Carotid automated ultrasound double line extraction system (CADLES) via Edge-Flow

Abstract

This paper presents a completely user-independent algorithm, that automatically extracts the far (distal) double line (lumen-intima and media-adventitia) in the carotid artery using an Edge Flow technique (a class of AtheroEdge™ systems) based on directional probability maps using the attributes of intensity and texture. The extracted double line translates into a measure of the intima-media thickness (IMT), a validated marker for the progression of atherosclerosis. The Carotid Automated Double Line Extraction System based on Edge-Flow (CADLES-EF) is characterized and validated by comparing the output of the algorithm with two other completely automatic techniques (CALEXia and CULEXsa) published by the same authors. Validation was performed on a multi-institutional database of 300 longitudinal B-mode carotid images with normal and pathologic arteries. CADLES-EF showed an intima-media thickness (IMT) bias of 0.043 ± 0.097 mm in comparison to CALEXia and CULEXsa that showed 0.134 ± 0.0.88 mm and 0.74 ± 0.092 mm, respectively. The system's Figure of Merit (FoM) showed an improvement when compared to previous automated methods: CALEXia and CULEXsa, leading to values of 84.7%, 91.5%, while our new approach, CADLES-EF performed the best with 94.8%.