Enhancing clot composition analysis in acute ischemic stroke with non-contrast spectral CT: A phantom-based study

Abstract

Acute ischemic stroke (AIS), caused by blood flow obstruction due to a clot, is a leading medical emergency, with mechanical thrombectomy (MT) being the standard of care intervention. The success of MT partly depends on the ability to effectively assess clot composition, as different clot types (e.g., erythrocyte-rich and fibrin-rich) can influence treatment outcomes. Conventional CT imaging struggles with accurately differentiating clot subtypes. Recent advancements in spectral CT imaging offer promising alternatives, but their diagnostic value remains underexplored. This study evaluates the potential of using electron density (ED) and effective atomic number (Z-effective) maps from non-contrast spectral CT (NCCT), that is, without iodine contamination for classifying clot compositions in a phantom model. Two experimental phases were conducted, with various clot analogs prepared in silicone tubes of varying diameters. Spectral CT scans were performed using a dual-layer detector system, generating conventional CT, ED, and Z-effective maps, along with virtual monoenergetic (MonoE) images. Our results revealed that ED and high-energy MonoE images (200 keV) provided superior discrimination of clot types, with larger effect sizes and higher area-under-curve (AUC) values compared to conventional CT and low-energy MonoE (40 keV) or Z-effective maps. These findings suggest that ED and high MonoE images enhance clot characterization and may improve treatment planning in AIS. However, smaller tube diameters and the inclusion of skull-equivalent material in the phantom model reduced the sensitivity and specificity of the spectral methods. This study highlights the potential of spectral CT to refine clot assessment in AIS, emphasizing the need for further clinical validation.