Publications

2025

1. MedIA 2025

   
   

   

   ×

   Joint coil sensitivity and motion correction in parallel MRI with a self-calibrating score-based diffusion model

   Lixuan Chen*,  Xuanyu Tian*, Jiangjie Wu, Ruimin Feng, Guoyan Lao, Yuyao Zhang, Hongen Liao, and Hongjiang Wei

   Medical Image Analysis 2025

   PDF Code
2. BSPC 2025

   
   

   

   ×

   Self-supervised denoising for high-dimensional magnetic resonance image

   Xuanyu Tian, Jiangjie Wu, Guoyan Lao, Chenhe Du, Changhao Jiang, Yanbin Li, Jeff L Zhang, Hongjiang Wei, and Yuyao Zhang

   Biomedical Signal Processing and Control 2025
3. AAAI 2025

   
   

   

   ×

   Unsupervised Self-Prior Embedding Neural Representation for Iterative Sparse-View CT Reconstruction

   Xuanyu Tian, Lixuan Chen, Qing Wu, Chenhe Du, Jingjing Shi, Hongjiang Wei, and Yuyao Zhang

   Proceedings of the AAAI Conference on Artificial Intelligence 2025

   Code
4. ICLR 2025

   
   

   Spotlight

   

   ×

   Moner: Motion Correction in Undersampled Radial MRI with Unsupervised Neural Representation

   Qing Wu*, Chenhe Du*,  Xuanyu Tian, Jingyi Yu, Yuyao Zhang, and Hongjiang Wei

   The 13th International Conference on Learning Representations 2025

   PDF

2024

1. MedIA 2024

   
   

   

   ×

   COLLATOR: Consistent spatial-temporal longitudinal atlas construction via implicit neural representation

   Lixuan Chen*,  Xuanyu Tian*, Jiangjie Wu, Guoyan Lao, Yuyao Zhang, and Hongjiang Wei

   Medical Image Analysis 2024

   Abs PDF Code

   Longitudinal brain atlases that present brain development trend along time, are essential tools for brain development studies. However, conventional methods construct these atlases by independently averaging brain images from different individuals at discrete time points. This approach could introduce temporal inconsistencies due to variations in ontogenetic trends among samples, potentially affecting accuracy of brain developmental characteristic analysis. In this paper, we propose an implicit neural representation (INR)-based framework to improve the temporal consistency in longitudinal atlases. We treat temporal inconsistency as a 4-dimensional (4D) image denoising task, where the data consists of 3D spatial information and 1D temporal progression. We formulate the longitudinal atlas as an implicit function of the spatial-temporal coordinates, allowing structural inconsistency over the time to be considered as 3D image noise along age. Inspired by recent self-supervised denoising methods (e.g. Noise2Noise), our approach learns the noise-free and temporally continuous implicit function from inconsistent longitudinal atlas data. Finally, the time-consistent longitudinal brain atlas can be reconstructed by evaluating the denoised 4D INR function at critical brain developing time points. We evaluate our approach on three longitudinal brain atlases of different MRI modalities, demonstrating that our method significantly improves temporal consistency while accurately preserving brain structures. Additionally, the continuous functions generated by our method enable the creation of 4D atlases with higher spatial and temporal resolution.

2. MICCAI 2024

   
   

   

   ×

   Zero-Shot Low-Field MRI Enhancement via Denoising Diffusion Driven Neural Representation

   Xiyue Lin*, Chenhe Du*, Qing Wu,  Xuanyu Tian, Jingyi Yu, Yuyao Zhang, and Hongjiang Wei

   International Conference on Medical Image Computing and Computer-Assisted Intervention 2024
3. IEEE TCI 2024

   
   

   

   ×

   Zero-Shot Image Denoising for High-Resolution Electron Microscopy

   Xuanyu Tian, Zhuoya Dong, Xiyue Lin, Yue Gao, Hongjiang Wei, Yanhang Ma, Jingyi Yu, and Yuyao Zhang

   IEEE Transactions on Computational Imaging 2024

   Abs PDF Code

   High-resolution electron microscopy (HREM) imaging technique is a powerful tool for directly visualizing a broad range of materials in real-space. However, it faces challenges in denoising due to ultra-low signal-to-noise ratio (SNR) and scarce data availability. In this work, we propose Noise2SR, a zero- shot self-supervised learning (ZS-SSL) denoising framework for HREM. Within our framework, we propose a super-resolution (SR) based self-supervised training strategy, incorporating the Random Sub-sampler module. The Random Sub-sampler is designed to generate approximate infinite noisy pairs from a single noisy image, serving as an effective data augmentation in zero-shot denoising. Noise2SR trains the network with paired noisy images of different resolutions, which is conducted via SR strategy. The SR-based training facilitates the network adopting more pixels for supervision, and the random sub-sampling helps compel the network to learn continuous signals enhancing the robustness. Meanwhile, we mitigate the uncertainty caused by random-sampling by adopting minimum mean squared error (MMSE) estimation for the denoised results. With the distinctive integration of training strategy and proposed designs, Noise2SR can achieve superior denoising performance using a single noisy HREM image. We evaluate the performance of Noise2SR in both simulated and real HREM denoising tasks. It outperforms state- of-the-art ZS-SSL methods and achieves comparable denoising performance with supervised methods. The success of Noise2SR suggests its potential for improving the SNR of images in material imaging domains.

4. Under Review

   
   

   

   ×

   DPER: Diffusion Prior Driven Neural Representation for Limited Angle and Sparse View CT Reconstruction

   Chenhe Du, Xiyue Lin, Qing Wu,  Xuanyu Tian, Ying Su, Zhe Luo, Rui Zheng, Yang Chen, Hongjiang Wei, S Kevin Zhou, Jingyi Yu, and Yuyao Zhang

   ArXiv preprint 2024

   Abs PDF

   Limited-angle and sparse-view computed tomography (LACT and SVCT) are crucial for expanding the scope of X-ray CT applications. However, they face challenges due to incomplete data acquisition, resulting in diverse artifacts in the reconstructed CT images. Emerging implicit neural representation (INR) techniques, such as NeRF, NeAT, and NeRP, have shown promise in under-determined CT imaging reconstruction tasks. However, the unsupervised nature of INR architecture imposes limited constraints on the solution space, particularly for the highly ill-posed reconstruction task posed by LACT and ultra-SVCT. In this study, we introduce the Diffusion Prior Driven Neural Representation (DPER), an advanced unsupervised framework designed to address the exceptionally ill-posed CT reconstruction inverse problems. DPER adopts the Half Quadratic Splitting (HQS) algorithm to decompose the inverse problem into data fidelity and distribution prior sub-problems. The two sub-problems are respectively addressed by INR reconstruction scheme and pre-trained score-based diffusion model. This combination first injects the implicit image local consistency prior from INR. Additionally, it effectively augments the feasibility of the solution space for the inverse problem through the generative diffusion model, resulting in increased stability and precision in the solutions. We conduct comprehensive experiments to evaluate the performance of DPER on LACT and ultra-SVCT reconstruction with two public datasets (AAPM and LIDC), an in-house clinical COVID-19 dataset and a public raw projection dataset created by Mayo Clinic. The results show that our method outperforms the state-of-the-art reconstruction methods on in-domain datasets, while achieving significant performance improvements on out-of-domain (OOD) datasets.

5. ISBI 2024

   
   

   

   ×

   Reconstructing Knee CT Volumes from Biplanar X-Rays Via Self-Supervised Neural Field

   Shuyang Lai,  Xuanyu Tian, Qing Wu, Chenhe Du, Xiaojun Xu, Hongjiang Wei, Xiaojun Guan, and Yuyao Zhang

   IEEE International Symposium on Biomedical Imaging (ISBI) 2024

2023

1. ISBI 2023

   
   

   

   ×

   Self-supervised High-Dimensional Magnetic Resonance Image Denoising Using Super-resolved Single Noisy Image

   Changhao Jiang*,  Xuanyu Tian*, Yanbin Li, Jiangjie Wu, Xin Mu, Lei Zhang, and Yuyao Zhang

   IEEE International Symposium on Biomedical Imaging 2023

   Abs

   Denoising of magnetic resonance image (MRI) is a critical step in MRI image processing and analysis. With the advantage of not requiring paired noisy-clean images for training, self-supervised denoising methods are emerging as competitive alternatives to supervised denoising methods in MRI denoising. However, current self-supervised image denoising methods are not effective enough for MRI. In this work, we propose Noise2SR-M (N2SR-M), a self-supervised denoising method for MR images, which is more efficient for high-dimensional MR images. N2SR-M is designed for training with paired noisy data of different sizes divided from a single high-dimensional noisy input image. Our N2SR-M model is able to utilize the redundant information from the additional image dimension to generate noisy image pairs for the denoising task. With the combination of additional dimension constraint and the effectiveness of SR method based training image pair generation, our model is more efficient for denoising high-dimensional MR images. The quantitative and qualitative improvements in blood oxygenation level dependent (BOLD) imaging denoising task demonstrate that N2SRM successfully restores detailed image contents and removes tiny structural noise and artifacts from noise-corrupted high-dimensional MRI. Moreover, the denoised BOLD image also induces more efficient R2* image computation.

2022

1. MICCAI 2022

   
   

   

   ×

   Noise2SR: Learning to Denoise from Super-Resolved Single Noisy Fluorescence Image

   Xuanyu Tian, Qing Wu, Hongjiang Wei, and Yuyao Zhang

   Medical Image Computing and Computer Assisted Intervention 2022

   Abs HTML PDF

   Fluorescence microscopy is a key driver to promote discoveries of biomedical research. However, with the limitation of microscope hardware and characteristics of the observed samples, the fluorescence microscopy images are susceptible to noise. Recently, a few self-supervised deep learning (DL) denoising methods have been proposed. However, the training efficiency and denoising performance of existing methods are relatively low in real scene noise removal. To address this issue, this paper proposed self-supervised image denoising method Noise2SR (N2SR) to train a simple and effective image denoising model based on single noisy observation. Our Noise2SR denoising model is designed for training with paired noisy images of different dimensions. Benefiting from this training strategy, Noise2SR is more efficiently self-supervised and able to restore more image details from a single noisy observation. Experimental results of simulated noise and real microscopy noise removal show that Noise2SR outperforms two blind-spot based self-supervised deep learning image denoising methods. We envision that Noise2SR has the potential to improve more other kind of scientific imaging quality.