ValidEIT - Validation of regional lung perfusion based on electrical impedance tomography (EIT) by computed tomography (CT) and invasive flow measurement (Swan-Ganz catheter)

• Applicants: Professor Dr.-Ing. Steffen Leonhardt; Privatdozent Dr. Sebastian Reinartz
• Subject Area: Medical Physics, Biomedical Technology ; Biomedical Systems Technology ; Nuclear Medicine, Radiotherapy, Radiobiology
• Term: from 2019 to 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 422367304

• Final Report Year: 2023

Final Report Abstract

The ValidEIT project aimed at enabling a non-invasive measurement of regional lung perfusion using Electrical Impedance Tomography (EIT). The project focused on understanding the cardiac components of the EIT signal to estimate regional perfusion. This was achieved through animal experiments, modelling, and algorithm development. Fifteen pigs were used to create a cardiopulmonary injury model, and EIT measurements were taken alongside reliable regional perfusion measurements from CT-P scans. The experiments involved inducing central to peripheral occlusions and modulating cardiac output. In silico simulations using dynamic finite element models were developed to understand the origins of the EIT cardiac-related signal (CRS) and generate realistic EIT signals. An innovative approach was used to analyze the data, employing nonlinear parametric semi-negative matrix factorization to decompose the signals and improve estimation accuracy. Traditional and Deep Learning approaches for source separation were compared. Harmonic-seeking algorithms proved to be more accurate, intelligible, and suitable for real-time implementation. The analysis of experimental data revealed that while the CRS signal is a good indicator of regional perfusion, its usefulness decreases as impairments become more distal. Overall, the project has advanced the state-of-the-art in the field, providing tools that are attractive for future clinical applications and generating valuable data for further advancements.

Publications

• “Prior-Based Unsupervised Separation of Cardiac-Related Signals,” in International Conference on Biomedical Applications of Electrical Impedance Tomography, Galway, 2021.
  D. F. Silva, U. Özmü, S. Leonhardt & C. Ngo
  
• “Top-to-Bottom Generator of EIT and CT Bolus Signals,” in International Conference on Bioelectromagnetism, Electrical Bioimpedance, and Electrical Impedance Tomography, Seoul, 2022.
  D. F. Silva, S. Leonhardt & C. Ngo
  
• “Unsupervised Clustering of Lung Perfusion Features,” in International Conference on Bioelectromagnetism, Electrical Bioimpedance, and Electrical Impedance Tomography, Seoul, 2022.
  D. F. Silva, C. Wemmer, M. Wolter, S. Leonhardt & C. Ngo
  
• A Real-Time Dual Heart and Respiratory Rate Estimator for Electrical Impedance Tomography. 2023 IEEE 36th International Symposium on Computer-Based Medical Systems (CBMS), 426-430. IEEE.
  Silva, Diogo Filipe & Leonhardt, Steffen
  
• Fast 4D FEM Model for EIT Source Separation Benchmarking. Current Directions in Biomedical Engineering, 9(1), 387-390.
  Silva, Diogo Filipe & Leonhardt, Steffen
  
• Hybrid Compartment Model Formulation for Accelerated Bolus Fitting. 2023 31st European Signal Processing Conference (EUSIPCO), 1120-1124. IEEE.
  Silva, Diogo Filipe & Leonhardt, Steffen
  
• “Estimation of Planar Direction of Flow as Additional Metric for EIT Perfusion Analysis,” in International Conference on Biomedical Applications of Electrical Impedance Tomography, Aachen, 2023.
  D. F. Silva & S. Leonhardt