Electrocardiographic imaging (ECGi)

ECGI Imaging

ECGi is an emerging imaging technique that allows beat-to-beat computational imaging of electrical activity in the heart using noninvasive surface ECG data in combination of patient-specific anatomical information. Over the last 10 years, we have been pushing the boundary of ECGi techniques through a combination of three general thrusts: methodological research to expand the capacity and improve the accuracy of ECGi, technical developments to lower the cost and improve the accessibility of ECGi, and experimental and clinical research to translate ECGi to improve the management of a variety of heart diseases. 

Funding sources: NIH R01HL145590 (2019-2024); NSF ACI-1350374 (2014-2019); NIH R21Hl125998 (2014-2018)

Collaborators: 

  • Drs. John L. Sapp and Milan Horacek, QEII Health Sciences Centre, Nova Scotia Health Authority (Halifax, NS, Canada)
  • Dr. Saman Nazarian, University of Pennsylvania School of Medicine, Philadelphia, PA
  • Drs. Ankur Kapoor and Vivek Singh, Siemens Medical Solutions, Princeton, PA

Students: Xiajun Jiang, Maryam Tolou, Omar Gharbia, Sandesh Ghimire (alumni), Jingjia Xu (alumni), Azar Rahimi (alumni)

Integrating Physics into ECGi

While traditional ECGi techniques image the electrical functioning at the external surface (epicardium) of the heart, we push the boundary for ECGi techniques that image beneath this surface throughout the depth of the heart wall. The main challenge of this problem is the lack of a physically-unique solution. Over the years, we have been developing various optimization and inference methods to overcome this challenge by incorporating physiologically rich knowledge into the reconstruction process. Major research directions and milestones include:

  • Deep learning of the physiological generation process and direct inference (2018-)
  • Adaptation of errors in the prior knowledge when constraining the data-driven inference (2015-)
  • Incorporation of low-dimensional structure of the electrophysiological process (2012-)
  • Incorporation of physiologically rich knowledge about the spatiotemporal electrophysiological process (2006-)

Example publications:

  • Xiajun Jiang, Sandesh Ghimire, Jwala Dhamala, Zhiyuan Li, Prashnna K. Gyawali, and Linwei Wang, Learning Geometry-Dependent and Physics-Based Inverse Image Reconstruction, Medical Image Computing and Computer-Assisted Intervention (MICCAI), 2020, accepted.
  • Sandesh Ghimire, Jwala Dhamala, Prashnna Gyawali, John L. Sapp, B. Milan Horacek, and Linwei Wang, Improving Generalization of Deep Networks for Inverse Reconstruction of Image Sequences, Information Processing in Medical Imaging (IPMI), Hong Kong, 2019, accepted (oral presentation) (url)
  • Sandesh Ghimire, Milan Horack, John L Sapp, and Linwei Wang, Noninvasive Reconstruction of Transmural Transmembrane Potential with Simultaneous Estimation of Prior Model Error, IEEE Transactions on Medical Imaging, 2019
  • Azar Rahimi, John Sapp, Jingjia Xu, Peter Bajorski, Milan Horacek, and Linwei Wang, Examining the Impact of Prior Models in Transmural Electrophysiological Imaging: A Hierarchical Multiple-Model Bayesian Approach, IEEE Transactions on Medical Imaging, 35(1): 229-243, 2016 (url).
  • Jingjia Xu, Azar Rahimi, Fei Gao, and Linwei Wang, Noninvasive Transmural Electrophysiological Imaging Based on Minimization of Total-Variational Functional, IEEE Transactions on Medical Imaging, 33(9): 1860-1874, 2014 (pdf)
  • Linwei Wang, Heye Zhang, Ken C.L. Wong, and Pengcheng Shi: Physiological-Model-Constrained Noninvasive Reconstruction of Volumetric Myocardial Transmembrane Potentials, IEEE Transactions on Biomedical Engineering, vol 57, no 2, pp 296-315, 2010 (Cover page). (pdf).

 

MRI-ECGi of Scar-related Ventricular Tachycardia

Ventricular tachycardia (VT) contributes to over 350,000 sudden deaths each year in the US. Malignant VTs involve an electrical “short circuit” in the heart, formed by narrow channels of surviving tissue inside myocardial scar. An important treatment is to use catheter ablation to "block" the channel that forms the circuit. Effective ablation requires imaging guidance to visualize the VT circuit relative to scar structures in 3D. Unfortunately, with conventional catheter mapping, up to 90% of the VT circuits are too short-lived to be mapped. For the 10% “mappable” VTs, their data are only available during ablation and limited to one ventricular surface. This inadequacy of functional VT data largely limits our knowledge about scar-related VT and ablation strategies, and reduces the ability of clinicians to identify ablation targets and assess ablation outcome. We have a keen interest in advancing ECGi -- as a conjunction to intra-procedural catheter mapping – to provide peri-procedural imaging of functional VT circuits integrated with 3D scar structure. We have been pushing forward our research agenda through a combination of methodological, animal model, and clinical investigations: 

  • Clinical human study: NIH R01HL145590 (2019-2024)
    • Pre-procedural & post-procedural use
    • Integration with scar imaging
  • Animal model validation (2014-2017): NIH R21Hl125998, collaboration with JHU, completed.

Example publications:

  • Linwei Wang, Omar A. Gharbia, B. Milan Horacek, Saman Nazarian, and John L. Sapp, Noninvasive Epicardial and Endocardial Electrocardiographic Imaging for Scar-related Ventricular Tachycardia, EP Europace, 20(FI2):f263-f272, 2018 (url, pdf), 2018.
  • Linwei Wang, Omar A. Gharbia, B. Milan Horacek, and John L. Sapp, Noninvasive Epicardial and Endocardial Electrocardiographic Imaging of Scar-related Ventricular Tachycardia, Journal of Electrocardiology, 49(6):887-893, 2016 (url)
  • Linwei Wang, Fady Dawoud, Sai-Kit Yeung, Pengcheng Shi, Ken C.L. Wong, Huafeng Liu, and Albert C. Lardo, Transmural Imaging of Ventricular Action Potentials and Post-Infarction Scars in Swine Hearts, IEEE Transactions on Medical Imaging, vol 32, no 4, pp 731-747, 2013 (pdf)

 

Making ECGi Less Expensive

Current ECGi techniques requires two types of input data: high-density surface ECG arrays, and patient-specific heart-torso geometrical models derived from medical tomographic scans. We aim to reduce the requirement of ECGi on these data to make it less expensive and more accessible. Two more research directions include:

  • Deep learning enabled Image-less ECGi (2019-)
  • 12-lead based ECGi (2018-)