Computerized assessment of coronary lumen and atherosclerotic plaque dimensions in three-dimensional intravascular ultrasound correlated with histomorphometry

Abstract

Intravascular ultrasound (IVUS), which depicts both lumen and plaque, offers the potential to improve on the limitations of angiography for the assessment of the natural history of atherosclerosis and progression or regression of the disease. To facilitate measuremenk and increase the reproducibility of quantitative IVUS analyses, a computerized contour detection system was developed that detects both the luminal and external vessel boundaries in 3dimensional sets of IVUS images. To validate this system, atherosclerotic human coronary segments (n = 13) with an area obstruction ~~40% (40% to 61%) were studied in vitro by IVUS. The computerized IVUS measurements (areas and volumes) of the lumen, total vessel, plaque-media complex, and percent obstruction were compared with findings by manual tracing of the NUS images and of the corresponding histologic cross sections obtained at 2-mm increments (n = 100). Both area and volume measurements by the contour detection system agreed well with the results obtained by manual tracing, showing low mean betweenmethod differences (-3.7% to 0.3%) with SDS not exceeding 6% and high correlation coeffrcienk (r = 0.97 to 0.99). Measurements of the lumen, total vessel, plaque-media complex, and percent obstruction by the contour detection system correlated well with histomorphometry of areas (r = 0.94,0.88,0.80, and 0.88) and volumes (r = 0.98,0.91,0.83, and 0.91). Systematic dii ferences between the resulk by the contour detection system and histomotphometry (29%, 13%, -%, and -22%, respectively) were found, most likely resulting born shrinkage during tissue fixation. The resulk of this study indicate that this computerized tVUS analysis sys tern is reliable for the assessment of coronary atherosclerosis in vivo. 0 7 996 by Excerpta Medica, Inc. (Am J Cardiol 1996;78: 1202-1209 Th e natural history and progression or regression of coronary atherosclerosis after pharmacologic and nonpharmacologic interventions have most often been assessed by quantitative coronary angiographyk3 However, the quantitative angiographic approach permits only the assessment of the luminal silhouette4 and indirect estimation of plaque burden. As a result of vessel remodeling, early atherosclerosis remains angiographically undetected until luminal encroachment starts and plaque occupies approximately 40% of the internal elastic membrane area.5 Intravascular ultrasound (IVUS) depicts both coronary lumen and vessel wall; measurements can be obtained by manually tracing the luminal and ex- ternal vascular boundaries.6-8 To reduce the time and subjectivity of manual tracing,' automated systems for quantitative analysis in 3-dimensional IVUS image sets have been developed.'0-14 As the available systems detect only the lumen, we developed a contour detection algorithm that detects both the luminal and external vascular boundaries of atherosclerotic coronar sets.'591 2 arteries in 3-dimensional IVUS image This approach allows the quantification of all IVUS images and permits even volumetric assessment which ha...

Quantification of Coronary Artery Lumen Volume by Digital Angiography

“…The bifurcations were chosen as landmarks for exact correspondence between casts and coronary arteriograms. Lumen volume was calculated from the weight and the known density (0.99 g/cm 3 ) of the casts.…”

“…Intravascular ultrasound (IVUS) and angioscopic techniques have been used for lumen volume reconstruction. [1][2][3] An inherent limitation of IVUS, however, is the wedging of the ultrasonic catheter in the case of severe stenosis and in small vessels. 4 This restricts the application of 3D IVUS to large vessels without severe stenosis.…”

A State-of-the-Art Review on Segmentation Algorithms in Intravascular Ultrasound (IVUS) Images

“…Despite the simplicity of some of the presented works [47]- [49], [53] encouraging results were produced, especially on images acquired with low-frequency transducers (20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30). However, in reality, the border detection problem is more complex, especially when high-frequency transducers (above 40 MHz) are used due to intrinsic and extrinsic image artifacts (see Section I-G) as well as local and global variations among luminal and intimal grayscale distributions, demanding for more sophisticated methods.…”

“…They studied inter-and intraobserver variability and showed high consistency of the method. They later evaluated the performance of the segmentation by comparing to histological images [20]. The algorithm performance was further refined and correlation between manual and automated traced contours was improved from (r = 0.91) to (r = 0.98) by employing electrocardiogram (ECG)-gated images [21].…”

Near real-time plaque segmentation of IVUS

“…Previous works in segmentation of IVUS images have shown different ways to segment lumen and to classify tissues [3], [4], [5]. However, these approaches usually are semi-automatic, sensitive to image artifacts and quite timeconsuming.…”