A novel myocardium segmentation approach based on neutrosophic active contour model

Area of Science:

• Medical Imaging
• Cardiology
• Image Processing

Background:

• Accurate myocardium delineation in echocardiography is crucial for diagnosing heart conditions.
• Challenges include low signal-to-noise ratio, poor contrast, and noise in echocardiographic images.
• Automatic detection can assist radiologists and improve diagnostic accuracy.

Purpose of the Study:

• To develop an automated method for detecting the myocardium region in left ventricle myocardial contrast echocardiography (LVMCE) images.
• To aid radiologists in diagnosing heart problems and measuring infarction size.
• To overcome limitations of existing methods in distinguishing myocardium from surrounding tissues.

Main Methods:

• Images are mapped to a neutrosophic similarity (NS) domain using intensity and homogeneity features.
• A neutrosophic active contour model (NACM) is proposed, with its energy function defined by NS values.
• Clustering algorithms initialize the ventricle region for improved accuracy and speed, followed by curve evolution for endocardium and epicardium detection.

Main Results:

• The NACM method demonstrated superior performance compared to eliminated particle swarm optimization (EPSO) and active contour model without edges (ACMWE).
• NACM achieved significantly lower Hdist and AvgDist metrics for both endocardium and epicardium detection.
• For endocardium, NACM resulted in Hdist of 4.55 ± 0.9mm and AvgDist of 0.58 ± 0.18mm, outperforming EPSO and ACMWE.

Conclusions:

• The proposed NACM method accurately detects myocardium in LVMCE images.
• This automated approach is valuable for clinical applications, particularly for measuring myocardial perfusion and infarct size.
• The method offers a significant improvement over existing techniques for cardiac image analysis.