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 journalArticle

11 Citations (Scopus)

Abstract

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
Volume23
Issue number12
DOIs
StatePublished - Dec 2012

Fingerprint

anatomy
Magnetic Resonance Imaging
pathology
Anatomy
Pathology
Magnetic resonance
Cardiac
magnetic resonance
Segmentation
Imaging techniques
Left Ventricle
Tissue
Enhancement
Gaussian Filter
Morphological Operations
myocardium
Fibrosis
snakes
fibrosis
scars

Keywords

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

ASJC Scopus subject areas

  • Instrumentation
  • Engineering (miscellaneous)
  • Applied Mathematics

Cite this

Automated segmentation and reconstruction of patient-specific cardiac anatomy and pathology from in vivo MRI. / Ringenberg, Jordan; Deo, Makarand; Devabhaktuni, Vijay; Filgueiras-Rama, David; Pizarro, Gonzalo; Ibañez, Borja; Berenfeld, Omer; Boyers, Pamela; Gold, Jeffrey.

In: Measurement Science and Technology, Vol. 23, No. 12, 125405, 12.2012.

Research output: Contribution to journalArticle

Ringenberg, Jordan ; Deo, Makarand ; Devabhaktuni, Vijay ; Filgueiras-Rama, David ; Pizarro, Gonzalo ; Ibañez, Borja ; Berenfeld, Omer ; Boyers, Pamela ; Gold, Jeffrey. / Automated segmentation and reconstruction of patient-specific cardiac anatomy and pathology from in vivo MRI. In: Measurement Science and Technology. 2012 ; Vol. 23, No. 12.
@article{dbc4d78e79084d2594cd7f514b89ee88,
title = "Automated segmentation and reconstruction of patient-specific cardiac anatomy and pathology from in vivo MRI",
abstract = "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.",
keywords = "delayed enhancement MRI, image-based modeling, ischemic heart disease",
author = "Jordan Ringenberg and Makarand Deo and Vijay Devabhaktuni and David Filgueiras-Rama and Gonzalo Pizarro and Borja Iba{\~n}ez and Omer Berenfeld and Pamela Boyers and Jeffrey Gold",
year = "2012",
month = "12",
doi = "10.1088/0957-0233/23/12/125405",
language = "English (US)",
volume = "23",
journal = "Measurement Science and Technology",
issn = "0957-0233",
publisher = "IOP Publishing Ltd.",
number = "12",

}

TY - JOUR

T1 - Automated segmentation and reconstruction of patient-specific cardiac anatomy and pathology from in vivo MRI

AU - Ringenberg, Jordan

AU - Deo, Makarand

AU - Devabhaktuni, Vijay

AU - Filgueiras-Rama, David

AU - Pizarro, Gonzalo

AU - Ibañez, Borja

AU - Berenfeld, Omer

AU - Boyers, Pamela

AU - Gold, Jeffrey

PY - 2012/12

Y1 - 2012/12

N2 - 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.

AB - 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.

KW - delayed enhancement MRI

KW - image-based modeling

KW - ischemic heart disease

UR - http://www.scopus.com/inward/record.url?scp=84870334697&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84870334697&partnerID=8YFLogxK

U2 - 10.1088/0957-0233/23/12/125405

DO - 10.1088/0957-0233/23/12/125405

M3 - Article

AN - SCOPUS:84870334697

VL - 23

JO - Measurement Science and Technology

JF - Measurement Science and Technology

SN - 0957-0233

IS - 12

M1 - 125405

ER -