We present a novel framework for high-fidelity novel view synthesis (NVS) from sparse images, addressing key limitations in recent feed-forward 3D Gaussian Splatting (3DGS) methods built on Vision Transformer (ViT) backbones. While ViT-based pipelines offer strong geometric priors, they are often constrained by low-resolution inputs due to computational costs. Moreover, existing generative enhancement methods tend to be 3D-agnostic, resulting in inconsistent structures across views, especially in unseen regions. To overcome these challenges, we design a Dual-Domain Detail Perception Module, which enables handling high-resolution images without being limited by the ViT backbone, and endows Gaussians with additional features to store high-frequency details. We develop a feature-guided diffusion network, which can preserve high-frequency details during the restoration process. We introduce a unified training strategy that enables joint optimization of the ViT-based geometric backbone and the diffusion-based refinement module. Experiments demonstrate that our method can maintain superior generation quality across multiple datasets.

Text-to-image (T2I) diffusion models excel at generating photorealistic images but often fail to render accurate spatial relationships. We identify two core issues underlying this common failure: 1) the ambiguous nature of data concerning spatial relationships in existing datasets, and 2) the inability of current text encoders to accurately interpret the spatial semantics of input descriptions. We propose CoMPaSS, a versatile framework that enhances spatial understanding in T2I models. It first addresses data ambiguity with the Spatial Constraints-Oriented Pairing (SCOP) data engine, which curates spatially-accurate training data via principled constraints. To leverage these priors, CoMPaSS also introduces the Token ENcoding ORdering (TENOR) module, which preserves crucial token ordering information lost by text encoders, thereby reinforcing the prompt's linguistic structure. Extensive experiments on four popular T2I models (UNet and MMDiT-based) show CoMPaSS sets a new state of the art on key spatial benchmarks, with substantial relative gains on VISOR (+98%), T2I-CompBench Spatial (+67%), and GenEval Position (+131%). Code is available at https://github.com/blurgyy/CoMPaSS.

We introduce BokehDiff, a novel lens blur rendering method that achieves physically accurate and visually appealing outcomes, with the help of generative diffusion prior. Previous methods are bounded by the accuracy of depth estimation, generating artifacts in depth discontinuities. Our method employs a physics-inspired self-attention module that aligns with the image formation process, incorporating depth-dependent circle of confusion constraint and self-occlusion effects. We adapt the diffusion model to the one-step inference scheme without introducing additional noise, and achieve results of high quality and fidelity. To address the lack of scalable paired data, we propose to synthesize photorealistic foregrounds with transparency with diffusion models, balancing authenticity and scene diversity.

Constructing an efficient and deployable anomaly detection system requires achieving high accuracy, low latency, and reliability. Existing methods either spend considerable time extracting rich spatio-temporal features to enhance anomaly detection performance, or blindly integrate multi-source features to boost accuracy, often neglecting the reliability of feature aggregation. The trade-off between the three objectives must be carefully considered when developing the model. To address these challenges, we introduce a novel Spatio-Temporal Weighted Graph Reasoning Learning (STWGRL) framework for multivariate time-series anomaly detection. Specifically, we propose a series-denoising receptance-weighted key value (D-RWKV) module to efficiently capture and model expressive long-term sequence information through a linear scaling mechanism. D-RWKV ensures compatibility by alleviating the memory bottleneck and enabling parallelized training. Furthermore, we design a targeted-awareness graph adaptive aggregation (TaGAA) module to learn the directed graph and adaptively enhance the signal’s intrinsic characteristics. Two graph-constraint losses are employed to strengthen the consistency and sparsity of the learned graphs. Experimental results on multiple benchmark tasks clearly demonstrate the effectiveness of the proposed framework. STWGRL achieves more accurate scores than most baselines, while containing fewer than 10K parameters.

Instance segmentation is data-hungry, and as model capacity increases, data scale becomes crucial for improving the accuracy. Most instance segmentation datasets today require costly manual annotation, limiting their data scale. Models trained on such data are prone to overfitting on the training set, especially for those rare categories. While recent works have delved into exploiting generative models to create synthetic datasets for data augmentation, these approaches do not efficiently harness the full potential of generative models. To address these issues, we introduce a more efficient strategy to construct generative datasets for data augmentation, termed DiverGen. Firstly, we provide an explanation of the role of generative data from the perspective of distribution discrepancy. We investigate the impact of different data on the distribution learned by the model. We argue that generative data can expand the data distribution that the model can learn, thus mitigating overfitting. Additionally, we find that the diversity of generative data is crucial for improving model performance and enhance it through various strategies, including category diversity, prompt diversity, and generative model diversity. With these strategies, we can scale the data to millions while maintaining the trend of model performance improvement. On the LVIS dataset, DiverGen significantly outperforms the strong model X-Paste, achieving +1.1 box AP and +1.1 mask AP across all categories, and +1.9 box AP and +2.5 mask AP for rare categories. Our codes are available at https://github.com/aim-uofa/DiverGen.

Fusing multi-modal radiology and pathology data with complementary information can improve the accuracy of tumor typing. However, collecting pathology data is difficult since it is high-cost and sometimes only obtainable after the surgery, which limits the application of multi-modal methods in diagnosis. To address this problem, we propose comprehensively learning multi-modal radiology-pathology data in training, and only using uni-modal radiology data in testing. Concretely, a Memory-aware Hetero-modal Distillation Network (MHD-Net) is proposed, which can distill well-learned multi-modal knowledge with the assistance of memory from the teacher to the student. In the teacher, to tackle the challenge in hetero-modal feature fusion, we propose a novel spatial-differentiated hetero-modal fusion module (SHFM) that models spatial-specific tumor information correlations across modalities. As only radiology data is accessible to the student, we store pathology features in the proposed contrast-boosted typing memory module (CTMM) that achieves type-wise memory updating and stage-wise contrastive memory boosting to ensure the effectiveness and generalization of memory items. In the student, to improve the cross-modal distillation, we propose a multi-stage memory-aware distillation (MMD) scheme that reads memory-aware pathology features from CTMM to remedy missing modal-specific information. Furthermore, we construct a Radiology-Pathology Thymic Epithelial Tumor (RPTET) dataset containing paired CT and WSI images with annotations. Experiments on the RPTET and CPTAC-LUAD datasets demonstrate that MHD-Net significantly improves tumor typing and outperforms existing multi-modal methods on missing modality situations.

Multi-modal prompt learning is a high-performance and cost-effective learning paradigm, which learns text as well as image prompts to tune pre-trained vision-language (V-L) models like CLIP for adapting multiple downstream tasks. However, recent methods typically treat text and image prompts as independent components without considering the dependency between prompts. Moreover, extending multi-modal prompt learning into the medical field poses challenges due to a significant gap between general- and medical-domain data. To this end, we propose a Multi-modal Collaborative Prompt Learning (MCPL) pipeline to tune a frozen V-L model for aligning medical text-image representations, thereby achieving medical downstream tasks. We first construct the anatomy-pathology (AP) prompt for multi-modal prompting jointly with text and image prompts. The AP prompt introduces instance-level anatomy and pathology information, thereby making a V-L model better comprehend medical reports and images. Next, we propose graph-guided prompt collaboration module (GPCM), which explicitly establishes multi-way couplings between the AP, text, and image prompts, enabling collaborative multi-modal prompt producing and updating for more effective prompting. Finally, we develop a novel prompt configuration scheme, which attaches the AP prompt to the query and key, and the text/image prompt to the value in self-attention layers for improving the interpretability of multi-modal prompts. Extensive experiments on numerous medical classification and object detection datasets show that the proposed pipeline achieves excellent effectiveness and generalization. Compared with state-of-the-art prompt learning methods, MCPL provides a more reliable multi-modal prompt paradigm for reducing tuning costs of V-L models on medical downstream tasks. Our code: https://github.com/CUHK-AIM-Group/MCPL.

Taking advantage of multi-modal radiology-pathology data with complementary clinical information for cancer grading is helpful for doctors to improve diagnosis efficiency and accuracy. However, radiology and pathology data have distinct acquisition difficulties and costs, which leads to incomplete-modality data being common in applications. In this work, we propose a Memory- and Gradient-guided Incomplete Modal-modal Learning (MGIML) framework for cancer grading with incomplete radiology-pathology data. Firstly, to remedy missing-modality information, we propose a Memory-driven Hetero-modality Complement (MH-Complete) scheme, which constructs modal-specific memory banks constrained by a coarse-grained memory boosting (CMB) loss to record generic radiology and pathology feature patterns, and develops a cross-modal memory reading strategy enhanced by a fine-grained memory consistency (FMC) loss to take missing-modality information from well-stored memories. Secondly, as gradient conflicts exist between missing-modality situations, we propose a Rotation-driven Gradient Homogenization (RG-Homogenize) scheme, which estimates instance-specific rotation matrices to smoothly change the feature-level gradient directions, and computes confidence-guided homogenization weights to dynamically balance gradient magnitudes. By simultaneously mitigating gradient direction and magnitude conflicts, this scheme well avoids the negative transfer and optimization imbalance problems. Extensive experiments on CPTAC-UCEC and CPTAC-PDA datasets show that the proposed MGIML framework performs favorably against state-of-the-art multi-modal methods on missing-modality situations.

To obtain clean images with salient structures from noisy observations, a growing trend in current denoising studies is to seek the help of additional guidance images with high signal-to-noise ratios, which are often acquired in different spectral bands such as near infrared. Although previous guided denoising methods basically require the input images to be well-aligned, a more common way to capture the paired noisy target and guidance images is to exploit a stereo camera system. However, current studies on cross-spectral stereo matching cannot fully guarantee the pixel-level registration accuracy, and rarely consider the case of noise contamination. In this work, for the first time, we propose a guided denoising framework for cross-spectral stereo images. Instead of aligning the input images via conventional stereo matching, we aggregate structures from the guidance image to estimate a clean structure map for the noisy target image, which is then used to regress the final denoising result with a spatially variant linear representation model. Based on this, we design a neural network, called as SANet, to complete the entire guided denoising process. Experimental results show that, our SANet can effectively transfer structures from an unaligned guidance image to the restoration result, and outperforms state-of-the-art denoisers on various stereo image datasets. Besides, our structure aggregation strategy also shows its potential to handle other unaligned guided restoration tasks such as super-resolution and deblurring. The source code is available at https://github.com/lustrouselixir/SANet.

Guided depth completion aims to recover dense depth maps by propagating depth information from the given pixels to the remaining ones under the guidance of RGB images. However, most of the existing methods achieve this using a large number of iterative refinements or stacking repetitive blocks. Due to the limited receptive field of conventional convolution, the generalizability with respect to different sparsity levels of input depth maps is impeded. To tackle these problems, we propose a feature point cloud aggregation framework to directly propagate 3D depth information between the given points and the missing ones. We extract 2D feature map from images and transform the sparse depth map to point cloud to extract sparse 3D features. By regarding the extracted features as two sets of feature point clouds, the depth information for a target location can be reconstructed by aggregating adjacent sparse 3D features from the known points using cross attention. Based on this, we design a neural network, called as PointDC, to complete the entire depth information reconstruction process. Experimental results show that, our PointDC achieves superior or competitive results on the KITTI benchmark and NYUv2 dataset. In addition, the proposed PointDC demonstrates its higher generalizability to different sparsity levels of the input depth maps and cross-dataset evaluation.

Immune checkpoint inhibitors (ICI) show high efficiency in a small fraction of advanced gastric cancer (GC). However, personalized immune subtypes have not been developed for the prediction of ICI efficiency in GC. Herein, we identified Pan-Immune Activation Module (PIAM), a curated gene expression profile (GEP) representing the co-infiltration of multiple immune cell types in tumor microenvironment of GC, which was associated with high expression of immunosuppressive molecules such as PD-1 and CTLA-4. We also identified Pan-Immune Dysfunction Genes (PIDG), a conservative PIAM-derivated GEP indicating the dysfunction of immune cell cooperation, which was associated with upregulation of metastatic programs (extracellular matrix receptor interaction, TGF-β signaling, epithelial-mesenchymal transition and calcium signaling) but downregulation of proliferative signalings (MYC targets, E2F targets, mTORC1 signaling, and DNA replication and repair). Moreover, we developed 'GSClassifier', an ensemble toolkit based on top scoring pairs and extreme gradient boosting, for population-based modeling and personalized identification of GEP subtypes. With PIAM and PIDG, we developed four Pan-immune Activation and Dysfunction (PAD) subtypes and a GSClassifier model 'PAD for individual' with high accuracy in predicting response to pembrolizumab (anti-PD-1) in advance GC (AUC = 0.833). Intriguingly, PAD-II (PIAMhighPIDGlow) displayed the highest objective response rate (60.0%) compared with other subtypes (PAD-I, PIAMhighPIDGhigh, 0%; PAD-III, PIAMlowPIDGhigh, 0%; PAD-IV, PIAMlowPIDGlow, 17.6%; P = 0.003), which was further validated in the metastatic urothelial cancer cohort treated with atezolizumab (anti-PD-L1) (P = 0.018). In all, we provided 'GSClassifier' as a refined computational framework for GEP-based stratification and PAD subtypes as a promising strategy for exploring ICI responders in GC. Metastatic pathways could be potential targets for GC patients with high immune infiltration but resistance to ICI therapy.

Compared to color images captured by conventional RGB cameras, monochrome (mono) images usually have higher signal-to-noise ratios (SNR) and richer textures due to the lack of color filter arrays in mono cameras. Therefore, using a mono-color stereo dual-camera system, we can integrate the lightness information of target monochrome images with the color information of guidance RGB images to accomplish image enhancement in a colorization manner. In this work, based on two assumptions, we introduce a novel probabilistic-concept guided colorization framework. First, adjacent contents with similar luminance are likely to have similar colors. By lightness matching, we can utilize colors of the matched pixels to estimate the target color value. Second, by matching multiple pixels from the guidance image, if more of these matched pixels have similar luminance values to the target one, we can estimate colors with more confidence. Based on the statistical distribution of multiple matching results, we retain the reliable color estimates as initial dense scribbles and then propagate them to the rest of the mono image. However, for a target pixel, the color information provided by its matching results is quite redundant. Hence, we introduce a patch sampling strategy to accelerate the colorization process. Based on the analysis of the posteriori probability distribution of the sampling results, we can use much fewer matches for color estimation and reliability assessment. To alleviate incorrect color propagation in the sparsely scribbled regions, we generate extra color seeds according to the existed scribbles to guide the propagation process. Experimental results show that, our algorithm can efficiently and effectively restore color images with higher SNR and richer details from the mono-color image pairs, and achieves good performance in solving the color bleeding problem.

Despite the tremendous advances in denoising techniques, it's still challenging to restore a clean image with salient structures based on one noisy observation, especially at high noise levels. In this work, we propose a frequency-domain guided denoising algorithm to conduct denoising with the help of a well-aligned guidance image. Thanks to their structural correlations, the frequency characteristics of the guidance image can indicate whether the frequency coefficients of the noisy target image are contributed by noise or textures. Therefore, the explicit frequency decomposition enables our denoising model to avoid over-smoothing detailed contents. However, as two input images are usually captured in different fields, their structures are not always consistent. Therefore, we model guided denoising with an optimization problem which considers both the representation model of the guidance image and the fidelity to the noisy target. Further, we design a convolutional neural network, called as FGDNet, to explore the optimal solution. Due to the visual masking phenomenon, human eyes are sensitive to noise in the flat areas, but may not perceive noise around edges or textures. Therefore, we expect to remove as much noise as possible to guarantee the spatial smoothness of flat contents, while also preserving high-frequency structures. Through frequency decomposition, our model can process the low-frequency and high-frequency contents separately. We also adopt a frequency-relevant loss function to train the network. Experimental results show that, compared with state-of-the-art guided and non-guided denoisers, our FGDNet achieves higher denoising accuracy and better visual quality in both flat and texture-rich regions.

Recent digital pathology workflows mainly focus on mono-modality histopathology image analysis. However, they ignore the complementarity between Haematoxylin & Eosin (H&E) and Immunohistochemically (IHC) stained images, which can provide comprehensive gold standard for cancer diagnosis. To resolve this issue, we propose a cross-boosted multi-target domain adaptation pipeline for multi-modality histopathology images, which contains Cross-frequency Style-auxiliary Translation Network (CSTN) and Dual Cross-boosted Segmentation Network (DCSN). Firstly, CSTN achieves the one-to-many translation from fluorescence microscopy images to H&E and IHC images for providing source domain training data. To generate images with realistic color and texture, Cross-frequency Feature Transfer Module (CFTM) is developed to pertinently restructure and normalize high-frequency content and low-frequency style features from different domains. Then, DCSN fulfills multi-target domain adaptive segmentation, where a dual-branch encoder is introduced, and Bidirectional Cross-domain Boosting Module (BCBM) is designed to implement cross-modality information complementation through bidirectional inter-domain collaboration. Finally, we establish Multi-modality Thymus Histopathology (MThH) dataset, which is the largest publicly available H&E and IHC image benchmark. Experiments on MThH dataset and several public datasets show that the proposed pipeline outperforms state-of-the-art methods on both histopathology image translation and segmentation.

Coronary optical coherence tomography (OCT) is a new high-resolution intravascular imaging technology that clearly depicts coronary artery stenosis and plaque information. Study of coronary OCT images is of significance in the diagnosis of coronary atherosclerotic heart disease (CAD). We introduce a new method based on the convolutional neural network (CNN) and an improved random walk (RW) algorithm for the recognition and segmentation of calcified, lipid and fibrotic plaque in coronary OCT images. First, we design CNN with three different depths (2, 4 or 6 convolutional layers) to perform the automatic recognition and select the optimal CNN model. Then, we device an improved RW algorithm. According to the gray-level distribution characteristics of coronary OCT images, the weights of intensity and texture term in the weight function of RW algorithm are adjusted by an adaptive weight. Finally, we apply mathematical morphology in combination with two RWs to accurately segment the plaque area. Compared with the ground truth of clinical segmentation results, the Jaccard similarity coefficient (JSC) of calcified and lipid plaque segmentation results is 0.864, the average symmetric contour distance (ASCD) is 0.375[Formula: see text]mm, the JSC and ASCD reliabilities are 88.33% and 92.50% respectively. The JSC of fibrotic plaque is 0.876, the ASCD is 0.349[Formula: see text]mm, the JSC and ASCD reliabilities are 90.83% and 95.83% respectively. In addition, the average segmentation time (AST) does not exceed 5 s. Reliable and significantly improved results have been achieved in this study. Compared with the CNN, traditional RW algorithm and other methods. The proposed method has the advantages of fast segmentation, high accuracy and reliability, and holds promise as an aid to doctors in the diagnosis of CAD.