradigm is designed for inter-modal manifold knowledge transfer to obtain effective cross-modal feature representations. Furthermore, MRL transfers the knowledge between both paired and unpaired data for robust learning on incomplete datasets. Experiments were conducted on two clinical datasets to validate the PI classification performance and generalization of GMRLNet. State-of-the-art comparisons show the higher accuracy of GMRLNet on incomplete datasets. Our method achieves 0.913 AUC and 0.904 balanced accuracy (bACC) for paired US and MFI images, as well as 0.906 AUC and 0.888 bACC for unimodal US images, illustrating its application potential in PI CAD systems.

NO donors on top of anti-hypertensive therapy reduces complications in chronic hypertensive pregnancies with hypodynamic circulation

ObjectivesChronic hypertension is associated with significant adverse maternal and fetal outcomes that appear to be often associated to a hypodynamic circulation. Treatment of hypertensive disorders of pregnancy tailored on maternal hemodynamics might reduce or mitigate these complications. Our purpose was to assess the hemodynamic modifications induced by the addition of NO donors and increased oral fluid intake on top of standard antihypertensive therapy in hypodynamic chronic hypertensive patients. We further evaluated if the possible hemodynamic modification induced by NO donors and increased oral fluid intake might be associated to a reduction of the severity and rate of complications vs. patients on antihypertensive standard treatment. Study DesignThis was a case-control study of 321 chronic hypertensive patients with a hypodynamic circulation at the echocardiographic evaluation at 24 weeks’ gestation. We included 160 controls (standard antihypertensive therapy) and 161 cases (standard therapy + NO donor patches + increased oral fluid intake). Student T test for paired and unpaired data, univariate logistic regression analysis, ROC curve analysis, and Cox Hazards Regression analysis were used as appropriate. ResultsAt enrollment the hemodynamic parameters were similar between the two groups. After 3–4 weeks stroke volume (77 ± 19 mL vs. 69 ± 19 mL; p < 0.001), and cardiac output (6.2 ± 1.7 L vs. 5.0 ± 1.6 L; p < 0.001) were higher and total peripheral vascular resistance (1465 ± 469 dyne·s·cm−5 vs. 1814 ± 524 dyne·s·cm−5; p < 0.001) was lower in the cases vs controls. Superimposed preeclampsia, preterm delivery before 34 weeks, abruptio placentae, HELLP Syndrome, fetal growth restriction, and perinatal death were more represented in the standard treatment group vs NO treated patients (81% vs 53%; p < 0.001). In particular, the standard treatment group showed 48% fetal growth restriction vs 34% in the NO treated group (p < 0.011). The Cox proportional-hazards regression showed a lower proportion of event-free pregnancies in controls on standard treatment (HR 2.6; 95% CI 2.0–3.5; p < 0.0001), and a prolongation of pregnancies in CH cases complicated by fetal growth restriction taking NO donors (HR 0.29; 95% CI 0.19–0.43; p = 0.0001). ConclusionsThe tailored treatment with NO donors and oral fluids of hypodynamic CH might have positive effects on the reduction or mitigations of adverse outcomes.

QS-ADN: quasi-supervised artifact disentanglement network for low-dose CT image denoising by local similarity among unpaired data

Deep learning has been successfully applied to low-dose CT (LDCT) image denoising for reducing potential radiation risk. However, the widely reported supervised LDCT denoising networks require a training set of paired images, which is expensive to obtain and cannot be perfectly simulated. Unsupervised learning utilizes unpaired data and is highly desirable for LDCT denoising. As an example, an artifact disentanglement network (ADN) relies on unpaired images and obviates the need for supervision but the results of artifact reduction are not as good as those through supervised learning. An important observation is that there is often hidden similarity among unpaired data that can be utilized. This paper introduces a new learning mode, called quasi-supervised learning, to empower ADN for LDCT image denoising. For every LDCT image, the best matched image is first found from an unpaired normal-dose CT (NDCT) dataset. Then, the matched pairs and the corresponding matching degree as prior information are used to construct and train our ADN-type network for LDCT denoising. The proposed method is different from (but compatible with) supervised and semi-supervised learning modes and can be easily implemented by modifying existing networks. The experimental results show that the method is competitive with state-of-the-art methods in terms of noise suppression and contextual fidelity. The code and working dataset are publicly available at https://github.com/ruanyuhui/ADN-QSDL.git.

Unsupervised Portrait Drawing Generation for Free Styles

Artistic portrait drawing (APDrawing) generation has seen progress in recent years. However, due to the naturally high scarcity and artistry, it is difficult to collect large-scale labeled and paired data and generally divide drawing styles into several specific recognized categories. Existing works suffer from the limited labeled data and naive manual division of drawing styles according to the corresponding artists. They cannot adapt to the actual situations, for example, a single artist might have multiple drawing styles and APDrawings from different artists might share similar styles. In this paper, we propose to use unlabeled and unpaired data and perform the task in an unsupervised manner. Without manual division of drawing styles, we take each portrait drawing as a unique style and introduce self-supervised feature learning to learn free styles for unlabeled portrait drawings. Besides, we devise a style bank and a decoupled cycle structure to take over two main considerations in the task: generation quality and style control. Extensive experiments show that our model is more adaptable to different style inputs than state-of-the-art methods.

The Increasing Level of Comfort and Decreasing Blood Pressure of Hypertensive Elderly with “TERTARIK” (Terapi Tertawa dengan Musik Rindik)

Changes in people's activity patterns during the pandemic caused discomfort responses that had implications for blood pressure fluctuations in hypertensive elderly. This phenomenon is significant because high blood pressure can cause complications and hospitalisation. This study aimed to determine the effect of TERTARIK on the comfort level and blood pressure of hypertensive elderly. A quantitative approach with a quasi-experimental pre-post test design with a control group was chosen. The sampling technique was purposive sampling with inclusion and exclusion criteria. The sample size obtained was 64 (32 per group), and 2 in the intervention group dropped out. Data was collected for four weeks with an intervention frequency of once a week. Comfort level and blood pressure variables are not normally distributed. These variables were tested with the Wilcoxon for paired data and the Mann-Whitney for unpaired data. The results obtained were that there was an influence of TERTARIK intervention on the comfort level (p-value 0,001), systolic blood pressure (p-value 0.001, and diastolic blood pressure (0.001) for hypertensive elderly. These results can guide nursing care for hypertensive elderly, especially with the complementary approach.

Single-shot image restoration via a model-enhanced network with unpaired supervision in an optical sparse aperture system.

We propose a model-enhanced network with unpaired single-shot data for solving the imaging blur problem of an optical sparse aperture (OSA) system. With only one degraded image captured from the system and one "arbitrarily" selected unpaired clear image, the cascaded neural network is iteratively trained for denoising and restoration. With the computational image degradation model enhancement, our method is able to improve contrast, restore blur, and suppress noise of degraded images in simulation and experiment. It can achieve better restoration performance with fewer priors than other algorithms. The easy selectivity of unpaired clear images and the non-strict requirement of a custom kernel make it suitable and applicable for single-shot image restoration of any OSA system.

Magnetic particle imaging deblurring with dual contrastive learning and adversarial framework

Magnetic particle imaging (MPI) is an emerging medical imaging technique that has high sensitivity, contrast, and excellent depth penetration. In MPI, x-space is a reconstruction method that transforms the measured voltages into particle concentrations. The reconstructed native image can be modeled as a convolution of the magnetic particle concentration with a point-spread function (PSF). The PSF is one of the important parameters in deconvolution. However, accurately measuring or modeling the PSF in the hardware used for deconvolution is challenging due to the various environment and magnetic particle relaxation. The inaccurate PSF estimation may lead to the loss of the content structure of the MPI image, especially in low gradient fields. In this study, we developed a Dual Adversarial Network (DAN) with patch-wise contrastive constraint to deblur the MPI image. This method can overcome the limitations of unpaired data in data acquisition scenarios and remove the blur around the boundary more effectively than the common deconvolution method. We evaluated the performance of the proposed DAN model on simulated and real data. Experimental results confirmed that our model performs favorably against the deconvolution method that is mainly used for deblurring the MPI image and other GAN-based deep learning models.

PUPC-GANs: A Novel Image Conversion Model using Modified CycleGANs in Healthcare

Introduction: Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) both have their areas of specialty in the medical imaging world. MRI is considered to be a safer modality as it exploits the magnetic properties of the hydrogen nucleus. Whereas a CT scan uses multiple X-rays, which is known to contribute to carcinogenesis and is associated with affecting the patient's health. Methods: In scenarios, such as radiation therapy, where both MRI and CT are required for medical treatment, a unique approach to getting both scans would be to obtain MRI and generate a CT scan from it. Current deep learning methods for MRI to CT synthesis purely use either paired data or unpaired data. Models trained with paired data suffer due to a lack of availability of wellaligned data. Results: Training with unpaired data might generate visually realistic images, although it still does not guarantee good accuracy. To overcome this, we proposed a new model called PUPCGANs (Paired Unpaired CycleGANs), based on CycleGANs (Cycle-Consistent Adversarial Networks). Conclusion: This model is capable of learning transformations utilizing both paired and unpaired data. To support this, a paired loss is introduced. Comparing MAE, MSE, NRMSE, PSNR, and SSIM metrics, PUPC-GANs outperform CycleGANs.

DC-cycleGAN: Bidirectional CT-to-MR synthesis from unpaired data.

Magnetic resonance (MR) and computer tomography (CT) images are two typical types of medical images that provide mutually-complementary information for accurate clinical diagnosis and treatment. However, obtaining both images may be limited due to some considerations such as cost, radiation dose and modality missing. Recently, medical image synthesis has aroused gaining research interest to cope with this limitation. In this paper, we propose a bidirectional learning model, denoted as dual contrast cycleGAN (DC-cycleGAN), to synthesize medical images from unpaired data. Specifically, a dual contrast loss is introduced into the discriminators to indirectly build constraints between real source and synthetic images by taking advantage of samples from the source domain as negative samples and enforce the synthetic images to fall far away from the source domain. In addition, cross-entropy and structural similarity index (SSIM) are integrated into the DC-cycleGAN in order to consider both the luminance and structure of samples when synthesizing images. The experimental results indicate that DC-cycleGAN is able to produce promising results as compared with other cycleGAN-based medical image synthesis methods such as cycleGAN, RegGAN, DualGAN, and NiceGAN. Code is available at https://github.com/JiayuanWang-JW/DC-cycleGAN.

Unsupervised underwater image enhancement via content-style representation disentanglement

The absorption and scattering properties of the water medium cause various types of distortion in underwater images, which seriously affects the accuracy and effectiveness of subsequent processing. The application of supervised learning algorithms in underwater image enhancement is limited by the difficulty of obtaining a large number of underwater paired images in practical applications. As a solution, we propose an unsupervised representation disentanglement based underwater image enhancement method (URD-UIE). URD-UIE disentangles content information (e.g., texture, semantics) and style information (e.g., chromatic aberration, blur, noise, and clarity) from underwater images and then employs the disentangled information to generate the target distortion-free image. Our proposed method URD-UIE adopts an unsupervised cycle-consistent adversarial translation architecture and combines multiple loss functions to impose specific constraints on the output results of each module to ensure the structural consistency of underwater images before and after enhancement. The experimental results demonstrate that the URD-UIE technique effectively enhances the quality of underwater images when training with unpaired data, resulting in a significant improvement in the performance of the standard model for underwater object detection and semantic segmentation.

Unsupervised dual-domain disentangled network for removal of rigid motion artifacts in MRI

Motion artifacts in magnetic resonance imaging (MRI) have always been a serious issue because they can affect subsequent diagnosis and treatment. Supervised deep learning methods have been investigated for the removal of motion artifacts; however, they require paired data that are difficult to obtain in clinical settings. Although unsupervised methods are widely proposed to fully use clinical unpaired data, they generally focus on anatomical structures generated by the spatial domain while ignoring phase error (deviations or inaccuracies in phase information that are possibly caused by rigid motion artifacts during image acquisition) provided by the frequency domain. In this study, a 2D unsupervised deep learning method named unsupervised disentangled dual-domain network (UDDN) was proposed to effectively disentangle and remove unwanted rigid motion artifacts from images. In UDDN, a dual-domain encoding module was presented to capture different types of information from the spatial and frequency domains to enrich the information. Moreover, a cross-domain attention fusion module was proposed to effectively fuse information from different domains, reduce information redundancy, and improve the performance of motion artifact removal. UDDN was validated on a publicly available dataset and a clinical dataset. Qualitative and quantitative experimental results showed that our method could effectively remove motion artifacts and reconstruct image details. Moreover, the performance of UDDN surpasses that of several state-of-the-art unsupervised methods and is comparable with that of the supervised method. Therefore, our method has great potential for clinical application in MRI, such as real-time removal of rigid motion artifacts.

CUI-Net: a correcting uneven illumination net for low-light image enhancement

Uneven lighting conditions often occur during real-life photography, such as images taken at night that may have both low-light dark areas and high-light overexposed areas. Traditional algorithms for enhancing low-light areas also increase the brightness of overexposed areas, affecting the overall visual effect of the image. Therefore, it is important to achieve differentiated enhancement of low-light and high-light areas. In this paper, we propose a network called correcting uneven illumination network (CUI-Net) with sparse attention transformer and convolutional neural network (CNN) to better extract low-light features by constraining high-light features. Specifically, CUI-Net consists of two main modules: a low-light enhancement module and an auxiliary module. The enhancement module is a hybrid network that combines the advantages of CNN and Transformer network, which can alleviate uneven lighting problems and enhance local details better. The auxiliary module is used to converge the enhancement results of multiple enhancement modules during the training phase, so that only one enhancement module is needed during the testing phase to speed up inference. Furthermore, zero-shot learning is used in this paper to adapt to complex uneven lighting environments without requiring paired or unpaired training data. Finally, to validate the effectiveness of the algorithm, we tested it on multiple datasets of different types, and the algorithm showed stable performance, demonstrating its good robustness. Additionally, by applying this algorithm to practical visual tasks such as object detection, face detection, and semantic segmentation, and comparing it with other state-of-the-art low-light image enhancement algorithms, we have demonstrated its practicality and advantages.

Synthesis of diffusion-weighted MRI scalar maps from FLAIR volumes using generative adversarial networks.

Acquisition and pre-processing pipelines for diffusion-weighted imaging (DWI) volumes are resource- and time-consuming. Generating synthetic DWI scalar maps from commonly acquired brain MRI sequences such as fluid-attenuated inversion recovery (FLAIR) could be useful for supplementing datasets. In this work we design and compare GAN-based image translation models for generating DWI scalar maps from FLAIR MRI for the first time. We evaluate a pix2pix model, two modified CycleGANs using paired and unpaired data, and a convolutional autoencoder in synthesizing DWI fractional anisotropy (FA) and mean diffusivity (MD) from whole FLAIR volumes. In total, 420 FLAIR and DWI volumes (11,957 images) from multi-center dementia and vascular disease cohorts were used for training/testing. Generated images were evaluated using two groups of metrics: (1) human perception metrics including peak signal-to-noise ratio (PSNR) and structural similarity (SSIM), (2) structural metrics including a newly proposed histogram similarity (Hist-KL) metric and mean squared error (MSE). Pix2pix demonstrated the best performance both quantitatively and qualitatively with mean PSNR, SSIM, and MSE metrics of 23.41 dB, 0.8, 0.004, respectively for MD generation, and 24.05 dB, 0.78, 0.004, respectively for FA generation. The new histogram similarity metric demonstrated sensitivity to differences in fine details between generated and real images with mean pix2pix MD and FA Hist-KL metrics of 11.73 and 3.74, respectively. Detailed analysis of clinically relevant regions of white matter (WM) and gray matter (GM) in the pix2pix images also showed strong significant (p < 0.001) correlations between real and synthetic FA values in both tissue types (R = 0.714 for GM, R = 0.877 for WM). Our results show that pix2pix's FA and MD models had significantly better structural similarity of tissue structures and fine details than other models, including WM tracts and CSF spaces, between real and generated images. Regional analysis of synthetic volumes showed that synthetic DWI images can not only be used to supplement clinical datasets, but demonstrates potential utility in bypassing or correcting registration in data pre-processing.

Progressively Unsupervised Generative Attentional Networks with Adaptive Layer-Instance Normalization for Image-to-Image Translation.

Unsupervised image-to-image translation has received considerable attention due to the recent remarkable advancements in generative adversarial networks (GANs). In image-to-image translation, state-of-the-art methods use unpaired image data to learn mappings between the source and target domains. However, despite their promising results, existing approaches often fail in challenging conditions, particularly when images have various target instances and a translation task involves significant transitions in shape and visual artifacts when translating low-level information rather than high-level semantics. To tackle the problem, we propose a novel framework called Progressive Unsupervised Generative Attentional Networks with Adaptive Layer-Instance Normalization (PRO-U-GAT-IT) for the unsupervised image-to-image translation task. In contrast to existing attention-based models that fail to handle geometric transitions between the source and target domains, our model can translate images requiring extensive and holistic changes in shape. Experimental results show the superiority of the proposed approach compared to the existing state-of-the-art models on different datasets.

Attentive continuous generative self-training for unsupervised domain adaptive medical image translation.

Self-training is an important class of unsupervised domain adaptation (UDA) approaches that are used to mitigate the problem of domain shift, when applying knowledge learned from a labeled source domain to unlabeled and heterogeneous target domains. While self-training-based UDA has shown considerable promise on discriminative tasks, including classification and segmentation, through reliable pseudo-label filtering based on the maximum softmax probability, there is a paucity of prior work on self-training-based UDA for generative tasks, including image modality translation. To fill this gap, in this work, we seek to develop a generative self-training (GST) framework for domain adaptive image translation with continuous value prediction and regression objectives. Specifically, we quantify both aleatoric and epistemic uncertainties within our GST using variational Bayes learning to measure the reliability of synthesized data. We also introduce a self-attention scheme that de-emphasizes the background region to prevent it from dominating the training process. The adaptation is then carried out by an alternating optimization scheme with target domain supervision that focuses attention on the regions with reliable pseudo-labels. We evaluated our framework on two cross-scanner/center, inter-subject translation tasks, including tagged-to-cine magnetic resonance (MR) image translation and T1-weighted MR-to-fractional anisotropy translation. Extensive validations with unpaired target domain data showed that our GST yielded superior synthesis performance in comparison to adversarial training UDA methods.

Development of Animated Videos Using Diary’s Virtual Reality as a Medium Of Distraction on The Anxiety

Background: Dental anxiety refers to anxiety about dental treatment procedures. Dental anxiety is common among pediatric patients because dental treatment procedures are usually a new experience. Therapy and desperate nonpharmacological techniques, which have already become widely used, can alleviate dental anxiety. Virtual reality is one of the various media considered to distract children during dental and oral treatment. Produce a viable Diary's Virtual Reality (VR) media model and its application effectively lowers the anxiety level of child patients upon deciduous tooth extraction action.&#x0D; Method: Experimental design pretest-posttest method with control group design. There were 64 respondents which were divided into; an intervention group given the treatment of watching Diary’s Virtual Reality video and a control group given the treatment of watching video using mobile phones. To assess the child’s level of anxiety when the deciduous tooth is extracted, an MDAS query was used. The test results data of the model were tested using wilcoxon paired test and a Mann-Whitney paired test.&#x0D; Result: The Expert Validation Test of Dary’s VR media earned an average feasibility score of 97,5% with excellent categories with p-value=0,000. The paired data effectiveness test results showed an intervention group p-value=0,000 and a control group=0,010. The test results of the effectiveness of unpaired variable data show that the p-value between the intervention group and the control group is p=0,000. The Diary’s VR model effectively lowers the anxiety level of child patients.&#x0D; Conclusion: Giving Intervention Diary’s VR lowers the anxiety level of child patients significantly compared to the control group.

AtacAnnoR: a reference-based annotation tool for single cell ATAC-seq data.

Here, we present AtacAnnoR, a two-round annotation method for scATAC-seq data using well-annotated scRNA-seq data as reference. We evaluate AtacAnnoR's performance against six competing methods on 11 benchmark datasets. Our results show that AtacAnnoR achieves the highest mean accuracy and the highest mean balanced accuracy and performs particularly well when unpaired scRNA-seq data are used as the reference. Furthermore, AtacAnnoR implements a 'Combine and Discard' strategy to further improve annotation accuracy when annotations of multiple references are available. AtacAnnoR has been implemented in an R package and can be directly integrated into currently popular scATAC-seq analysis pipelines.