MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Influence of Imaging Preset On Grayscale Texture Features of Ultrasound Carotid Plaque Images and Their Relationship to Serum Lipid Levels

C Steffel*, J Stein , C Korcarz , O Okonkwo , C Mitchell , University of Wisconsin-Madison School of Medicine and Public Health, Madison, WI

Presentations

(Tuesday, 7/31/2018) 4:30 PM - 6:00 PM

Room: Room 207

Purpose: Texture features from ultrasound brightness-mode images of carotid plaque are associated with cardiovascular disease risk factors. We evaluated whether imaging preset affected the relationships between plaque texture features and serum lipid levels.

Methods: Carotid plaque images (n=116) were acquired from 35 individuals with two different imaging presets. Preset 1 utilized high dynamic range (90 dB) and the most linear grayscale map possible. Preset 2 utilized moderate dynamic range (65 dB) and grayscale map optimized for carotid intima-media thickness measurement. Texture features of Plaque Area and Black Area (mm²), Grayscale Median (GSM) Value, Gray Level Difference Statistics (GLDS) Contrast, Plaque Type (1-4, lower type associated with pixels having grayscale values below 25), Percent Black Area in One Component of Plaque (BAOC), and Percent Black Area in All Plaque (BAAP) were extracted from segmented plaque using specialized plaque analysis software. Serum concentrations of total cholesterol, LDL-cholesterol, HDL-cholesterol, non-HDL cholesterol, and triglycerides were measured within 2 weeks of imaging. Non-parametric statistical tests were used. Wilcoxon Signed Rank test assessed differences between texture features by imaging preset. Spearman’s rho partial-correlation coefficient (age, gender) assessed relationship between texture features and serum lipid levels by imaging preset.

Results: Participants were median(interquartile range[IQR])=67.3(9.9) years-old, 49 percent female. No significant differences between presets and texture features were found apart from plaque area (Preset 1 median(IQR)=16.26(16.02), Preset 2=15.36(15.10), p=0.007). Preset 1 yielded significant correlations (correlation coefficient, p-value) between serum triglycerides and texture features: BAOC (0.352, p=0.014) and BAAP (0.373, p=0.009). Correlations with other lipid fractions and between serum lipid levels and Preset 2 were not significant. Plaque type classification changed between presets for 37 percent of plaques (Cohen’s kappa=0.259, p=0.038).

Conclusion: Associations between plaque characteristics and serum triglycerides were identified only using images acquired with Preset 1 (highest dynamic range and most linear grayscale map).

Funding Support, Disclosures, and Conflict of Interest: This investigation was supported by Alzheimer's Association NIRGD-305257, Extendicare Foundation, and F31HL141008. C. Steffel, C. Korcarz, O. Okonwko have nothing to disclose. J. Stein, Wisconsin Alumni Research Foundation, patent related to carotid wall thickness and vascular age. C. Mitchell, Davies Publishing Inc, authorship textbook; Elsevier, Wolters-Kluwer, author textbook chapters, royalties.

Keywords

Ultrasonics, Quantitative Imaging, Stroke

Taxonomy

IM- Ultrasound : Quantitative imaging/analysis

Contact Email