Knowledge directed left ventricular boundary detection in equilibrium radionuclide angiocardiography

Area of Science:

• Medical Imaging
• Cardiovascular Imaging
• Image Analysis

Background:

• Accurate delineation of the left ventricular (LV) boundary is crucial for assessing cardiac function.
• Traditional methods for LV boundary detection can be time-consuming and operator-dependent.
• Equilibrium radionuclide angiocardiography (ERNA) provides functional information but requires robust image analysis techniques.

Purpose of the Study:

• To develop and validate an automated method for left ventricular (LV) boundary detection in equilibrium radionuclide angiocardiographic (ERNA) images.
• To improve the efficiency and objectivity of LV boundary delineation in cardiac imaging studies.
• To demonstrate the applicability of the method across different ERNA views and temporal sequences.

Main Methods:

• Utilized directional gradient edge operators and an edge point linking scheme for lower-level image analysis.
• Implemented a knowledge-based strategy with heuristic cost functions incorporating local and global information for higher-level boundary finding.
• Applied the methodology to end-diastolic frames and tracked LV boundaries through entire temporal sequences in multiple ERNA views.

Main Results:

• Successfully automated the detection of the left ventricular (LV) boundary in various equilibrium radionuclide angiocardiographic (ERNA) image views.
• Demonstrated accurate LV boundary delineation in a detailed example from a left lateral view ERNA sequence.
• Illustrated the method's capability to follow LV boundaries throughout temporal sequences in both left lateral and left anterior oblique ERNA studies.

Conclusions:

• The proposed automated method effectively delineates the left ventricular (LV) boundary in equilibrium radionuclide angiocardiographic (ERNA) images.
• This approach offers a reliable and efficient tool for quantitative analysis of cardiac function from ERNA studies.
• The knowledge-based strategy enhances boundary detection accuracy and robustness across different imaging scenarios.