Automated segmentation and reconstruction of patient-specific cardiac anatomy and pathology from in vivo MRI

Jordan Ringenberg, Makarand Deo, Vijay Devabhaktuni, David Filgueiras-Rama, Gonzalo Pizarro, Borja Ibañez, Omer Berenfeld, Pamela Boyers, Jeffrey Gold

Research output: Contribution to journalArticlepeer-review

12 Scopus citations


This paper presents an automated method to segment left ventricle (LV) tissues from functional and delayed-enhancement (DE) cardiac magnetic resonance imaging (MRI) scans using a sequential multi-step approach. First, a region of interest (ROI) is computed to create a subvolume around the LV using morphological operations and image arithmetic. From the subvolume, the myocardial contours are automatically delineated using difference of Gaussians (DoG) filters and GSV snakes. These contours are used as a mask to identify pathological tissues, such as fibrosis or scar, within the DE-MRI. The presented automated technique is able to accurately delineate the myocardium and identify the pathological tissue in patient sets. The results were validated by two expert cardiologists, and in one set the automated results are quantitatively and qualitatively compared with expert manual delineation. Furthermore, the method is patient-specific, performed on an entire patient MRI series. Thus, in addition to providing a quick analysis of individual MRI scans, the fully automated segmentation method is used for effectively tagging regions in order to reconstruct computerized patient-specific 3D cardiac models. These models can then be used in electrophysiological studies and surgical strategy planning.

Original languageEnglish (US)
Article number125405
JournalMeasurement Science and Technology
Issue number12
StatePublished - Dec 2012
Externally publishedYes


  • delayed enhancement MRI
  • image-based modeling
  • ischemic heart disease

ASJC Scopus subject areas

  • Instrumentation
  • Engineering (miscellaneous)
  • Applied Mathematics


Dive into the research topics of 'Automated segmentation and reconstruction of patient-specific cardiac anatomy and pathology from in vivo MRI'. Together they form a unique fingerprint.

Cite this