A computer vision deep learning segmentation approach successfully detected carotid arteries with 98% accuracy and identified stenosis with 90% accuracy in ultrasound images.
Does a computer vision deep learning method accurately detect carotid arteries and stenoses in ultrasound images?
Computer vision deep learning techniques demonstrate high accuracy in detecting carotid arteries and stenoses on ultrasound, offering a potential tool for automated vascular disease diagnosis.
With the change of living conditions and dietary habits, there has been an increase in vascular diseases in recent years. These diseases often lead to peripheral narrowing and occlusion of the arteries. In the management of vascular diseases, which can lead to serious consequences such as stroke and even death, it is very valuable to use appropriate imaging techniques for early and accurate diagnosis. However, the large and complex vascular network makes it difficult to analyze accurate data during diagnostic procedures. In our study, we aimed to easily, quickly and effectively recognize carotid arteries and stenoses ultrasonographically using computer vision deep learning techniques. A dataset containing 3 different levels of the carotid artery was created by obtaining 3401 US images from 120 cases whose age and gender were randomly selected. The width of the carotid artery and the vessels where stenosis was detected were measured using the computer vision deep learning method. Computer vision deep learning segmentation successfully detected carotid arteries with a rate of 98% and stenosis in the carotid artery with a rate of 90%, which is considered high.
Bilici et al. (Mon,) conducted a other in Carotid artery stenosis (n=120). Computer vision deep learning method (U-Net) and Gaussian-based adaptive thresholding vs. Manual labeling by radiologist was evaluated on Accuracy of carotid artery detection. A computer vision deep learning segmentation approach successfully detected carotid arteries with 98% accuracy and identified stenosis with 90% accuracy in ultrasound images.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: