Maximum likelihood correction of shape bias arising from imaging protocol: Application to cardiac MRI

Pau Medrano-Gracia, David A. Bluemke, Brett R. Cowan, J. Paul Finn, Carissa G. Fonseca, João A.C. Lima, Avan Suinesiaputra, Alistair A. Young

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

To establish a fair comparison between shape models derived from different imaging protocols, a mapping correcting local bias must be applied. In this paper, a multi-dimensional statistical model has been investigated to correct the systematic differences between Steady-State Free Precession (SSFP) and Gradient Recalled Echo (GRE) cardiac MRI protocols. This statistical model makes use of the Maximum Likelihood (ML) approach to estimate the local parameters of the respective GRE and SSFP distributions. Once those parameters are known, a local mapping can be applied. We applied this method to 46 normal volunteers who were imaged with both protocols. The SSFP model was estimated from the corresponding GRE model and validation was performed with leave-one-out experiments. The error was examined in both the local model parameters and the clinically important global mass and volume estimates. Results showed that the systematic bias around the apex and papillary muscles could be locally corrected and that the mapping also provided a global correction in cavity volume (average error of 0.4 ±12.4 ml) and myocardial mass (-1.2 ±11.1 g).

Original languageEnglish (US)
Title of host publicationStatistical Atlases and Computational Models of the Heart
Subtitle of host publicationImaging and Modelling Challenges - Second International Workshop, STACOM 2011, Held in Conjunction with MICCAI 2011, Revised Selected Papers
Pages214-223
Number of pages10
DOIs
StatePublished - Mar 21 2012
Event2nd International Workshop on Statistical Atlases and Computational Models of the Heart: Imaging and Modelling Challenges, STACOM 2011, Held in Conjunction with MICCAI 2011 - Toronto, ON, Canada
Duration: Sep 22 2011Sep 22 2011

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7085 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other2nd International Workshop on Statistical Atlases and Computational Models of the Heart: Imaging and Modelling Challenges, STACOM 2011, Held in Conjunction with MICCAI 2011
Country/TerritoryCanada
CityToronto, ON
Period9/22/119/22/11

Keywords

  • Cardiac Magnetic Resonance Imaging (MRI)
  • Finite Element Modelling
  • Gradient Recalled Echo (GRE)
  • Protocol Correction
  • Statistical Model
  • Steady-State Free Precession (SSFP)

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Maximum likelihood correction of shape bias arising from imaging protocol: Application to cardiac MRI'. Together they form a unique fingerprint.

Cite this