Automatic image-driven segmentation of the ventricles in cardiac cine MRI
Autor: | Max A. Viergever, Wiro J. Niessen, Evert-Jan Vonken, Chris A. Cocosco, Thomas Netsch, Gunnar K. Lund, Alexander Stork |
---|---|
Přispěvatelé: | Radiology & Nuclear Medicine |
Rok vydání: | 2008 |
Předmět: |
Adult
medicine.medical_specialty Heart Diseases Computer science Heart Ventricles Magnetic Resonance Imaging Cine Image processing Image (mathematics) Automation Imaging Three-Dimensional medicine Image Processing Computer-Assisted Humans Radiology Nuclear Medicine and imaging Computer vision Segmentation medicine.diagnostic_test business.industry Magnetic resonance imaging Steady-state free precession imaging Cine mri Automatic segmentation Radiology Artificial intelligence Mr images business Algorithms |
Zdroj: | Journal of Magnetic Resonance Imaging, 28(2), 366-374. John Wiley & Sons Inc. |
ISSN: | 1053-1807 |
Popis: | 3Purpose: To propose and to evaluate a novel method for theautomatic segmentation of the heart’s two ventricles fromdynamic (“cine”) short-axis “steady state free precession”(SSFP) MR images. This segmentation task is of significantclinical importance. Previously published automatedmethods have various disadvantages for routine clinicaluse.Materials and Methods: The proposed method is primarilyimage-driven: it exploits the spatiotemporal informationprovided by modern 3D time SSFP cardiac MRI, andmakes only few and plausible assumptions about the imageacquisition and about the imaged heart. Specifically, themethod does not require previously trained statisticalshape models or gray-level appearance models, as oftenused by other methods.Results: The performance of the segmentation method wasdemonstrated through a qualitative visual validation on 32clinical exams: no gross failures for the left-ventricle (right-ventricle) on 31 (29) of the exams were found. A validationof resulting quantitative cardiac functional parametersshowed good agreement with a manual quantification of 19clinical exams.Conclusion: The proposed method is feasible, fast, androbust against anatomical variability and image contrastvariations.Key Words: cardiac MRI; automatic segmentation; image-driven segmentationJ. Magn. Reson. Imaging 2008;28:366–374.© 2008 Wiley-Liss, Inc. |
Databáze: | OpenAIRE |
Externí odkaz: |