Left ventricular myocardium segmentation on delayed phase of multi-detector row computed tomography

Area of Science:

• Medical Imaging
• Cardiology
• Computational Anatomy

Background:

• Advanced ischemic heart disease often leads to left ventricular (LV) myocardial volume loss.
• Abnormal enhancement patterns in delayed-phase multi-detector row computed tomography (MDCT) are characteristic of this condition.
• Accurate estimation of LV myocardial volume is crucial for diagnosis and treatment planning.

Purpose of the Study:

• To develop and evaluate an adaptive segmentation method for contouring the myocardium in delayed-phase MDCT.
• To enable accurate computation of LV myocardial volume from these images.
• To provide a tool for radiologists and physicians to assist in volume estimation.

Main Methods:

• Anisotropic diffusion filtering preprocesses MDCT images to enhance contrast and reduce noise.
• Two contouring modes are used on a lead slice, with region-growing and convex-hull algorithms refining contours across slices.
• Morphological operators and width properties are utilized for final LV myocardial volume calculation.

Main Results:

• The proposed method was evaluated on 27 healthy patients.
• Computer simulations demonstrated that the method reliably identifies myocardial contours comparable to manual delineations by experienced experts.
• The results indicate high stability and accuracy in contouring.

Conclusions:

• The developed adaptive segmentation method provides robust contouring of the LV myocardium on delayed-phase MDCT.
• This technique has the potential to significantly reduce the time and effort required for manual contour sketching.
• It offers a precise and stable approach for LV myocardial volume assessment.