Premium
Does Two‐Dimensional Image Reconstruction from Three‐Dimensional Full Volume Echocardiography Improve the Assessment of Left Ventricular Morphology and Function?
Author(s) -
Amzulescu Mihaela Silvia,
Slavich Massimo,
Florian Anca,
Goetschalckx Kaatje,
Voigt JensUwe
Publication year - 2013
Publication title -
echocardiography
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.404
H-Index - 62
eISSN - 1540-8175
pISSN - 0742-2822
DOI - 10.1111/j.1540-8175.2012.01800.x
Subject(s) - intersection (aeronautics) , ventricle , ideal (ethics) , position (finance) , volume (thermodynamics) , artificial intelligence , orientation (vector space) , ventricular function , function (biology) , plane (geometry) , geometry , computer vision , mathematics , computer science , physics , medicine , cardiology , cartography , philosophy , epistemology , finance , quantum mechanics , economics , evolutionary biology , biology , geography
Aims: (i) To investigate geometric differences between apical views of the left ventricle ( LV ) derived from standard 2D echocardiography (std2D) and triplane ( TP ) views, as well as the “ideally” reconstructed 2D (rec2D) views derived from 3D full volume (3 DFV ) acquisitions, and their influence on the assessment of LV morphology and function. (ii) To determine the feasibility and accuracy of the automatic reconstruction of 2D apical views from 3 DFV datasets. Methods and results: In 59 patients with structurally normal, dilated, and hypertrophic hearts, rec2D was reconstructed manually and automatically and compared to std2D, TP , and 3 DFV regarding the image plane orientation (true vs. ideal probe position, plane intersection angles), LV dimensions, volumes, and EF . The ideal probe position deviated from the true one by 6.9 ± 4.1 mm and 9.5 ± 4.5 mm, for manually and automatically rec2D, respectively, regardless of LV geometry. The mean difference± SD between manual and automatic reconstruction was −2.5 ± 4.4 mm. LV long axis was measured minimally, but significantly longer in rec2D than std2D and TP . LV volumes and EF did not differ between methods. The intersection angle of the two‐chamber view and the three‐chamber view with the four‐chamber view for manual and automatic reconstruction was 53°±7° and 129°±7° and 60° and 130°, respectively. Conclusion: Ideal reconstruction of nonforeshortened 2D images from 3 DFV does not lead to a relevant improvement in image geometry or the assessment of LV morphology and function. The automatic reconstruction algorithm deviates only slightly from manual results.