As a cost-effective and robust technology, automotive radar has seen steady improvement during the last years. Radio frequency (RF) images, serving as a radar data format with rich semantic information, have attracted considerable interest in radar object detection. Previous RF-based models heavily rely on convolutional neural networks, leading to the high computational cost. To solve this problem, we propose a model called Mask-RadarNet to fully utilize the hierarchical semantic features from the RF image sequences. Mask-RadarNet exploits the combination of interleaved convolution and attention operations in the encoder. In addition, patch shift is introduced to Mask-RadarNet for efficient spatial-temporal feature learning. By shifting part of patches with a specific mosaic pattern in the temporal dimension, Mask-RadarNet achieves competitive performance while reducing the computational burden of the spatial-temporal modeling. In order to capture the spatial-temporal semantic contextual information, we design the class masking attention module (CMAM) in our encoder. Moreover, a lightweight auxiliary decoder is added to our model to aggregate prior maps generated from the CMAM. Experiments on the CRUW dataset demonstrate that the proposed Mask-RadarNet achieves state-of-the-art performance with relatively lower computational complexity and fewer parameters.

Accurate segmentation of cervical structures in transvaginal ultrasound (TVS) is critical for assessing the risk of spontaneous preterm birth (PTB), yet the scarcity of labeled data limits the performance of supervised learning approaches. This paper introduces the Fetal Ultrasound Grand Challenge (FUGC), the first benchmark for semi-supervised learning in cervical segmentation, hosted at ISBI 2025. FUGC provides a dataset of 890 TVS images, including 500 training images, 90 validation images, and 300 test images. Methods were evaluated using the Dice Similarity Coefficient (DSC), Hausdorff Distance (HD), and runtime (RT), with a weighted combination of 0.4/0.4/0.2. The challenge attracted 10 teams with 82 participants submitting innovative solutions. The best-performing methods for each individual metric achieved 90.26% mDSC, 38.88 mHD, and 32.85 ms RT, respectively. FUGC establishes a standardized benchmark for cervical segmentation, demonstrates the efficacy of semi-supervised methods with limited labeled data, and provides a foundation for AI-assisted clinical PTB risk assessment.

Accurately and preoperatively predicting survival for high-grade gliomas (HGGs) is important for optimizing treatment strategies. Increasing evidence suggests that brain structural and functional connectivity networks derived from advanced magnetic resonance imaging (MRI) are promising predictors for HGG survival. However, advanced MRIs (e.g., diffusion MRI and functional MRI) are generally clinically inaccessible for HGG patients before initiating therapy. To compensate for lack of advanced MRI modalities in brain network studies, in this paper we evaluate the feasibility and performance of predicting HGG survival using exclusively preoperative multi-modal basic structural MRI (sMRI, e.g., T1- and T2-weighted MRI) based brain regional radiomics similarity networks (R2SNs). To this end, we propose a new cooperative multiplex graph neural network (GNN) based multi-modal R2SN integration framework for preoperative HGG survival prediction. First, multi-modal R2SNs are represented by a multiplex network, where each modality-specific R2SN forms one multiplex layer and nodes (i.e., brain regions of interest (ROIs)) are coupled to their replicas across multiplex layers. This facilitates flexible inter-ROI communications both within and between R2SNs. Second, a cooperative GNN is applied to capture intra-modal node feature propagations within each multiplex layer, followed by attention mechanisms used to capture inter-modal node feature interactions across multiplex layers. Finally, a tailored tumor-aware graph pooling is developed to assemble features from the tumor-intersecting ROIs for survival prediction. Extensive experiments on a collected HGG database with three basic sMRI modalities demonstrate the superiority of our method over state-of-the-art baselines in survival stratification. The code is available at https://github.com/ZiLaoTou/TCM-GNN.

Wei Wang, Zhaowei Li, Qi Xu, Linfeng Li, YiQing Cai, Botian Jiang, Hang Song, Xingcan Hu, Pengyu Wang, Li Xiao. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Functional connectivity networks (FCNs), as graph-structured data derived from functional magnetic resonance imaging (fMRI), are essential for understanding how brain functions coordinate with behavior and cognition. However, the utility of these FCNs is often limited by the brain atlas, since the predefined regions of interest by the atlas represent nodes in FCNs. To address these limitations and enhance the comparability of functional connectivity analyses across different atlases, we introduce the Spatio-Temporal Mapping Generative Adversarial Network (STMap-GAN) based on generative modeling. Convolutional networks and long short-term memory modules are used in the generator to improve the spatio and temporal consistency of generated fMRI time series for target brain atlases. The transformer module in the discriminator can effectively capture different features in fMRI time series, thus accurately distinguishing the generated time series from ground truth. This study demonstrates the ability of STMap-GAN to maintain high fidelity in FCN mapping across various atlases, ensuring consistency and replicability in neuroscience research.

Resting-state functional connectivity networks (rs-FCNs) have been most frequently used for brain network analysis in neuroscience. However, a body of evidence indicates that task-state FCNs (ts-FCNs) are better associated with individual differences in behavior than rs-FCN. Until now there have been no studies of ascertaining to what extent rs-FCNs can account for ts-FCNs. In this paper, we propose a Multiple Graph Autoencoder based Generative Adversarial Network (MGAE-GAN) model to enable the inference of ts-FCNs from rs-FCNs. The generator of MGAE-GAN is built upon several graph autoencoders to learn and adaptively combine multiple implicit relationships between rs-FCNs and ts-FCNs. To ensure the authenticity of the predicted ts-FCNs, we design the discriminator of MGAE-GAN based on graph metric learning. Additionally, we incorporate a correlation loss and a subject-similarity-preserving loss to maintain overall correlation and between-subject similarities before and after the generator, respectively. Experimental results on the Human Connectome Project (HCP) S1200 demonstrate the effectiveness of our MGAE-GAN for predicting ts-FCNs from rs-FCNs.

Autism Spectrum Disorder (ASD) is a complex neurodevelopmental disorder characterized by abnormalities in social interaction, communication, and behavioral patterns. Research on brain functional connectivity networks (FCNs) derived from functional magnetic resonance imaging (fMRI) has become a prominent approach for understanding and diagnosing ASD, as well as advancing effective treatments. Despite significant advancements in related FCN studies, many challenges remain unresolved. On the one hand, existing methods face limitations in capturing the complex interactions between brain regions; on the other hand, modeling brain functional connectivity remains insufficient and lacks interpretability. To address these issues, we propose a novel spatio-temporal aware graph convolutional network (STGNet) for ASD diagnosis. In the STGNet, we first introduce a transformer-inspired Node Attention Encoder module to mitigate the compression of temporal information and further generate an interpretable FCN matrix. Moreover, leveraging the community structure of FCNs, a partition-global node embedding architecture is developed to hierarchically extract fine-grained features of functional subnetworks and global connectivity patterns. Meanwhile, the GCN used for downstream prediction combines both such spatial and temporal features. Extensive experiments on the Autism Brain Imaging Data Exchange (ABIDE) demonstrate that our STGNet outperforms several existing models for ASD diagnosis. Additionally, interpretability validation confirms our model’s ability to identify potential biomarkers, aligning well with findings from previous neuroimaging studies.

Patients with glioblastoma (GBM) have a poor survival rate. In order to facilitate early interventions and personalized therapeutic treatment, there is a pressing need for employing routine non-invasive MRI for preoperative GBM survival prediction. In this paper, we investigate to what extent regional radiomics similarity networks (R2SNs) can be used to predict overall survival (OS) time in GBM. Different from the widely used MRI-derived radiomics features that focus on single or several brain regions independently, the R2SNs can take into account the potential associations among brain regions with radiomics similarity for improved survival prediction. Specifically, we first introduce a distance correlation based R2SN (DC-R2SN), where distance correlation (instead of Pearson’s correlation in the traditional R2SN) is adopted to measure the more complex interactions between a pair of brain regions defined by the corresponding radiomics features. A graph neural network (GNN) framework is then proposed for fusing DCR2SNs and clinical data to predict OS time of GBM patients. Experimental results on the publicly available UPenn-GBM database demonstrate the effectiveness of our proposed GNN based survival prediction framework with the DC-R2SNs.

Recently, functional connectivity network (FCN) analysis via graph convolutional networks (GCNs) has greatly boosted diagnostic performance of brain diseases on a population graph for subject classification. However, most existing methods only focus on FCNs based on a single brain atlas (ignoring complementary information among multiatlas-based FCNs), and the population graph structure is preconstructed and fixed during the GCN training (not truly reflecting the relation between subjects). In this paper, we propose an adaptive multiview community-preserved graph convolutional network (CP-GCN) method to accommodate multiatlas-based FCNs. Specifically, we first introduce a multiview FCN fusion module to obtain multiatlas FCN embeddings via concatenating both intra- and inter-atlas embeddings that are extracted separately from fully connected layers. We then develop a multihead similarity learning module to adaptively learn the population graph structure, best serving GCN for node classification. Finally, under the learned graph structure, a CP-GCN based node classification module is applied for subject classification through designing a community-preserved constraint on the GCN. Experimental results on the ABIDE validate the effectiveness of our method for autism identification, and our findings related to autism can be easily traced back with biological interpretability.

Graph convolutional deep learning has emerged as a promising method to explore the functional organization of the human brain in neuroscience research. This paper presents a novel framework that utilizes the gated graph transformer (GGT) model to predict individuals' cognitive ability based on functional connectivity (FC) derived from fMRI. Our framework incorporates prior spatial knowledge and uses a random-walk diffusion strategy that captures the intricate structural and functional relationships between different brain regions. Specifically, our approach employs learnable structural and positional encodings (LSPE) in conjunction with a gating mechanism to efficiently disentangle the learning of positional encoding (PE) and graph embeddings. Additionally, we utilize the attention mechanism to derive multi-view node feature embeddings and dynamically distribute propagation weights between each node and its neighbors, which facilitates the identification of significant biomarkers from functional brain networks and thus enhances the interpretability of the findings. To evaluate our proposed model in cognitive ability prediction, we conduct experiments on two large-scale brain imaging datasets: the Philadelphia Neurodevelopmental Cohort (PNC) and the Human Connectome Project (HCP). The results show that our approach not only outperforms existing methods in prediction accuracy but also provides superior explainability, which can be used to identify important FCs underlying cognitive behaviors.

Recent studies show that multi-modal data fusion techniques combine information from diverse sources for comprehensive diagnosis and prognosis of complex brain disorder, often resulting in improved accuracy compared to single-modality approaches. However, many existing data fusion methods extract features from homogeneous networs, ignoring heterogeneous structural information among multiple modalities. To this end, we propose a Hypergraph-based Multi-modal data Fusion algorithm, namely HMF. Specifically, we first generate a hypergraph similarity matrix to represent the high-order relationships among subjects, and then enforce the regularization term based upon both the inter- and intra-modality relationships of the subjects. Finally, we apply HMF to integrate imaging and genetics datasets. Validation of the proposed method is performed on both synthetic data and real samples from schizophrenia study. Results show that our algorithm outperforms several competing methods, and reveals significant interactions among risk genes, environmental factors and abnormal brain regions.

Functional magnetic resonance imaging (fMRI) is a powerful technique with the potential to estimate individual variations in behavioral and cognitive traits. Joint learning of multiple datasets can utilize their complementary information so as to improve learning performance, but it also gives rise to the challenge for data fusion to effectively integrate brain patterns elicited by multiple fMRI data. However, most of the current data fusion methods analyze each single dataset separately and further infer the relationship among them, which fail to utilize the multidimensional structure inherent across modalities and may ignore complex but important interactions. To address this issue, we propose a novel sparse tensor decomposition method to integrate multiple task-stimulus (paradigm) fMRI data. Seeing each paradigm fMRI as one modality, our proposed method considers the relationships across subjects and modalities simultaneously. In specific, a third-order tensor is first modeled by using the functional network connectivity (FNC) of subjects in multiple fMRI paradigms. A novel sparse tensor decomposition with the regularization terms is designed to factorize the tensor into a series of rank-one components, which can extract the shared components across modalities as the embedded features. The L2,1-norm regularizer (i.e., group sparsity) is enforced to select a few common features among multiple subjects. Validation of the proposed method is performed on realistic three paradigm fMRI datasets from the Philadelphia Neurodevelopmental Cohort (PNC) study, for the study of the relationship between the FNC and human cognitive abilities. Experimental results show our method outperforms several other competing methods in the prediction of individuals with different cognitive behaviors via the wide range achievement test (WRAT). Furthermore, our method discovers the FNC related to the cognitive behaviors, such as the connectivity associated with the default mode network (DMN) for three paradigms, and the connectivity between DMN and visual (VIS) domains within the emotion task.

Human brain development is a complex and dynamic process caused by several factors such as genetics, sex hormones, and environmental changes. A number of recent studies on brain development have examined functional connectivity (FC) defined by the temporal correlation between time series of different brain regions. We propose to add the directional flow of information during brain maturation. To do so, we extract effective connectivity (EC) through Granger causality (GC) for two different groups of subjects, i.e., children and young adults. The motivation is that the inclusion of causal interaction may further discriminate brain connections between two age groups and help to discover new connections between brain regions. The contributions of this study are threefold. First, there has been a lack of attention to EC-based feature extraction in the context of brain development. To this end, we propose a new kernel-based GC (KGC) method to learn nonlinearity of complex brain network, where a reduced Sine hyperbolic polynomial (RSP) neural network was used as our proposed learner. Second, we used causality values as the weight for the directional connectivity between brain regions. Our findings indicated that the strength of connections was significantly higher in young adults relative to children. In addition, our new EC-based feature outperformed FC-based analysis from Philadelphia neurocohort (PNC) study with better discrimination of different age groups. Moreover, the fusion of these two sets of features (FC + EC) improved brain age prediction accuracy by more than 4%, indicating that they should be used together for brain development studies.

Recently, a hypergraph constructed from functional magnetic resonance imaging (fMRI) was utilized to explore brain functional connectivity networks (FCNs) for the classification of neurodegenerative diseases. Each edge of a hypergraph (called hyperedge) can connect any number of brain regions-of-interest (ROIs) instead of only two ROIs, and thus characterizes high-order relations among multiple ROIs that cannot be uncovered by a simple graph in the traditional graph based FCN construction methods. Unlike the existing hypergraph based methods where all hyperedges are assumed to have equal weights and only certain topological features are extracted from the hypergraphs, we propose a hypergraph learning based method for FCN construction in this paper. Specifically, we first generate hyperedges from fMRI time series based on sparse representation, then employ hypergraph learning to adaptively learn hyperedge weights, and finally define a hypergraph similarity matrix to represent the FCN. In our proposed method, weighting hyperedges results in better discriminative FCNs across subjects, and the defined hypergraph similarity matrix can better reveal the overall structure of brain network than using those hypergraph topological features. Moreover, we propose a multi-hypergraph learning based method by integrating multi-paradigm fMRI data, where the hyperedge weights associated with each fMRI paradigm are jointly learned and then a unified hypergraph similarity matrix is computed to represent the FCN. We validate the effectiveness of the proposed method on the Philadelphia Neurodevelopmental Cohort dataset for the classification of individuals' learning ability from three paradigms of fMRI data. Experimental results demonstrate that our proposed approach outperforms the traditional graph based methods (i.e., Pearson's correlation and partial correlation with the graphical Lasso) and the existing unweighted hypergraph based methods, which sheds light on how to optimize estimation of FCNs for cognitive and behavioral study.

In this paper, with respect to multichannel synthetic aperture radar (SAR), we first formulate the problems of Doppler ambiguities on the radial velocity (RV) estimation of a ground moving target in the range-compressed domain, the range-Doppler domain, and the image domain, respectively. It is revealed that in these problems, the cascaded time–space Doppler ambiguity (CTSDA) may arise; that is, the time domain Doppler ambiguity in each channel arises first and then the spatial domain Doppler ambiguity among multichannels arises second. Accordingly, the multichannel SAR systems with different parameters are investigated in three cases with different Doppler ambiguity properties. Then, a multifrequency SAR is proposed for the RV estimation by solving the ambiguity problem based on the Chinese remainder theorem (CRT). In the first two cases, the ambiguity problem can be solved by the existing closed-form robust CRT. In the third case, it is found that the problem is different from the conventional CRT problem and we call it a double remaindering problem in this paper. We then propose a sufficient condition under which the double remaindering problem, i.e., the CTSDA, can also be solved by the closed-form robust CRT. When the sufficient condition is not satisfied, a searching-based method is proposed. Finally, some results of numerical experiments are provided to demonstrate the effectiveness of the proposed methods.

This paper investigates polynomial remainder codes with non-pairwise coprime moduli. We first propose a robust reconstruction for polynomials from erroneous residues when the degrees of all residue errors are small, namely robust Chinese Remainder Theorem (CRT) for polynomials. It basically says that a polynomial can be reconstructed from erroneous residues such that the degree of the reconstruction error is upper bounded by τ whenever the degrees of all residue errors are upper bounded by τ, where a sufficient condition for τ and a reconstruction algorithm are obtained. By relaxing the constraint that all residue errors have small degrees, another robust reconstruction is then presented when there are multiple unrestricted errors and an arbitrary number of errors with small degrees in the residues. By making full use of redundancy in moduli, we obtain a stronger residue error correction capability in the sense that apart from the number of errors that can be corrected in the previous existing result, some errors with small degrees can be also corrected in the residues. With this newly obtained result, improvements in uncorrected error probability and burst error correction capability in a data transmission are illustrated.

Chinese remainder theorem (CRT) provides an undersampling method to detect the frequency of a complex sinusoid. The detection of the multiple frequencies in a signal formed by the superposition of multiple complex sinusoids is a task frequently encountered in several applications such as phase unwrapping in radar signal processing and multiwavelength optical interferometry. A generalized CRT for multiple integers has recently been studied. In this letter, we complement it by giving two new conditions that ensure the maximal possible dynamic range for the multiple integers, i.e., the least common multiple (lcm) of all the moduli. Then, two corresponding determination algorithms are also proposed.

This paper investigates polynomial remainder codes with non-pairwise coprime moduli. We first consider a robust reconstruction problem for polynomials from erroneous residues when the degrees of all residue errors are assumed small, namely, the robust Chinese Remainder Theorem (CRT) for polynomials. It basically says that a polynomial can be reconstructed from its erroneous residues such that the degree of the reconstruction error is upper bounded by$\tau$whenever the degrees of all residue errors are upper bounded by$\tau$, where a sufficient condition for$\tau$and a reconstruction algorithm are obtained. By relaxing the constraint that all residue errors have small degrees, another robust reconstruction is then presented when there are multiple unrestricted errors and an arbitrary number of errors with small degrees in the residues. We finally obtain a stronger residue error correction capability in the sense that apart from the number of errors that can be corrected in the previous existing result, some errors with small degrees can be also corrected in the residues. With this newly obtained result, improvements in uncorrected error probability and burst error correction capability in data transmission are illustrated.

A generalized Chinese remainder theorem (CRT) for the determination of two integers is studied in this letter, where the correspondence between the remainders and the two integers in each residue set is not known. A better range than the existing known ones of two integers that can be uniquely determined from their residue sets is first obtained. Then, a closed-form and simple determination algorithm is proposed. Finally, a better sufficient condition on the range of determinable two integers is obtained when the number of erroneous residue sets is given. The study is motivated and has applications in the determination of multiple frequencies from multiple undersampled waveforms.