Myocardial border detection by branch-and-bound dynamic programming in magnetic resonance images

Area of Science:

• Medical imaging analysis
• Computational algorithms
• Cardiovascular system imaging

Background:

• Magnetic resonance imaging (MRI) provides clear, noninvasive visualization but generates large datasets.
• Manual measurement of dynamic cardiovascular MRI is impractical due to data volume.
• Efficient algorithms are crucial for online computer-aided diagnosis using MRI.

Purpose of the Study:

• To apply a branch-and-bound dynamic programming (DP) technique for detecting left ventricular endocardial borders.
• To reduce the computational time required for border detection in cardiac MRI.
• To evaluate the performance of the proposed method against conventional techniques.

Main Methods:

• Implementation of a branch-and-bound dynamic programming algorithm.
• Application to detect endocardial borders in left ventricular MRI.
• Comparison of computational time with exhaustive search methods using statistical tests.

Main Results:

• The branch-and-bound DP technique significantly reduces computational time compared to exhaustive search.
• Statistical tests confirm the CPU time performance advantage of the proposed method.
• The algorithm demonstrates effectiveness in analyzing dynamic cardiac MRI data.

Conclusions:

• The branch-and-bound DP method offers a faster and more efficient approach for endocardial border detection in cardiac MRI.
• This advancement supports the development of effective online computer-aided measurement and diagnosis systems.
• The technique addresses the challenges of analyzing large datasets from dynamic organ imaging.