Time continuous tracking and segmentation of cardiovascular magnetic resonance images using multidimensional dynamic...

Area of Science:

• Medical Imaging
• Cardiovascular Imaging
• Computational Anatomy

Background:

• Accurate delineation of cardiac structures in magnetic resonance imaging (MRI) is crucial for quantitative analysis.
• Existing methods for cardiac contour detection face challenges in achieving time-continuous accuracy across all cardiac phases.

Purpose of the Study:

• To develop and evaluate a semiautomatic, time-continuous contour detection method for cardiac MRI.
• To assess the accuracy and clinical applicability of the proposed method for cardiac volume and ejection fraction quantification.

Main Methods:

• A semiautomatic method based on multidimensional dynamic programming for time-continuous contour detection.
• Shape parameterization and cost hypercube generation using image-derived cost functions.
• Optimal path searching through hypercubes with adjustable constraints on parameter changes.

Main Results:

• Achieved average border positioning errors of 1.77 mm (epicardial) and 1.86 mm (endocardial) across all subjects and phases.
• Demonstrated low average errors in end-diastolic and end-systolic volumes (-4.36 to 4.24 mL).
• Reported an average error of 4.82% in ejection fraction, comparable to state-of-the-art methods.

Conclusions:

• The proposed multidimensional dynamic programming method offers accurate and robust time-continuous contour detection in cardiac MRI.
• The method shows significant potential for reliable application in routine clinical practice for cardiac assessment.
• Quantitative results indicate high precision in volume and ejection fraction measurements, supporting clinical utility.