Measurements of carotid intima media thickness in non-invasive high-frequency ultrasound images: the effect of dynamic range setting

Ultrasound (US) measurements of carotid intima-media thickness (CIMT) are used to assess the extent of arteriosclerosis and are a biomarker for the future risk of cardiovascular disease (CVD) and stroke [1‐6]. Rate of CIMT change is also used as a surrogate endpoint for CVD in interventional studies to assess early effects of drug therapies [7, 8].

In US images the carotid artery wall is depicted as three separate layers: intima, media and adventitia (Figure 1). The CIMT is defined as the distance from the lumen-intima interface to the media-adventitia interface [9]. The intima-media complex can easily be distinguished from the surrounding tissue in US images and the distinct borders allows for manual as well as automatic measurements of the CIMT.

Reproducibility is a prerequisite for a reliable biomarker, and standardization of imaging and measurement of CIMT will minimize variability between measurements and between readers. Development of US protocols for measuring CIMT has been focusing on choice of vessel segment, near or far wall measurements, and frequency setting as well as performance of the technicians and the observers [5]. The updated Mannheim Carotid Intima-Media Thickness and Plaque Consensus [5] suggests that the dynamic range (DR) should be set at approximately 60 decibel (dB). The DR defines the number of US echo intensities included in the image formation process. Increased DR will increase the range of signal intensities resulting in a low-contrast image. Oppositely, decreased DR will decrease the range of signal intensities resulting in a high-contrast image.

It has previously been shown that change in DR has a significant effect on measurements of wall thickness and lumen diameter in phantoms [10], however the documentation on effects in human arteries are lacking..

The purpose of this study was to quantify the effects of changing DR in non-invasive US CIMT measurements in the common carotid arteries (CCA).

A total of 313 CCAs were included in the analysis: 181 CCAs from participants in the ABD-study and 132 CCAs from participants in the INFO-study. Population mean CIMT_mean was 0.529, 0.564, 0.590 and 0.605 mm with DR of 40, 65, 70 and 85 dB, respectively. Corresponding results for CIMT_max was 0.626, 0.667, 0.698 and 0.716. Differences between the DR steps were all statistically significant (p < 0.01) (Tables 2 and 3). The relative increases in CIMT_mean between DR steps were equal to the relative increase in CIMT_max for the same steps (p > 0.44). The relative increases for CIMT_mean and CIMT_max were 6.7% and 7.0% for 40–55 dB, 4.9% and 4.8% for 55–70 dB, and 2.6% and 2.7% for 70–85 dB, respectively.

Average change in CIMT_mean from 40 to 85 dB was 0.076 mm corresponding to approximately 0.008 mm for each 5 dB increase. For all measurements mean CIMT_sd was less than 0.05 mm.

The CV for CIMT_mean was 0.65% (95% CI; 0.48-0.83) for 40 dB, 0.42% (95% CI; 0.28-0.56) for 55 dB, 0.86% (95% CI; 0.51-1.21) for 70 dB and 0.59% (95% CI; 0.37-0.82) for 85 dB. For CIMT_max the CVs were 0.59% (95% CI; 0.33-0.85), 0.85% (95% CI; 0.16-1.36), 1.36 (95%CI; 0.37-2.37) and 1.04% (95% CI; 0.41-1.67) for 40, 55, 70 and 85 dB, respectively.

This study showed that an increase in DR in US image acquisition yields increased CIMT values and that the effect of increasing DR is largest for lower DRs. The DR should therefore be specified in all US protocols where CIMT values are used either directly or as a predictive biomarker, especially in longitudinal follow-up assessments.

As DR increases the range of signal intensities included in the image formation increases. This increase yields reduced image contrast and less defined interfaces between the intima, media and adventitia layers in the artery wall. Consequently, it is anticipated that CIMT would increase with increasing DR. In the interval between 40 and 85 dB the average increase in CIMT_mean per 5 dB increase in DR was 0.008 mm. This is approximately 2.5 times higher than the values found by Potter et al. in a phantom study [10]. The phantoms were made of tissue-mimicking agar with wall thicknesses of 0.5, 1.0 and 1.5 mm and acoustic properties similar to tissue. In our study, the changes in CIMT measurements due to DR variation were anticipated to be in the same range as the phantom with 0.5 mm wall thickness. The discrepancy in findings may be related to the composition of the wall (phantom versus human arteries). In contrast to human arteries, the phantom walls were homogenous with defined interfaces and consequently, interface echo signals will be more distinct in phantoms than in human arteries. The relative change in CIMT_mean for CCAs was largest between 40 and 55 dB (6.7%) and smallest between 70 and 85 dB (2.6%) indicating a declining dependency for increasing DR. The relative increase found by Potter et al. for DR of 40 to 60 dB was about 3% and as such our findings suggest an underestimation of DR dependency in phantoms, especially for low DRs. The algorithms used for the automatic edge detection as well as larger variation in the tissue composition of human arteries than in the phantoms might have contributed to the differences between the studies. We speculate that the variation in the human tissue causes a broader spectrum of echo signals, and that the effect of widening the DR therefore will be more prominent in this setting. However, despite differences in magnitude both Potter et al. and we found that CIMT increased for increasing DR.

The CIMT_mean and CIMT_max found for different DRs in our study emphasize the need for reporting DR used in US protocols as CIMT and change in CIMT over time are used as tools for risk-stratification, as surrogate end-points, and as references for clinical decision-making: European Guidelines on cardiovascular disease prevention in clinical practise 2012 states that CIMT > 0.9 mm is considered abnormal [11]. The extreme DR values (40 and 85 dB) in our study yielded a difference in CIMT_mean and CIMT_max of 8-10% of the suggested reference value. Knowledge of DR setting is important for patients where CIMT values are close to cut-off values or reference values of clinical importance. Different DR between baseline and follow-up examinations may lead to under- or overestimation of CIMT change. Outcome of multicentre studies may also be compromised. Thus, by including information of DR in guidelines, reference data and study protocols will increase the validity of the results.

A shortcoming of our study is the large DR steps (15 dB). Smaller DR steps would probably give more precise results of the relation between measured CIMT and DR setting. However, the wide range of DRs included allowed identification of the non-linearity between change in CIMT and change in DR that would probably not have been detected if DR had been restricted to 40 to 60 dB, as in Potter et al. [10]. The large number of participants and the fact that the relative CIMT changes for different DRs were similar for the two study populations differing in age and gender mixture strengthens the external validity of our findings.

In conclusion, DR significantly changes CIMT measurements and the changes are most prominent for lower DRs. The effect of changing DR is larger in human arteries than in phantoms. Reporting the DR will therefore increase the validity of CIMT data, especially in prospective clinical trials.