IMPACT OF CONVOLUTION FILTERS ON CORONARY PLAQUE CHARACTERIZATION IN MULTISLICE COMPUTED TOMOGRAPHY

Abstract

Objective: This study aims to evaluate the impact of different convolution filters on the variability of coronary plaque attenuation in ex vivo left coronary arteries imaged using Multislice Computed Tomography (AC-CTMS).

Methods:  The research involved three post-mortem left coronary arteries, each introduced with a diluted contrast medium and imaged under standardized conditions using four convolution filters (B30f, B36f, B46f, B60f). Identification of plaque and attenuation measurements in Hounsfield Units (HU) were performed across various regions of interest (ROIs), including the Lumen, surrounding Oil (simulating epicardial fat), and calcified and noncalcified plaque components.

Results: The study found significant variability in attenuation values across the convolution filters (p<0.001). Statistical analysis revealed marked differences in the precision of structural differentiation between the plaque components based on the sharpness of the filter used. The sharper filters (B46f, B60f) provided more transparent and abrupt transitions among different plaque structures than the softer filters (B30f, B36f).

Conclusions: Convolution filter selection critically affects the attenuation values of coronary plaques in AC-CTMS and subsequently influences the accuracy and reliability of plaque characterization. The results underscore the need for careful selection of convolution filters in clinical settings to enhance the assessment of coronary artery disease and plaque vulnerability. Further research with more diverse samples is essential to confirm these findings and their applicability in clinical practice.