Methods
The deep learning-based algorithm was applied on static carotid ultrasound images obtained from the database hosted by Signal Processing Laboratory \cite{laboratory} . Details about the data set are provided on that website location and reproduced here. The database contains images of common carotid artery (CCA) of ten volunteers (mean age 27.5 ± 3.5 years) with different weight (mean weight 76.5 ± 9.7 kg). Images (usually eight images per volunteer) were acquired with Sonix OP ultrasound scanner with different set-up of depth, gain, time gain compensation (TGC) curve and different linear array transducers. The image database contains 84 B-mode ultrasound images of CCA in longitudinal section. The resolution of images is approximately 390x330 pixels. Two different linear array transducers with different frequencies (10MHz and 14MHz) were used. These frequencies were chosen because of their suitability for superficial vascular scanning. All images were taken by specialists with five years’ experience in vascular ultrasound. Images were captured in accordance to the standard protocol with patients lying in the supine position and with the neck rotated to the left side while the right CCA was examined.