banner image
No data available.
Please log in to see this content.
You have no subscription access to this content.
No metrics data to plot.
The attempt to load metrics for this article has failed.
The attempt to plot a graph for these metrics has failed.
Carotid artery recognition system: A comparison of three automated paradigms for ultrasound images
Rent this article for


Image of FIG. 1.
FIG. 1.

Automated cropping of the ultrasound DICOM image. The original image frame (left) is cropped (right). The white circle indicates the scanning depth, which could be used for manual computation of the calibration factor.

Image of FIG. 2.
FIG. 2.

CARSgd procedure for ADF tracing. (a) Original cropped image. (b) Down-sampled image. (c) Despeckled image. (d) Image after convolution with first-order Gaussian derivative (sigma = 8). (e) Intensity profile of the column indicated by the vertical dashed line in panel d (ADF indicates the position of the far adventitia wall.). (f) Cropped image with far adventitia profile overlaid.

Image of FIG. 3.
FIG. 3.

CARSia (integrated approach) recognition strategy. (a) Original image. (b) Automatically identified line segments (black lines). (c) Final ADF tracing after line segments validation and combination.

Image of FIG. 4.
FIG. 4.

CARSsa (signal processing) strategy for far wall adventitial tracing. (a) Original image. (b) 2DH. The gray portion of the 2DH denotes the region in which we suppose to find only lumen pixels. (c) Original image with lumen pixels overlaid in gray. (d) Sample processing of one column, with the marker points of the far (ADF) adventitia layer and of the lumen (L).

Image of FIG. 5.
FIG. 5.

The left column reports a sample of automated carotid recognition by CARSgd (a), CARSia (c), and CARSsa (e) compared with the human traced ADF profile. In the right column, the CARSgd (b), CARSia (d), and CARSsa (f) ADF profile is compared with the ground truth LI/MA boundaries.

Image of FIG. 6.
FIG. 6.

Samples of automated far adventitia tracing by the three techniques. CARSgd is represented by panels a, d, and g. CARSia is shown in panels b, e, and h. CARSsa is represented in panels c, f, and i. The white dashed line represents the GTLI profile, the black dashed line represents the GTMA. The continuous black line represents the ADF. The first row is relative to a straight and horizontal artery; the second row to a curved artery; the third row to a straight but inclined artery.

Image of FIG. 7.
FIG. 7.

Effect of ADF distance from MA on the IMT measurement error. The ADF profile was progressively shifted toward the bottom of the image (i.e., the HD from MA increased). (a) Average IMT measurement error for the three techniques. (b) Standard deviation of the IMT measurement error for the three techniques. The tolerance zone, where the error remains stable, is between 0 and 6 pixels of shift. Around 6–8 pixels, there is a break point and the error increases (CARSia—black dashed line; CARSsa—gray line; CARSgd —black line.)

Image of FIG. 8.
FIG. 8.

Distribution of the HD between ADF profiles and the LI (left panels) and MA (right panels) interfaces. The top row is relative to CARSia, the middle to CARSsa, the bottom to CARSgd.

Image of FIG. 9.
FIG. 9.

Samples of inaccurate ADF tracings by CARSgd (a); CARSia (b); and CARSsa (c). Dashed lines are the average GTLI (white) and GTMA (black). The continuous black line represents the ADF. The white arrows indicate the inaccurate adventitia tracing.


Generic image for table

Image database and patient demographics—Characteristics of the image dataset coming from four different institutions and relative patient demographics. The first column reports the institution, the second the number of images, the third the conversion factor, the fourth the scanner used. Finally, the last two columns report the number of patients and their demographics.

Generic image for table

Hausdorff distance between the average GT ADF (first row), LI (second row), and MA (third row) boundaries and the ADF automated tracings by the three considered techniques.

Generic image for table

Performance of three techniques using visual inspection of CCA recognition


Article metrics loading...


Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Carotid artery recognition system: A comparison of three automated paradigms for ultrasound images