Related Experiment Videos
Automated cardiac MR image segmentation: theory and measurement evaluation
M F Santarelli1, V Positano, C Michelassi
1C.N.R. Institute of Clinical Physiology, Via Moruzzi, 1, Loc. S. Cataldo, 56124 Pisa, Italy. santarel@ifc.cnr.it
Medical Engineering & Physics
|January 23, 2003
Summary
This study introduces an automated method for segmenting cardiac Magnetic Resonance images, accurately identifying both endocardium and epicardium. The novel approach effectively handles grayscale inhomogeneity, improving cardiac image analysis.
Area of Science:
- Medical Imaging
- Image Processing
- Cardiovascular Research
Background:
- Cardiac Magnetic Resonance (CMR) imaging segmentation is crucial for assessing heart function.
- Grayscale inhomogeneity presents a significant challenge in accurate CMR segmentation.
- Existing methods struggle with precise delineation of both endocardial and epicardial borders.
Purpose of the Study:
- To develop and validate a novel automated segmentation method for cardiac MRI.
- To address the challenge of grayscale inhomogeneity in cardiac image analysis.
- To enable accurate tracking of both endocardium and epicardium in cardiac MRI studies.
Main Methods:
- Anisotropic diffusion filtering for selective edge preservation and inhomogeneity reduction.
- Gradient-Vector-Flow (GVF) snake model for robust boundary detection, including concave regions.
- Automated multiphase, multislice segmentation procedure for endocardial and epicardial borders.
Main Results:
- The method successfully reduced grayscale inhomogeneity while preserving critical myocardial edges.
- The GVF snake effectively captured and adapted to complex cardiac boundaries.
- High agreement (P<0.001) was achieved between automated and manual segmentation across 907 cardiac images.
Conclusions:
- The proposed automated segmentation approach is accurate and reliable for cardiac MRI.
- This technique offers a significant advancement for quantitative analysis of cardiac function and perfusion.
- The method demonstrates potential for widespread clinical application in cardiovascular imaging.