Proceedings. IEEE International Symposium on Biomedical Imaging最新文献

Parkinson's Disease (PD) affects millions globally, impacting movement. Prior research utilized deep learning for PD prediction, primarily focusing on medical images, neglecting the data's underlying manifold structure. This work proposes a multimodal approach encompassing both image and non-image features, leveraging contrastive cross-view graph fusion for PD classification. We introduce a novel multimodal co-attention module, integrating embeddings from separate graph views derived from low-dimensional representations of images and clinical features. This enables more robust and structured feature extraction for improved multi-view data analysis. Additionally, a simplified contrastive loss-based fusion method is devised to enhance cross-view fusion learning. Our graph-view multimodal approach achieves an accuracy of 91% and an area under the receiver operating characteristic curve (AUC) of 92.8% in five-fold cross-validation. It also demonstrates superior predictive capabilities on non-image data compared to solely machine learning-based methods.

Real-time dynamic MRI is important for visualizing time-varying processes in several applications, including cardiac imaging, where it enables free-breathing images of the beating heart without ECG gating. However, current real-time MRI techniques commonly face challenges in achieving the required spatio-temporal resolutions due to limited acceleration rates. In this study, we propose a deep learning (DL) technique for improving the estimation of stationary outer-volume signal from shifted time-interleaved undersampling patterns. Our approach utilizes the pseudo-periodic nature of the ghosting artifacts arising from the moving organs. Subsequently, this estimated outer-volume signal is subtracted from individual timeframes of the real-time MR time series, and each timeframe is reconstructed individually using physics-driven DL methods. Results show improved image quality at high acceleration rates, where conventional methods fail.

Morphometricity examines the global statistical association between brain morphology and an observable trait, and is defined as the proportion of the trait variation attributable to brain morphology. In this work, we propose an accurate morphometricity estimator based on the generalized random effects (GRE) model, and perform morphometricity analyses on five cognitive traits in an Alzheimer's study. Our empirical study shows that the proposed GRE model outperforms the widely used LME model on both simulation and real data. In addition, we extend morphometricity estimation from the whole brain to the focal-brain level, and examine and quantify both global and regional neuroanatomical signatures of the cognitive traits. Our global analysis reveals 1) a relatively strong anatomical basis for ADAS13, 2) intermediate ones for MMSE, CDRSB and FAQ, and 3) a relatively weak one for RAVLT.learning. The top associations identified from our regional morphometricity analysis include those between all five cognitive traits and multiple regions such as hippocampus, amygdala, and inferior lateral ventricles. As expected, the identified regional associations are weaker than the global ones. While the whole brain analysis is more powerful in identifying higher morphometricity, the regional analysis could localize the neuroanatomical signatures of the studied cognitive traits and thus provide valuable information in imaging and cognitive biomarker discovery for normal and/or disordered brain research.

Skull-stripping is the removal of background and non-brain anatomical features from brain images. While many skull-stripping tools exist, few target pediatric populations. With the emergence of multi-institutional pediatric data acquisition efforts to broaden the understanding of perinatal brain development, it is essential to develop robust and well-tested tools ready for the relevant data processing. However, the broad range of neuroanatomical variation in the developing brain, combined with additional challenges such as high motion levels, as well as shoulder and chest signal in the images, leaves many adult-specific tools ill-suited for pediatric skull-stripping. Building on an existing framework for robust and accurate skull-stripping, we propose developmental SynthStrip (d-SynthStrip), a skull-stripping model tailored to pediatric images. This framework exposes networks to highly variable images synthesized from label maps. Our model substantially outperforms pediatric baselines across scan types and age cohorts. In addition, the <1-minute runtime of our tool compares favorably to the fastest baselines. We distribute our model at https://w3id.org/synthstrip.

U-shaped networks and its variants have demonstrated exceptional results for medical image segmentation. In this paper, we propose a novel dual self-distillation (DSD) framework for U-shaped networks for 3D medical image segmentation. DSD distills knowledge from the ground-truth segmentation labels to the decoder layers and also between the encoder and decoder layers of a single U-shaped network. DSD is a generalized training strategy that could be attached to the backbone architecture of any U-shaped network to further improve its segmentation performance. We attached DSD on two state-of-the-art U-shaped backbones, and extensive experiments on two public 3D medical image segmentation datasets demonstrated significant improvement over those backbones, with negligible increase in trainable parameters and training time. The source code is publicly available at https://github.com/soumbane/DualSelfDistillation.

Positron Emission Tomography (PET) is a valuable imaging method for studying molecular-level processes in the body, such as hyperphosphorylated tau (p-tau) protein aggregates, a hallmark of several neurodegenerative diseases including Alzheimer's disease. P-tau density and cerebral perfusion can be quantified from PET data using tracer kinetic modeling techniques. However, noise in PET images leads to uncertainty in the estimated kinetic parameters. This can be quantified in a Bayesian framework by the posterior distribution of kinetic parameters given PET measurements. Markov Chain Monte Carlo (MCMC) techniques can be employed to estimate the posterior distribution, although with significant computational needs. In this paper, we propose to leverage deep learning inference efficiency to infer the posterior distribution. A novel approach using denoising diffusion probabilistic model (DDPM) is introduced. The performance of the proposed method was evaluated on a [18F]MK6240 study and compared to an MCMC method. Our approach offered significant reduction in computation time (over 30 times faster than MCMC) and consistently predicted accurate (< 0.8 % mean error) and precise (< 5.77 % standard deviation error) posterior distributions.

Children with Autism Spectrum Disorder (ASD) frequently exhibit comorbid anxiety, which contributes to impairment and requires treatment. Therefore, it is critical to investigate co-occurring autism and anxiety with functional imaging tools to understand the brain mechanisms of this comorbidity. Multidimensional Anxiety Scale for Children, 2nd edition (MASC-2) score is a common tool to evaluate the daily anxiety level in autistic children. Predicting MASC-2 score with Functional Magnetic Resonance Imaging (fMRI) data will help gain more insights into the brain functional networks of children with ASD complicated by anxiety. However, most of the current graph neural network (GNN) studies using fMRI only focus on graph operations but ignore the spectral features. In this paper, we explored the feasibility of using spectral features to predict the MASC-2 total scores. We proposed SpectBGNN, a graph-based network, which uses spectral features and integrates graph spectral filtering layers to extract hidden information. We experimented with multiple spectral analysis algorithms and compared the performance of the SpectBGNN model with CPM, GAT, and BrainGNN on a dataset consisting of 26 typically developing and 70 ASD children with 5-fold cross-validation. We showed that among all spectral analysis algorithms tested, using the Fast Fourier Transform (FFT) or Welch's Power Spectrum Density (PSD) as node features performs significantly better than correlation features, and adding the graph spectral filtering layer significantly increases the network's performance.

Normative models in neuroimaging learn the brain patterns of healthy population distribution and estimate how disease subjects like Alzheimer's Disease (AD) deviate from the norm. Existing variational autoencoder (VAE)-based normative models using multimodal neuroimaging data aggregate information from multiple modalities by estimating product or averaging of unimodal latent posteriors. This can often lead to uninformative joint latent distributions which affects the estimation of subject-level deviations. In this work, we addressed the prior limitations by adopting the Mixture-of-Product-of-Experts (MoPoE) technique which allows better modelling of the joint latent posterior. Our model labelled subjects as outliers by calculating deviations from the multimodal latent space. Further, we identified which latent dimensions and brain regions were associated with abnormal deviations due to AD pathology.

Parallel transmission (pTx) is an important technique for reducing transmit field inhomogeneities at ultrahigh-field (UHF) MRI. pTx typically involves solving an optimization problem for radiofrequency pulse design, with hard constraints on specific-absorption rate (SAR) and/or power, which may be time-consuming. In this work, we propose a novel approach towards incorporating hard constraints to physics-driven neural networks. Our method unrolls an extension of the log-barrier method, where the central path problems are solved via the gradient descent method whose optimal step sizes are learned with a neural network. Results indicate that our method is substantially faster compared to traditional convex optimization techniques, while achieving similar performance.

Digital Breast Tomosynthesis (DBT) is a widely used medical imaging modality for breast cancer screening and diagnosis, offering higher spatial resolution and greater detail through its 3D-like breast volume imaging capability. However, the increased data volume also introduces pronounced data imbalance challenges, where only a small fraction of the volume contains suspicious tissue. This further exacerbates the data imbalance due to the case-level distribution in real-world data and leads to learning a trivial classification model that only predicts the majority class. To address this, we propose a novel method using view-level contrastive Self-supervised Initialization and Fine-Tuning for identifying abnormal DBT images, namely SIFT-DBT. We further introduce a patch-level multi-instance learning method to preserve spatial resolution. The proposed method achieves 92.69% volume-wise AUC on an evaluation of 970 unique studies.