Author + information
- Mika Bando, MD,
- Hirotsugu Yamada, MD, PhD∗ (, )
- Kenya Kusunose, MD, PhD,
- Daiju Fukuda, MD, PhD,
- Rie Amano, RMS,
- Rina Tamai, RMS,
- Yuta Torii, RMS,
- Yukina Hirata, RMS,
- Susumu Nishio, RMS,
- Junichiro Satomi, MD, PhD,
- Shinji Nagahiro, MD, PhD and
- Masataka Sata, MD, PhD
- ↵∗Department of Cardiovascular Medicine, Tokushima University Hospital, 2-50-1 Kuramoto, Tokushima, Japan
Ultrasound imaging of carotid artery plaques not only identifies the existence and size of the plaques but also provides information on the histological features of the plaques such as lipid pools and fibrous or calcified tissue. We attempted to develop a novel system to obtain the characteristics of carotid plaques using noninvasive ultrasound and high-resolution color-coded display maps based on integrated backscatter (IB) values (the “iPlaque” system).
We prospectively enrolled 17 patients for this study. Seven patients (n = 7 men, 71 ± 5 years old) who underwent carotid artery endarterectomy (CEA) from September 14, 2010 to October 25, 2011 were the group in which we determined the IB thresholds. For validation of the imaging software in our iPlaque system, data were obtained from the following 10 patients (mean age 70 ± 7 years, n = 7 men) who underwent CEA from July 6, 2012 to May 17, 2013. This study was approved by the institutional review board of the University of Tokushima, and each patient gave written informed consent.
Offline imaging analytical software was designed to extract IB values from each pixel of the ultrasound images, which were obtained with commercially available ultrasound diagnostic equipment (Logiq 7, General Electric Medical Systems, Milwaukee, Wisconsin). Cross-sectional scanning of the carotid lesion was performed before surgery, and >20 images/cm that included the largest plaque was stored digitally as RAW data.
In a cross-section of the resected plaques from 7 patients who underwent CEA, we identified 10 sites in each plaque where the histology showed calcification, dense fibrosis, fibrosis, or lipid. Calcification was identified as purple cellular crystals by hematoxylin and eosin staining and brown crystals by Von Kossa staining. Dense fibrosis was identified as intense green staining by Masson trichrome stain or thick orange staining by microscopic polarization after picrosirius red staining. The fibrous portion of the plaque was stained less intensely green by Masson trichrome. The lipid content was visualized as cholesterol clefts on hematoxylin and eosin, Masson trichrome, and oil red O staining.
The IB values of the corresponding locations in the same cross-sectional ultrasound image were calculated. The histological specimen that corresponded to the ultrasonic image was carefully selected by matching the diameter of the vessel and the thickness and shape of the plaque. The IB value was calculated as (tissue value − 255)/255 × dynamic range. The dynamic range was fixed as 40 dB throughout this study. To compensate for the effect of ultrasound attenuation, the IB data in the image were adjusted by 2.0 dB/mm. Each IB value was normalized to the calibrated value by setting the IB value of blood near the plaque at −70 dB. The plaque in the image was manually traced, and then the program calculated the plaque area, area of each tissue component categorized by the IB threshold, and the amount of each component as a percent of the entire plaque area.
The average normalized IB value in 70 sampling points from 7 patients was −37.99 ± 4.93 dB (range −30.14 to −46.00 dB) in dense fibrosis tissue, −52.34 ± 3.99 dB (range −44.73 to −61.23 dB) in fibrosis tissue, and −69.55 ± 5.16 dB (range −60.08 to −79.33 dB) in lipid pool. With a cutoff value of −46.18 dB, dense fibrosis and fibrosis could be discriminated with 97.1% sensitivity and 100% specificity. With a cutoff value of −61.23 dB, fibrosis and lipid pool could be identified with 97.1% sensitivity and 100% specificity. Because the maximum IB value of dense fibrosis was −30.14 dB, a tissue with IB > −30.14 dB was considered to be calcification. Thus, we assigned the following IB thresholds for tissue characterization: calcification −30.14 dB ≤ IB; dense fibrosis −46.18 dB ≤ IB < −30.14 dB; fibrosis −61.23 dB ≤ IB < −46.18 dB; and lipid pool IB < −61.23 dB. With our iPlaque software, calcification is displayed in red, dense fibrosis in yellow, fibrosis in green, and lipid pool in blue. Figure 1 shows the color-coded maps and various histological stains of resected plaques.
Using the next 10 patients who underwent CEA, we validated our iPlaque imaging analysis by comparing it with the histology of the carotid plaque specimens. The overall agreements for tissue characterization between iPlaque and histological analyses in each tissue type were good (calcification r = 0.98; dense fibrosis r = 0.82; fibrosis r = 0.62; lipid r = 0.61; all p < 0.001).
We developed and validated a novel imaging system, iPlaque, which produces an automatic color map of the tissue characteristics of carotid plaques with much higher spatial resolution (0.1 vs. 2.0 mm) compared with a previous similar system (1). Our iPlaque system allows noninvasive quantitative tissue characterization of carotid plaques and may become a useful tool for the management of patients with atherosclerosis.
Please note: This study was supported in part by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan (Knowledge Cluster and New Research Area) and JSPS KAKENHI grants 25293184, 25670390, 24659392, and 22500437.
- American College of Cardiology Foundation
- Kawasaki M.,
- Takatsu H.,
- Noda T.,
- et al.