Adapting Vision Foundation Models for Real-time Ultrasound Image Segmentation

We propose a novel approach that adapts hierarchical vision foundation models for real-time ultrasound image segmentation. Existing ultrasound segmentation methods often struggle with adaptability to new tasks, relying on costly manual annotations, while real-time approaches generally fail to match state-of-the-art performance. To overcome these limitations, we introduce an adaptive framework that leverages the vision foundation model Hiera to extract multi-scale features, interleaved with DINOv2 representations to enhance visual expressiveness. These enriched features are then decoded to produce precise and robust segmentation. We conduct extensive evaluations on six public datasets and one in-house dataset, covering both cardiac and thyroid ultrasound segmentation. Experiments show that our approach outperforms state-of-the-art methods across multiple datasets and excels with limited supervision, surpassing nnUNet by over 20\% on average in the 1\% and 10\% data settings. Our method achieves $\sim$77 FPS inference speed with TensorRT on a single GPU, enabling real-time clinical applications.

Progressive Test Time Energy Adaptation for Medical Image Segmentation

We propose a model-agnostic, progressive test-time energy adaptation approach for medical image segmentation. Maintaining model performance across diverse medical datasets is challenging, as distribution shifts arise from inconsistent imaging protocols and patient variations. Unlike domain adaptation methods that require multiple passes through target data - impractical in clinical settings - our approach adapts pretrained models progressively as they process test data. Our method leverages a shape energy model trained on source data, which assigns an energy score at the patch level to segmentation maps: low energy represents in-distribution (accurate) shapes, while high energy signals out-of-distribution (erroneous) predictions. By minimizing this energy score at test time, we refine the segmentation model to align with the target distribution. To validate the effectiveness and adaptability, we evaluated our framework on eight public MRI (bSSFP, T1- and T2-weighted) and X-ray datasets spanning cardiac, spinal cord, and lung segmentation. We consistently outperform baselines both quantitatively and qualitatively.

Causal Modeling of fMRI Time-series for Interpretable Autism Spectrum Disorder Classification

Autism spectrum disorder (ASD) is a neurological and developmental disorder that affects social and communicative behaviors. It emerges in early life and is generally associated with lifelong disabilities. Thus, accurate and early diagnosis could facilitate treatment outcomes for those with ASD. Functional magnetic resonance imaging (fMRI) is a useful tool that measures changes in brain signaling to facilitate our understanding of ASD. Much effort is being made to identify ASD biomarkers using various connectome-based machine learning and deep learning classifiers. However, correlation-based models cannot capture the non-linear interactions between brain regions. To solve this problem, we introduce a causality-inspired deep learning model that uses time-series information from fMRI and captures causality among ROIs useful for ASD classification. The model is compared with other baseline and state-of-the-art models with 5-fold cross-validation on the ABIDE dataset. We filtered the dataset by choosing all the images with mean FD less than 15mm to ensure data quality. Our proposed model achieved the highest average classification accuracy of 71.9% and an average AUC of 75.8%. Moreover, the inter-ROI causality interpretation of the model suggests that the left precuneus, right precuneus, and cerebellum are placed in the top 10 ROIs in inter-ROI causality among the ASD population. In contrast, these ROIs are not ranked in the top 10 in the control population. We have validated our findings with the literature and found that abnormalities in these ROIs are often associated with ASD.

The Silent Majority: Demystifying Memorization Effect in the Presence of Spurious Correlations

Machine learning models often rely on simple spurious features -- patterns in training data that correlate with targets but are not causally related to them, like image backgrounds in foreground classification. This reliance typically leads to imbalanced test performance across minority and majority groups. In this work, we take a closer look at the fundamental cause of such imbalanced performance through the lens of memorization, which refers to the ability to predict accurately on \textit{atypical} examples (minority groups) in the training set but failing in achieving the same accuracy in the testing set. This paper systematically shows the ubiquitous existence of spurious features in a small set of neurons within the network, providing the first-ever evidence that memorization may contribute to imbalanced group performance. Through three experimental sources of converging empirical evidence, we find the property of a small subset of neurons or channels in memorizing minority group information. Inspired by these findings, we articulate the hypothesis: the imbalanced group performance is a byproduct of ``noisy'' spurious memorization confined to a small set of neurons. To further substantiate this hypothesis, we show that eliminating these unnecessary spurious memorization patterns via a novel framework during training can significantly affect the model performance on minority groups. Our experimental results across various architectures and benchmarks offer new insights on how neural networks encode core and spurious knowledge, laying the groundwork for future research in demystifying robustness to spurious correlation.

A Flow-based Truncated Denoising Diffusion Model for Super-resolution Magnetic Resonance Spectroscopic Imaging

Magnetic Resonance Spectroscopic Imaging (MRSI) is a non-invasive imaging technique for studying metabolism and has become a crucial tool for understanding neurological diseases, cancers and diabetes. High spatial resolution MRSI is needed to characterize lesions, but in practice MRSI is acquired at low resolution due to time and sensitivity restrictions caused by the low metabolite concentrations. Therefore, there is an imperative need for a post-processing approach to generate high-resolution MRSI from low-resolution data that can be acquired fast and with high sensitivity. Deep learning-based super-resolution methods provided promising results for improving the spatial resolution of MRSI, but they still have limited capability to generate accurate and high-quality images. Recently, diffusion models have demonstrated superior learning capability than other generative models in various tasks, but sampling from diffusion models requires iterating through a large number of diffusion steps, which is time-consuming. This work introduces a Flow-based Truncated Denoising Diffusion Model (FTDDM) for super-resolution MRSI, which shortens the diffusion process by truncating the diffusion chain, and the truncated steps are estimated using a normalizing flow-based network. The network is conditioned on upscaling factors to enable multi-scale super-resolution. To train and evaluate the deep learning models, we developed a 1H-MRSI dataset acquired from 25 high-grade glioma patients. We demonstrate that FTDDM outperforms existing generative models while speeding up the sampling process by over 9-fold compared to the baseline diffusion model. Neuroradiologists' evaluations confirmed the clinical advantages of our method, which also supports uncertainty estimation and sharpness adjustment, extending its potential clinical applications.

AI-powered multimodal modeling of personalized hemodynamics in aortic stenosis

Aortic stenosis (AS) is the most common valvular heart disease in developed countries. High-fidelity preclinical models can improve AS management by enabling therapeutic innovation, early diagnosis, and tailored treatment planning. However, their use is currently limited by complex workflows necessitating lengthy expert-driven manual operations. Here, we propose an AI-powered computational framework for accelerated and democratized patient-specific modeling of AS hemodynamics from computed tomography. First, we demonstrate that our automated meshing algorithms can generate task-ready geometries for both computational and benchtop simulations with higher accuracy and 100 times faster than existing approaches. Then, we show that our approach can be integrated with fluid-structure interaction and soft robotics models to accurately recapitulate a broad spectrum of clinical hemodynamic measurements of diverse AS patients. The efficiency and reliability of these algorithms make them an ideal complementary tool for personalized high-fidelity modeling of AS biomechanics, hemodynamics, and treatment planning.

Assessment of Clonal Hematopoiesis of Indeterminate Potential from Cardiac Magnetic Resonance Imaging using Deep Learning in a Cardio-oncology Population

Background: We propose a novel method to identify who may likely have clonal hematopoiesis of indeterminate potential (CHIP), a condition characterized by the presence of somatic mutations in hematopoietic stem cells without detectable hematologic malignancy, using deep learning techniques. Methods: We developed a convolutional neural network (CNN) to predict CHIP status using 4 different views from standard delayed gadolinium-enhanced cardiac magnetic resonance imaging (CMR). We used 5-fold cross validation on 82 cardio-oncology patients to assess the performance of our model. Different algorithms were compared to find the optimal patient-level prediction method using the image-level CNN predictions. Results: We found that the best model had an area under the receiver operating characteristic curve of 0.85 and an accuracy of 82%. Conclusions: We conclude that a deep learning-based diagnostic approach for CHIP using CMR is promising.

STNAGNN: Spatiotemporal Node Attention Graph Neural Network for Task-based fMRI Analysis

Task-based fMRI uses actions or stimuli to trigger task-specific brain responses and measures them using BOLD contrast. Despite the significant task-induced spatiotemporal brain activation fluctuations, most studies on task-based fMRI ignore the task context information aligned with fMRI and consider task-based fMRI a coherent sequence. In this paper, we show that using the task structures as data-driven guidance is effective for spatiotemporal analysis. We propose STNAGNN, a GNN-based spatiotemporal architecture, and validate its performance in an autism classification task. The trained model is also interpreted for identifying autism-related spatiotemporal brain biomarkers.

2.5D Multi-view Averaging Diffusion Model for 3D Medical Image Translation: Application to Low-count PET Reconstruction with CT-less Attenuation Correction

Positron Emission Tomography (PET) is an important clinical imaging tool but inevitably introduces radiation hazards to patients and healthcare providers. Reducing the tracer injection dose and eliminating the CT acquisition for attenuation correction can reduce the overall radiation dose, but often results in PET with high noise and bias. Thus, it is desirable to develop 3D methods to translate the non-attenuation-corrected low-dose PET (NAC-LDPET) into attenuation-corrected standard-dose PET (AC-SDPET). Recently, diffusion models have emerged as a new state-of-the-art deep learning method for image-to-image translation, better than traditional CNN-based methods. However, due to the high computation cost and memory burden, it is largely limited to 2D applications. To address these challenges, we developed a novel 2.5D Multi-view Averaging Diffusion Model (MADM) for 3D image-to-image translation with application on NAC-LDPET to AC-SDPET translation. Specifically, MADM employs separate diffusion models for axial, coronal, and sagittal views, whose outputs are averaged in each sampling step to ensure the 3D generation quality from multiple views. To accelerate the 3D sampling process, we also proposed a strategy to use the CNN-based 3D generation as a prior for the diffusion model. Our experimental results on human patient studies suggested that MADM can generate high-quality 3D translation images, outperforming previous CNN-based and Diffusion-based baseline methods.

Calibrating Multi-modal Representations: A Pursuit of Group Robustness without Annotations

Fine-tuning pre-trained vision-language models, like CLIP, has yielded success on diverse downstream tasks. However, several pain points persist for this paradigm: (i) directly tuning entire pre-trained models becomes both time-intensive and computationally costly. Additionally, these tuned models tend to become highly specialized, limiting their practicality for real-world deployment; (ii) recent studies indicate that pre-trained vision-language classifiers may overly depend on spurious features -- patterns that correlate with the target in training data, but are not related to the true labeling function; and (iii) existing studies on mitigating the reliance on spurious features, largely based on the assumption that we can identify such features, does not provide definitive assurance for real-world applications. As a piloting study, this work focuses on exploring mitigating the reliance on spurious features for CLIP without using any group annotation. To this end, we systematically study the existence of spurious correlation on CLIP and CILP+ERM. We first, following recent work on Deep Feature Reweighting (DFR), verify that last-layer retraining can greatly improve group robustness on pretrained CLIP. In view of them, we advocate a lightweight representation calibration method for fine-tuning CLIP, by first generating a calibration set using the pretrained CLIP, and then calibrating representations of samples within this set through contrastive learning, all without the need for group labels. Extensive experiments and in-depth visualizations on several benchmarks validate the effectiveness of our proposals, largely reducing reliance and significantly boosting the model generalization.