Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention最新文献

Multi-site brain MRI heterogeneity caused by differences in scanner field strengths, acquisition protocols, and software versions poses a significant challenge for consistent analysis. Image-level harmonization, leveraging advanced learning methods, has attracted increasing attention. However, existing methods often rely on paired data (e.g., human traveling phantoms) for training, which are not always available. Some methods perform MRI harmonization by transferring target-style features to source images but require explicitly learning disentangled image styles (e.g., contrast) via encoder-decoder networks, which increases computational complexity. This paper presents an unpaired MRI harmonization (UMH) framework based on a new image style-guided diffusion model. UMH operates in two stages: (1) a coarse harmonizer that aligns multi-site MRIs to a unified domain via a conditional latent diffusion model while preserving anatomical content; and (2) a fine harmonizer that adapts coarsely harmonized images to a specific target using style embeddings derived from a pre-trained Contrastive Language-Image Pre-training (CLIP) encoder, which captures semantic style differences between the original MRIs and their coarsely-aligned counterparts, eliminating the need for paired data. By leveraging rich semantic style representations of CLIP, UMH avoids learning image styles explicitly, thereby reducing computation costs. We evaluate UMH on 4,123 MRIs from three distinct multi-site datasets, with results suggesting its superiority over several state-of-the-art (SOTA) methods across image-level comparison, downstream classification, and brain tissue segmentation tasks.

Understanding the organization of human brain networks has become a central focus in neuroscience, particularly in the study of functional connectivity, which plays a crucial role in diagnosing neurological disorders. Advances in functional magnetic resonance imaging and machine learning techniques have significantly improved brain network analysis. However, traditional machine learning approaches struggle to capture the complex relationships between brain regions, while deep learning methods, particularly Transformer-based models, face computational challenges due to their quadratic complexity in long-sequence modeling. To address these limitations, we propose a Core-Periphery State-Space Model (CP-SSM), an innovative framework for functional connectome classification. Specifically, we introduce Mamba, a selective state-space model with linear complexity, to effectively capture long-range dependencies in functional brain networks. Furthermore, inspired by the core-periphery (CP) organization, a fundamental characteristic of brain networks that enhances efficient information transmission, we design CP-MoE, a CP-guided Mixture-of-Experts that improves the representation learning of brain connectivity patterns. We evaluate CP-SSM on two benchmark fMRI datasets: ABIDE and ADNI. Experimental results demonstrate that CP-SSM surpasses Transformer-based models in classification performance while significantly reducing computational complexity. These findings highlight the effectiveness and efficiency of CP-SSM in modeling brain functional connectivity, offering a promising direction for neuroimaging-based neurological disease diagnosis. Our code is available at https://github.com/m1nhengChen/cpssm.

$\text{A}\beta$ Positron Emission Tomography (PET) is often used to manage Alzheimer's disease (AD). To better understand $\text{A}\beta$ progression, we introduce and evaluate a mathematical model that couples $\text{A}\beta$ at parcellated gray matter regions. We term this model LNODE for "latent network ordinary differential equations". At each region, we track normal $\text{A}\beta$ , abnormal $\text{A}\beta$ , and $m$ latent states that intend to capture unobservable mechanisms coupled to $\text{A}\beta$ progression. LNODE is parameterized by subject-specific parameters and cohort parameters. We jointly invert for these parameters by fitting the model to $\text{A}\beta$ -PET data from 585 subjects from the ADNI dataset. Although underparameterized, our model achieves population $R^2\ge 98\%$ compared to $R^2\le 60\%$ when fitting without latent states. Furthermore, these preliminary results suggest the existence of different subtypes of $\text{A}\beta$ progression.

Language promptable X-ray image segmentation would enable greater flexibility for human-in-the-loop workflows in diagnostic and interventional precision medicine. Prior efforts have contributed task-specific models capable of solving problems within a narrow scope, but expanding to broader use requires additional data, annotations, and training time. Recently, language-aligned foundation models (LFMs) - machine learning models trained on large amounts of highly variable image and text data thus enabling broad applicability - have emerged as promising tools for automated image analysis. Existing foundation models for medical image analysis focus on scenarios and modalities where large, richly annotated datasets are available. However, the X-ray imaging modality features highly variable image appearance and applications, from diagnostic chest X-rays to interventional fluoroscopy, with varying availability of data. To pave the way toward an LFM for comprehensive and language-aligned analysis of arbitrary medical X-ray images, we introduce FluoroSAM, a language-promptable variant of the Segment-Anything Model, trained from scratch on 3M synthetic X-ray images from a wide variety of human anatomies, imaging geometries, and viewing angles. These include pseudo-ground truth masks for 128 organ types and 464 tools with associated text descriptions. FluoroSAM is capable of segmenting myriad anatomical structures and tools based on natural language prompts, thanks to the novel incorporation of vector quantization (VQ) of text embeddings in the training process. We demonstrate FluoroSAM's performance quantitatively on real X-ray images and showcase on several applications how FluoroSAM is a key enabler for rich human-machine interaction in the X-ray image acquisition and analysis context. Code is available at https://github.com/arcadelab/fluorosam.

Multi-tracer positron emission tomography (PET), which assesses key neurological biomarkers such as tau pathology, neuroinflammatory, $\beta$ -amyloid deposition, and glucose metabolism, plays a vital role in diagnosing neurological disorders by providing complementary insights into the brain's molecular and functional state. Acquiring multi-tracer PET scans remains challenging due to high costs, radiation exposure, and limited tracer availability. Recent studies have attempted to synthesize multi-tracer PET images from structural MRI. However, these approaches typically either rely on direct mappings to individual tracers or lack distributional constraints, leading to inconsistencies in image quality across tracers. To this end, we propose a normalized diffusion framework (NDF) to generate high-quality multi-tracer PET images from a single MRI through a distribution-guided class-conditioned weighted diffusion model. Specifically, a diffusion model conditioned on MRI and tracer-specific class labels is trained to synthesize PET images of multiple tracers, and a pre-trained normalizing flow model refines these outputs by mapping them into a shared distribution space. This mapping ensures that the subject-specific high-level features across different PET tracers are preserved, resulting in more consistent and accurate synthesis. Experiments on a total of 425 subjects with multi-tracer PET scans demonstrate that our NDF outperforms current state-of-the-art methods, indicating its potential for advancing multi-tracer PET synthesis.

Major Adverse Cardiovascular Events (MACE) remain the leading cause of mortality globally, as reported in the Global Disease Burden Study 2021. Opportunistic screening leverages data collected from routine health check-ups and multimodal data can play a key role to identify at-risk individuals. Chest X-rays (CXR) provide insights into chronic conditions contributing to major adverse cardiovascular events (MACE), while 12-lead electrocardiogram (ECG) directly assesses cardiac electrical activity and structural abnormalities. Integrating CXR and ECG could offer a more comprehensive risk assessment than conventional models, which rely on clinical scores, computed tomography (CT) measurements, or biomarkers, which may be limited by sampling bias and single modality constraints. We propose a novel predictive modeling framework - MOSCARD, multimodal causal reasoning with co-attention to align two distinct modalities and simultaneously mitigate bias and confounders in opportunistic risk estimation. Primary technical contributions are - (i) multimodal alignment of CXR with ECG guidance; (ii) integration of causal reasoning; (iii) dual back-propagation graph for deconfounding. Evaluated on internal, shift data from emergency department (ED) and external MIMIC datasets, our model outperformed single (ED) and external MIMIC datasets, our model outperformed single modality and state-of-the-art foundational models - AUC: 0.75, 0.83, 0.71 respectively. Proposed cost-effective opportunistic screening enables early intervention, improving patient outcomes and reducing disparities.

For the early diagnosis of Alzheimer's disease (AD), it is essential that we have effective multiclass classification methods that can distinct subjects with mild cognitive impairment (MCI) from cognitively normal (CN) subjects and AD patients. However, significant overlaps of biomarker distributions among these groups make this a difficult task. In this work, we propose a novel framework for multi-modal, multiclass AD diagnosis that can integrate information from diverse and complex modalities to resolve ambiguity among the disease groups and hence enhance classification performances. More specifically, our approach integrates T1-weighted MRI, tau PET, fiber orientation distribution (FOD) from diffusion MRI (dMRI), and Montreal Cognitive Assessment (MoCA) scores to classify subjects into AD, MCI, and CN groups. We introduce a Swin-FOD model to extract order-balanced features from FOD and use contrastive learning to align MRI and PET features. These aligned features and MoCA scores are then processed with a Tabular Prior-data Fitted In-context Learning (TabPFN) method, which selects model parameters based on the alignment between input data and prior data during pre-training, eliminating the need for additional training or fine-tuning. Evaluated on the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset ( $n=1147$ ), our model achieved a diagnosis accuracy of 73.21%, outperforming all comparison models ( $n=10$ ). We also performed Shapley analysis and quantitatively evaluated the essential contributions of each modality.

Alzheimer's Disease (AD) and Lewy Body Dementia (LBD) often exhibit overlapping pathologies, leading to common symptoms that make diagnosis challenging and protracted in clinical settings. While many studies achieve promising accuracy in identifying AD and LBD at earlier stages, they often focus on discrete classification rather than capturing the gradual nature of disease progression. Since dementia develops progressively, understanding the continuous trajectory of dementia is crucial, as it allows us to uncover hidden patterns in cognitive decline and provides critical insights into the underlying mechanisms of disease progression. To address this gap, we propose a novel multi-scale learning framework that leverages hierarchical anatomical features to model the continuous relationships across various neurodegenerative conditions, including Mild Cognitive Impairment, AD, and LBD. Our approach employs the proposed hierarchical graph embedding fusion technique, integrating anatomical features, cortical folding patterns, and structural connectivity at multiple scales. This integration captures both fine-grained and coarse anatomical details, enabling the identification of subtle patterns that enhance differentiation between dementia types. Additionally, our framework projects each subject onto continuous tree structures, providing intuitive visualizations of disease trajectories and offering a more interpretable way to track cognitive decline. To validate our approach, we conduct extensive experiments on our in-house dataset of 308 subjects spanning multiple groups. Our results demonstrate that the proposed tree-based model effectively represents dementia progression, achieves promising performance in intricate classification task of AD and LBD, and highlights discriminative brain regions that contribute to the differentiation between dementia types. Our code is available at https://github.com/tongchen2010/haff.

Accurate modeling of disease progression is essential for comprehending the heterogeneous neuropathologies such as Alzheimer's Disease (AD). Traditional neuroimaging analysis often confound disease effects with normal aging, complicating the differential diagnosis. Recent advancements in deep learning have catalyzed the development of disentanglement techniques in Autoencoder networks, aiming to segregate longitudinal changes attributable to aging from those due to disease-specific alterations within the latent space. However, existing longitudinal disentanglement methods usually model disease as a single axis factor which ignores the complexity and heterogeneity of Alzheimer's Disease. In response to this issue, we propose a novel Surface-based Multi-axis Disentanglement framework.This framework posits multiple disease axes within the latent space, enhancing the model's capacity to encapsulate the multifaceted nature of AD, which includes various disease trajectories. To assign axes to data trajectories without explicit ground truth labels, we implement a longitudinal contrastive loss leveraging self-supervision, thereby refining the separation of disease trajectories. Evaluated on the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset ( $N=1321$ ), our model demonstrates superior performance in delineating between cognitively normal (CN), mild cognitive impairment (MCI), and AD subjects,classification of stable MCI vs converting MCI and Amyloid status, compared to the single-axis model. This is further substantiated through an ablation study on the contrastive loss, underscoring the utility of our multi-axis approach in capturing the complex progression patterns of AD. The code is available at: https://github.com/jianweizhang17/MultiAxisDisentanglement.git.

During the process of brain aging, the changes of white matter structural connectivity are closely correlated with the cognitive traits and brain function. Genes have strong controls over this transition of structural connectivity-altering, which influences brain health and may lead to severe dementia disease, e.g., Alzheimer's disease. In this work, we introduce a novel deep-learning diagram, an oblique genomics mixture of experts(OG-MoE), designed to address the prediction of brain disease diagnosis, with awareness of the structural connectivity changes over time, and coupled with the genomics influences. By integrating genomics features into the dynamic gating router system of MoE layers, the model specializes in representing the structural connectivity components in separate parameter spaces. We pretrained the model on the self-regression task of brain connectivity predictions and then implemented multi-task supervised learning on brain disorder predictions and brain aging prediction. Compared to traditional associations analysis, this work provided a new way of discovering the soft but intricate inter-play between brain connectome phenotypes and genomic traits. It revealed the significant divergence of this correlation between the normal brain aging process and neurodegeneration.