Label-Free Physics-Informed Image Sequence Reconstruction with Disentangled Spatial-Temporal Modeling

Jiang, Xiajun; Missel, Ryan; Toloubidokhti, Maryam; Li, Zhiyuan; Gharbia, Omar A.; Sapp, John L.; Wang, Linwei

doi:10.1007/978-3-030-87231-1_35

Cited by 8 publications

(3 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Context-Set Feed-Forward Embedding: We use the meta-model q ζ (c|D c ) to get the feed-forward embedding for each context set. First, each sample y 1:T ∈ D c is embedded through a neural function h φ (y 1:T ) that uses a GCN-GRU cell [16] to obtain the sequential information from the graph, and aggregate it across time with a linear layer. We then average all latent embedding in D c : which then parameterizes q ζ = N (µ c , σ 2 c ) via two separate linear layers.…”

Section: Meta-model For Amortized Variational Inferencementioning

confidence: 99%

Few-shot Generation of Personalized Neural Surrogates for Cardiac Simulation via Bayesian Meta-Learning

Jiang,

Li,

Missel

et al. 2022

Preprint

View full text Add to dashboard Cite

Clinical adoption of personalized virtual heart simulations faces challenges in model personalization and expensive computation. While an ideal solution is an efficient neural surrogate that at the same time is personalized to an individual subject, the state-of-the-art is either concerned with personalizing an expensive simulation model, or learning an efficient yet generic surrogate. This paper presents a completely new concept to achieve personalized neural surrogates in a single coherent framework of meta-learning (metaPNS). Instead of learning a single neural surrogate, we pursue the process of learning a personalized neural surrogate using a small amount of context data from a subject, in a novel formulation of few-shot generative modeling underpinned by: 1) a setconditioned neural surrogate for cardiac simulation that, conditioned on subject-specific context data, learns to generate query simulations not included in the context set, and 2) a meta-model of amortized variational inference that learns to condition the neural surrogate via simple feed-forward embedding of context data. As test time, metaPNS delivers a personalized neural surrogate by fast feed-forward embedding of a small and flexible number of data available from an individual, achieving -for the first time -personalization and surrogate construction for expensive simulations in one end-to-end learning framework. Synthetic and real-data experiments demonstrated that metaPNS was able to improve personalization and predictive accuracy in comparison to conventionallyoptimized cardiac simulation models, at a fraction of computation.

show abstract

Section: Meta-model For Amortized Variational Inferencementioning

confidence: 99%

Few-shot Generation of Personalized Neural Surrogates for Cardiac Simulation via Bayesian Meta-Learning

Jiang,

Li,

Missel

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…These methods only modeled the static relationship from TMP to BSPM at a single moment in time without considering the dynamic activation information, resulting in a poorer TMP recovery performance. There is also work (Jiang et al 2021) that use the forward transformation matrix H as the objective function for unsupervised learning, which considers the dynamic activation process of the encoded latent space, but does not take into account the dynamic activation process of the TMP itself, which is actually still a static reconstruction.…”

Section: Introductionmentioning

confidence: 99%

Neural network aided extended Kalman filtering for inverse imaging of cardiac transmembrane potential

Ran,

Hu,

Huang

et al. 2024

Phys. Med. Biol.

View full text Add to dashboard Cite

Objective: The aim of this study is to address the limitations in reconstructing the electrical activity of the heart from the body surface electrocardiogram, which is an ill-posed inverse problem. Current methods often assume values commonly used in the literature in the absence of a priori knowledge, leading to errors in the model. Furthermore, most methods ignore the dynamic activation process inherent in cardiomyocytes during the cardiac cycle.Approach: To overcome these limitations, we propose an extended Kalman filter (EKF)-based neural network approach to dynamically reconstruct cardiac transmembrane potential. Specifically, a recurrent neural network is used to establish the state estimation equation of the EKF, while a convolutional neural network is used as the measurement equation. The Jacobi matrix of the network undergoes a correction feedback process to obtain the Kalman gain.Main Results: After repeated iterations, the final estimated state vector, i.e., the reconstructed image of the transmembrane potential, is obtained. The results from both the final simulation and real experiments demonstrate the robustness and accurate quantification of the model.Significance: This study presents a new approach to cardiac transmembrane potential reconstruction that offers higher accuracy and robustness compared to traditional methods. The use of neural networks and EKFs allows dynamic modelling that takes into account the activation processes inherent in cardiomyocytes and does not require a priori knowledge of inputs such as forward transition matrices.

show abstract

“…Deterministic approaches in cardiac electrophysiology involve minimizing a cost function that quantifies the discrepancy between the observed data and the model predictions. For robust inverse, spatial and/ or temporal regularization [13], [14] and physics-informed regularization [15], [16] have been widely used. Probabilistic methods rely on Bayesian inference theory and numerical techniques to generate posterior distributions for the model parameters [17], [18].…”

Section: Introductionmentioning

confidence: 99%

Deep Computational Model for the Inference of Ventricular Activation Properties

Camps

Banerjee

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Myocardial infarction (MI) demands precise and swift diagnosis. Cardiac digital twins (CDTs) have the potential to offer individualized evaluation of cardiac function in a non-invasive manner, making them a promising approach for personalized diagnosis and treatment planning of MI. The inference of accurate myocardial tissue properties is crucial in creating a reliable CDT platform, and particularly in the context of studying MI. In this work, we investigate the feasibility of inferring myocardial tissue properties from the electrocardiogram (ECG), focusing on the development of a comprehensive CDT platform specifically designed for MI.The platform integrates multi-modal data, such as cardiac MRI and ECG, to enhance the accuracy and reliability of the inferred tissue properties. We perform a sensitivity analysis based on computer simulations, systematically exploring the effects of infarct location, size, degree of transmurality, and electrical activity alteration on the simulated QRS complex of ECG, to establish the limits of the approach. We subsequently propose a deep computational model to infer infarct location and distribution from the simulated QRS. The in silico experimental results show that our model can effectively capture the complex relationships between the QRS signals and the corresponding infarct regions, with promising potential for clinical application in the future. The code will be released publicly once the manuscript is accepted for publication.

show abstract

Label-Free Physics-Informed Image Sequence Reconstruction with Disentangled Spatial-Temporal Modeling

Cited by 8 publications

References 19 publications

Few-shot Generation of Personalized Neural Surrogates for Cardiac Simulation via Bayesian Meta-Learning

Few-shot Generation of Personalized Neural Surrogates for Cardiac Simulation via Bayesian Meta-Learning

Neural network aided extended Kalman filtering for inverse imaging of cardiac transmembrane potential

Deep Computational Model for the Inference of Ventricular Activation Properties

Contact Info

Product

Resources

About