Effects of ultrasound technology advances on measurement of carotid intima–media thickness: A review

Abstract

In this review, we describe how technological advances in ultrasound imaging related to transducer construction and image processing fundamentally alter generation of ultrasound images to produce better quality images with higher resolution. However, carotid intima–media thickness (IMT) measurements made from images acquired on modern ultrasound systems are not comparable to historical population nomograms that were used to determine wall thickness thresholds that inform atherosclerotic cardiovascular disease risk. Because it is nearly impossible to replicate instrumentation settings that were used to create the reference carotid IMT nomograms and to place an individual’s carotid IMT value in or above a clinically relevant percentile, carotid IMT measurements have a very limited role in clinical medicine, but remain a useful research tool when instrumentation, presets, image acquisition, and measurements can be standardized. In addition to new validation studies, it would be useful for the ultrasound imaging community to reach a consensus regarding technical aspects of ultrasound imaging acquisition, processing, and display for blood vessels so standard presets and imaging approaches could reliably yield the same measurements.

Introduction

Measuring the combined thickness of the intimal and media layers of the carotid artery (carotid intima–media thickness; IMT) using B-mode ultrasound is an established research technique for assessing arterial injury, atherosclerotic cardiovascular disease (ASCVD) risk, and the effects of interventions that reduce ASCVD risk for over three decades.^1–15 Carotid IMT measurement, especially in conjunction with plaque detection, also has been used clinically to refine ASCVD risk estimates for over two decades, though currently coronary artery calcium is the dominant technology used for this purpose in the United States.^1,3,16–19 Guideline and consensus recommendations that adapted carotid IMT research techniques for the clinical setting relied upon the clinician’s ability to map a common carotid artery IMT value from an individual patient onto a reference population’s IMT distribution with increased ASCVD risk indicated by values ⩾ 75th percentile based on a patient’s age, sex, and race.^1,18,20 Carotid IMT reference nomograms differed by study populations, imaging protocols, and ultrasound instrumentation (e.g. transducers and image processing techniques). ¹ Although efforts to match a patient to a reference nomogram and to match imaging techniques were identified, transducer technology, image acquisition, and image processing techniques have changed dramatically in the past two decades. Because of significant improvements in image frame rates and resolution, it is nearly impossible to replicate the instrumentation settings that were used to create the reference carotid IMT nomograms and to confidently place an individual’s carotid IMT value in or above a clinically relevant percentile.^21,22 In this review, we describe how technological advances in ultrasound imaging related to transducer construction and image processing fundamentally alter generation of ultrasound images to produce better quality images with higher resolution. The aim of this review is to provide background and a framework for future studies that examine how ultrasound systems and instrumentation may influence carotid IMT measurements and to encourage more validation and standardization of modern ultrasound image acquisition and processing.

Advances in transducer technology and image processing techniques

Advances in transducer technologies

Since the 1980s and 1990s when most of the epidemiological carotid IMT reference databases were created, ¹ ultrasound transducers have undergone several technological improvements, including development of broad bandwidth multi-frequency transducers and changes in transducer element construction. ²³ Modern ultrasound transducers include composite elements that improve bandwidth, resolution, and sensitivity, and lower impedance, ²³ as well as more uniform ‘PureWave’ crystals used by one manufacturer. ²⁴ Some ultrasound systems also use capacitive micromachined ultrasonic transducers (CMUT) and/or transducers constructed using silicon chip technology. Compared to traditional lead zirconate titanate (PZT) and composites, CMUT transducers allow for broader bandwidth and improved axial and lateral image resolution.^25,26 Additional advances permit greater use of multi-row and two-dimensional arrays instead of single row arrays. These advances in transducer construction led to improved image resolution without compromising penetration compared to imaging performed with transducers used in earlier eras.

Advances in ultrasound image processing

In the 1980s and 1990s, ultrasonographers primarily focused on adjusting imaging frequency and focal zone settings to improve detail resolution (i.e. the ultrasound system’s ability to detect two separate structures and display them as two separate echoes) of arterial images. ²¹ Modern ultrasound systems use technology that can improve contrast resolution (i.e. an ultrasound system’s ability to discern differences between small anatomic structures that differ in their scattering strength) or echogenicity ²⁷ based on differences in image brightness.^21,28 Recent advances include spatial compounding, tissue harmonic imaging, and real-time adaptive filtering (speckle reduction) techniques. ²¹ Spatial compounding uses electronic beam steering to acquire several overlapping images, thereby improving image quality and contrast resolution by reducing speckle and clutter.^21,23,28 Surfaces of biological structures are presented with increased clarity and improved contrast because they are imaged from multiple angles.^22,23 Tissue harmonic imaging filters the fundamental transmitted frequency echoes and only accepts second harmonic frequency echoes, which improves detail resolution because the harmonic frequency is a multiple of the original frequency and the beam is narrower.^{22,23,29–31} Harmonic imaging also improves image quality by reducing grating lobe artifacts. ²³ Real-time adaptive filters apply mathematical algorithms to improve edge detection, smooth tissue texture, and reduce noise and artifacts while preserving temporal resolution. ²¹ Real-time adaptive filtering occurs within a single frame and across successive frames by scanning each pixel in an image and adjusting pixel display characteristics for signal versus artifact. ²¹

Use of these advanced technologies can improve image quality for superficial structures, such as the carotid artery, and enhance visualization of the carotid IMT. ²² They significantly alter ultrasound image quality compared to traditional systems without these advanced features, including most of the scanners used in historical epidemiological studies. These advanced image processing techniques vary across systems, making it difficult to compare the resolution and resulting carotid IMT values from modern ultrasound systems to images acquired with older technologies.

Advances in computing speed and power of ultrasound systems

Prior to 2010, most ultrasound systems consisted of a beam-former, signal processor, image processor, and display. These systems operated by having the transducer send an ultrasound pulse along a focused beam into the body.^23,32–34 The pulse interacted with tissues and a stream of echoes was sent back to the transducer.^23,32–34 Based on the direction of the pulse and the time for the echoes to return, the ultrasound system processed the returning echo signals from the active elements to produce a line of data (scan line) and, with each pulse, one or more scan lines were written. This process was repeated many times to create an anatomic image.^23,32–34 Increasing the number of scan lines used to create an image improves spatial detail (lateral) resolution ³⁵ and image quality. Thus, higher line density (more scan lines) provides better detail resolution, but it takes longer to create the image and it lowers the frame rate, which decreases temporal resolution.^23,35

As image processing techniques advanced, some scanners used enhanced computing power and parallel processing to send broad, overlapping beams into the body, collect and store individual channel data following each transmitted beam, and process these data to create a virtual transmitted beam. Different approaches use divergent waves, plane waves, and convergent waves, ³⁶ but all use enhanced parallel processing to create a synthetic transmit focus and extend the use of dynamic reception to focus on each pixel,^23,36 such that there no longer is a one-to-one correspondence between the transmitted pulse and scan lines written.^23,32–34 Instead, all channel data are collected and, through computational processing, the location and strength of each echo signal are determined and an image constructed from the stored data. Virtual transmitted beams are effectively narrower than most physical beams produced with older ultrasound systems. ²³ The end result of these enhancements is an image with improved detail resolution due to a thinner virtual beam, improved contrast resolution, and improved temporal resolution, findings that further diminish comparability between ultrasound images generated on modern systems compared to those used in historical carotid IMT studies.^23,32–34

Influence of ultrasound system and instrumentation settings on measurements

We searched the PubMed and CINAHL databases to identify articles that reported changes in measurements with changes in ultrasound system instrumentation settings and discovered that the effects of instrumentation on carotid IMT measurements have not been investigated rigorously. A small human study showed minor effects of linear ultrasound transducer frequencies (8 MHz vs 14 MHz) on carotid IMT and a study in ultrasound phantoms showed significant effects of small changes in dynamic range on wall thickness estimates.^37,38 Another study demonstrated that spatial compound imaging, tissue harmonic imaging, and real-time adaptive image processing (speckle reduction) demonstrated improved resolution of carotid wall and plaque borders. ²² A small study of individuals with human immunodeficiency virus infection examined how fundamental imaging, harmonic imaging, and crossbeam imaging affected carotid IMT measurements using the same ultrasound system. ³⁹ The authors concluded that there was not a statistically significant difference between measurements of images acquired using fundamental imaging and harmonic imaging or between crossbeam imaging and fundamental imaging. ³⁹ However, carotid IMT measurements from images that were acquired using harmonic imaging were statistically different from those using crossbeam imaging and carotid IMT measurements using crossbeam imaging did not demonstrate the same association with age that was observed with fundamental imaging. ³⁹ In that study, crossbeam imaging yielded the lowest carotid IMT values and the authors concluded that advanced imaging instrumentation settings yielded different carotid IMT measurements. ³⁹ No study that we are aware of has examined how these settings affect carotid IMT measurements in a clinical setting and the extent to which they would result in different risk assessments.

Other ultrasound disciplines also have shown that differences in ultrasound system and instrumentation settings influence measurements. For example, significant differences in fetal nuchal translucency thickness were observed when measurements obtained by fundamental imaging were compared to harmonic imaging. ⁴⁰ In that study, the same ultrasound system and transducer were used for all image acquisitions. ⁴⁰ In a study using ultrasound phantoms, significant differences in grayscale median values were observed between ultrasound systems and after changes in overall gain. ⁴¹ The results of these small studies suggest that advanced ultrasound technologies and instrumentation settings influence how an ultrasound image is displayed and significantly affect measurement values. To illustrate these findings with examples, we acquired representative images in our laboratory and identified a marked influence of ultrasound system settings on carotid IMT measurements (see online Appendix A). The influence was large and in the range of 25–50 percentile points based on nomograms from large epidemiological studies that used older ultrasound systems.

Implications for clinical and research use of carotid IMT measurements

Changes in ultrasound image processing and transducer technology have drastically altered image generation and display^23,32–34 so modern carotid IMT measurements no longer can be reliably compared to historical population nomograms that have been used to determine thresholds for normal or increased ASCVD risk. Effects on carotid artery plaque detection have not been studied, but improved resolution may increase plaque detection ²¹ and consequently affect prevalence and/or the predictive value of plaque detection. Although carotid IMT data generated with these technologies no longer can be referenced against older nomograms in the clinical setting, modern devices remain important and useful in research settings because they definitively detect and quantify early stages of arterial injury and atherosclerosis, and their responses to changes in risk factors and medications. Moreover, they are inexpensive and do not expose participants to ionizing radiation, intravenous medications, or iodinated contrast. ⁴² They especially are useful for research on determinants of arterial injury and atherosclerosis in children and young adults. ⁴³

Future directions

Changes in ultrasound technology in the past two decades provide the opportunity to develop new carotid IMT and plaque nomograms using modern equipment and contemporary patient populations. Because ultrasound is inexpensive, portable, and does not use ionizing radiation, it could be a useful technology for ASCVD risk screening, especially in children and young adults in whom coronary artery calcium often is absent and avoiding radiation is desirable. Until new reference nomograms are generated, use of arbitrary, higher carotid IMT thresholds that unequivocally indicate higher risk even if over-estimated (perhaps over the 90th or 95th percentile), could still be considered for clinical risk stratification, especially if combined with plaque screening. In addition, there is an important opportunity to continue to investigate the influence of ultrasound system and instrumentation settings on carotid IMT and other arterial measurements. Previous validation experiments that compared ultrasound systems and presets could be repeated to provide additional insight into which settings and ultrasound systems yield the most accurate measurements;^8,15 gold standards could include histopathology, a standardized ultrasound phantom, or, in vivo measurements using magnetic resonance imaging or computed tomography. Ideally, the ultrasound imaging community would use these data to reach a consensus regarding technical aspects of ultrasound image acquisition, processing, and display for blood vessels so that standard presets and imaging approaches could reliably yield the same measurements. A standardized test phantom could be used by all manufacturers and border detection software programs to validate their carotid IMT measurements.

Conclusions

Advances in transducer technology, ultrasound image processing, and ultrasound system computing speed and power produce arterial images with much higher resolution than in the past. Although these advances improve image quality and appearance, carotid ultrasound images generated on modern ultrasound systems no longer are comparable to those from historical studies that generated normal ranges for carotid IMT measurement. It is nearly impossible to replicate the instrumentation settings that were used to create the reference carotid IMT nomograms and to place an individual’s carotid IMT value in or above a clinically relevant percentile. Until new population-based nomograms with modern equipment and contemporary participant cohorts are created, carotid IMT measurements will remain useful research tools but have a very limited role in clinical medicine for assessing ASCVD risk, since absolute wall thickness measurements and their changes or differences may reflect the ultrasound system and instrumentation rather than true biological phenomena. More research and validation studies of newer ultrasound systems are needed and could inform a consensus regarding technical aspects of ultrasound image acquisition, processing, and display of arterial images.

Supplemental Material

Supplementary_Material_Appendix_A – Supplemental material for Effects of ultrasound technology advances on measurement of carotid intima–media thickness: A review

Supplemental material, Supplementary_Material_Appendix_A for Effects of ultrasound technology advances on measurement of carotid intima–media thickness: A review by Carol Mitchell, Claudia E Korcarz, James A Zagzebski and James H Stein in Vascular Medicine