Solving Inaccuracies in Anatomical Models for Electrocardiographic Inverse Problem Resolution by Maximizing Reconstruction Quality

IEEE Trans Med Imaging. 2018 Mar;37(3):733-740. doi: 10.1109/TMI.2017.2707413. Epub 2017 May 23.


Electrocardiographic Imaging has become an increasingly used technique for non-invasive diagnosis of cardiac arrhythmias, although the need for medical imaging technology to determine the anatomy hinders its introduction in the clinical practice. This paper explores the ability of a new metric based on the inverse reconstruction quality for the location and orientation of the atrial surface inside the torso. Body surface electrical signals from 31 realistic mathematical models and four AF patients were used to estimate the optimal position of the atria inside the torso. The curvature of the L-curve from the Tikhonov method, which was found to be related to the inverse reconstruction quality, was measured after application of deviations in atrial position and orientation. Independent deviations in the atrial position were solved by finding the maximal L-curve curvature with an error of 1.7 ± 2.4 mm in mathematical models and 9.1 ± 11.5 mm in patients. For the case of independent angular deviations, the error in location by using the L-curve was 5.8±7.1° in mathematical models and 12.4° ± 13.2° in patients. The ability of the L-curve curvature was tested also under superimposed uncertainties in the three axis of translation and in the three axis of rotation, and the error in location was of 2.3 ± 3.2 mm and 6.4° ± 7.1° in mathematical models, and 7.9±10.7 mm and 12.1°±15.5° in patients. The curvature of L-curve is a useful marker for the atrial position and would allow emending the inaccuracies in its location.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adult
  • Aged
  • Algorithms
  • Electrocardiography / methods*
  • Female
  • Heart Atria / diagnostic imaging*
  • Humans
  • Image Processing, Computer-Assisted / methods*
  • Middle Aged
  • Models, Anatomic*