Three-dimensional automatic quantitative analysis of intravascular ultrasound images

doi:10.1016/S0301-5629(99)00167-2

Ultrasound in Medicine & Biology

Volume 26, Issue 4, May 2000, Pages 527-537

https://doi.org/10.1016/S0301-5629(99)00167-2 Get rights and content

Abstract

Intravascular ultrasound (IVUS) has established itself as a useful tool for coronary assessment. The vast amount of data obtained by a single IVUS study renders manual analysis impractical for clinical use. A computerized method is needed to accelerate the process and eliminate user-dependency. In this study, a new algorithm is used to identify the lumen border and the media-adventitia border (the external elastic membrane). Setting an initial surface on the IVUS catheter perimeter and using active contour principles, the surface inflates until virtual force equilibrium defined by the surface geometry and image features is reached. The method extracts these features in three dimensions (3-D). Eight IVUS procedures were performed using an automatic pullback device. Using the ECG signal for synchronization, sets of images covering the entire studied region and corresponding to the same cardiac phase were sampled. Lumen and media-adventitia border contours were traced manually and compared to the automatic results obtained by the suggested method. Linear regression results for vessel area enclosed by the lumen and media-adventitia border indicate high correlation between manual vs. automatic tracings (y = 1.07 × −0.38; r = 0.98; SD = 0.112 mm²; n = 88). These results indicate that the suggested algorithm may potentially provide a clinical tool for accurate lumen and plaque assessment.

Introduction

Coronary angiography has been the gold standard for imaging and diagnostics in the interventional catheterization laboratory. This method depicts coronary arteries as a silhouette of the lumen. The information regarding the vessel wall and its contents is limited and sometimes misleading. In the last few years, intravascular ultrasound (IVUS) has established itself as a useful tool for coronary assessment (Schawarzacher et al. 1997). IVUS is a catheter-based technique that provides two-dimensional cross-sectional images of the arterial wall. IVUS is based on a highly miniaturized transducer element located on the catheter tip. IVUS images provide information regarding luminal cross-sectional area, wall thickness, wall structure (intima, media and adventitia layers) and compensate for the lack of knowledge provided by angiography. This makes it a valuable addition to the interventional procedure.

Quantitative analysis of IVUS images, such as area measurements, can be done manually. However, manual tracing is a very time-consuming procedure subjected to operator variability. Thus, to reduce the time of analysis and the subjectivity of manual tracing, an automatic method is needed. Several algorithms have been developed to trace automatically Mojsilovic et al 1997, Sonka et al 1994 or semi-automatically Li et al 1992, Li et al 1994 the wall boundaries in two dimensions (2-D) Li et al 1992, Mojsilovic et al 1997, Sonka et al 1994 or three dimensions (3-D) Li et al 1994, Von Birgelen et al 1996a, Von Birgelen et al 1996b, Von Birgelen et al 1997. These algorithms have used texture-based approaches Dixon et al 1997, Mojsilovic et al 1997, Hough transform (Van Horn et al. 1994), cross-correlation Li et al 1992, Sonka et al 1994 and other methods for image segmentation. Segmentation in the 3-D framework provides essential information to the clinician by allowing volumetric measurements in addition to the area measurements evaluated in 2-D. In addition, knowing the stenosis path along the vessel can contribute to the interventional procedure.

IVUS images are characterized by relatively round-shaped features arising from the natural characteristics of the vessel. Image noise in IVUS consists of speckles, shadowing (due to calcium or an implanted stent) and image artifacts (e.g., misregistered echoes or signal loss). The properties of active contour algorithms (or “snakes”) proposed by Kass et al. (1987) make them suitable for handling such problems. These algorithms take into consideration the shape of the contour as well as the image properties, thus overcoming image distortion. In this study, a new method is suggested for identifying the lumen border and the media-adventitia border (also defined as the external elastic membrane). This method allows extraction of these borders in 3-D by employing active contour principles.

Section snippets

Data acquisition

IVUS image acquisition of the 3-D data was carried out by using a 30-MHz catheter on a “CVIS” (Boston-Scientific) ultrasound imaging system. A pullback device, which was calibrated to pull the IVUS transducer in the longitudinal direction at a constant velocity of 0.5 mm/s was utilized while scanning the vessel segment. The ECG signal was recorded along with the image and the procedure was recorded on an S-VHS VCR. A VideoPort Professional^TM frame grabber manufactured by MRT micro Inc. was used

Quantitative results

Human vessels. The algorithm was applied to the sampled sets of images, yielding automatically traced sets of contours. The list of the parameter values used here is outlined in Table 1. Forty-four images taken from the data set obtained from the human coronaries were arbitrarily selected. To avoid any bias toward longer segments, about the same number of slices were taken from each studied segment regardless of its length. The lumen and media-adventitia borders were manually traced in each

Discussion

In this study, an automatic method for lumen and media-adventitia border extraction has been introduced. The method is based on active contour principles (“snakes”), and the extraction of the features of interest is performed in 3-D. High correlation (r = 0.978) and low variability (15.2 17.4% and 6.5 ± 7.6% for the lumen and the media-adventitia borders, respectively) was found between automatic and manual tracings, indicating good consistency of the suggested algorithm. The results indicate

References (16)

K.M. Coy et al.
Intravascular ultrasound imagingA current perspective
J Am Coll Cardiol
(1991)
A.J. Thrush et al.
An evaluation of the potential and limitations of three-dimensional reconstruction from intravascular ultrasound images
Ultrasound Med Biol
(1997)
C. Von Birgelen et al.
Computerized assessment of coronary lumen and atherosclerotic plaque dimensions in three-dimensional intravascular ultrasound correlated with histomorphometry
Am J Cardiol
(1996)
C. Von Birgelen et al.
Morphometric analysis in three-dimensional intracoronary ultrasoundAn in vitro and in vivo study performed with a novel system for the contour detection of lumen and plaque
Am Heart J
(1996)
Dixon KJ, Vince DG, Cothern RM, Cornhill JF. Characterization of coronary plaque in intravascular ultrasound using...
Gunn SR. Dual active contour. PhD thesis, Faculty of Engineering and Applied Science, Department of Electronics and...
M. Kass et al.
SnakesActive contour models
Int J Comp Vision
(1987)
Li W, Bosch JG, Zhong Y, Gussenhoven WJ, Rijsterborgh H, Reiber JHC, Bom N. Semiautomatic frame to frame tracking of...

There are more references available in the full text version of this article.

Cited by (73)

Ultrasound technology assisted colloidal nanocrystal synthesis and biomedical applications
2024, Ultrasonics Sonochemistry
Non-invasive and high spatiotemporal resolution mythologies for the diagnosis and treatment of disease in clinical medicine promote the development of modern medicine. Ultrasound (US) technology provides a non-invasive, real-time, and cost-effective clinical imaging modality, which plays a significant role in chemical synthesis and clinical translation, especially in in vivo imaging and cancer therapy. On the one hand, the US treatment is usually accompanied by cavitation, leading to high temperature and pressure, so-called “hot spot”, playing a significant role in sonochemical-based colloidal synthesis. Compared with the classical nucleation synthetic method, the sonochemical synthesis strategy presents high efficiency for the fabrication of colloidal nanocrystals due to its fast nucleation and growth procedure. On the other hand, the US is attractive for in vivo and medical treatment, with applications increasing with the development of novel contrast agents, such as the micro and nano bubbles, which are widely used in neuromodulation, with which the US can breach the blood–brain barrier temporarily and safely, opening a new door to neuromodulation and therapy. In terms of cancer treatment, sonodynamic therapy and US-assisted synergetic therapy show great effects against cancer and sonodynamic immunotherapy present unparalleled potentiality compared with other synergetic therapies. Further development of ultrasound technology can revolutionize both chemical synthesis and clinical translation by improving efficiency, precision, and accessibility while reducing environmental impact and enhancing patient care. In this paper, we review the US-assisted sonochemical synthesis and biological applications, to promote the next generation US technology-assisted applications.
Convolutional networks for the segmentation of intravascular ultrasound images: Evaluation on a multicenter dataset
2022, Computer Methods and Programs in Biomedicine
The delineation of the lumen contour and external elastic lamina (EEL) in intravascular ultrasound (IVUS) images is crucial for the quantitative analysis of coronary atherosclerotic plaques. However, the presence of ultrasonic shadows and anatomical structures (such as bifurcations and calcified plaque) complicates the automatic delineation of the lumen contour and EEL. The purpose of this paper is to evaluate the IVUS segmentation performances of different convolutional networks and the impact factors on a large-scale multiple-center dataset.
A total of 6516 cross-sectional images from 175 IVUS pullbacks acquired in different centers by different IVUS imaging catheters were screened from a corelab to evaluate the segmentation methods. The IVUS images included bifurcation, side branch ostia, and various image artifacts to reflect the general image characteristics in routine clinical acquisition. We compared three generic fully convolutional networks (FCNs) and two FCNs specifically designed for the segmentation of IVUS images and explored the factors impacting the segmentation performance, including the training images and the input of consecutive images to the models. The performance of the FCNs was evaluated by using the Dice similarity coefficient (DSC), the Jaccard index (JI), the Hausdorff distance (HD), linear regression and Bland-Altman analysis.
The 4-cascaded RefineNet and DeepLabv3+ outperformed U-net and IVUS-net in the segmentation of the lumen contour and EEL on IVUS images. DeepLabv3+ had the best segmentation performance, with DSCs of 0.927 and 0.944, JIs of 0.911 and 0.933, and HDs of 0.336 mm and 0.367 mm for delineation of the lumen and EEL, respectively. Excellent agreement between DeepLabv3+ and the manual delineation was found in the quantification of the coronary plaque area (r = 0.98).
The convolutional network architecture is effective in the automatic segmentation of IVUS images. It might contribute to the clinical application of quantitative IVUS analysis in real-world as well as the efficient assessment of coronary atherosclerosis.
Advanced deep learning methodology for accurate, real-time segmentation of high-resolution intravascular ultrasound images
2021, International Journal of Cardiology
The aim of this study is to develop and validate a deep learning (DL) methodology capable of automated and accurate segmentation of intravascular ultrasound (IVUS) image sequences in real-time.
IVUS segmentation was performed by two experts who manually annotated the external elastic membrane (EEM) and lumen borders in the end-diastolic frames of 197 IVUS sequences portraying the native coronary arteries of 65 patients. The IVUS sequences of 177 randomly-selected vessels were used to train and optimise a novel DL model for the segmentation of IVUS images. Validation of the developed methodology was performed in 20 vessels using the estimations of two expert analysts as the reference standard.
The mean difference for the EEM, lumen and plaque area between the DL-methodology and the analysts was ≤0.23mm² (standard deviation ≤0.85mm²), while the Hausdorff and mean distance differences for the EEM and lumen borders was ≤0.19 mm (standard deviation≤0.17 mm). The agreement between DL and experts was similar to experts' agreement (Williams Index ranges: 0.754–1.061) with similar results in frames portraying calcific plaques or side branches.
The developed DL-methodology appears accurate and capable of segmenting high-resolution real-world IVUS datasets. These features are expected to facilitate its broad adoption and enhance the applications of IVUS in clinical practice and research.
Vessel membrane segmentation and calcification location in intravascular ultrasound images using a region detector and an effective selection strategy
2020, Computer Methods and Programs in Biomedicine
Segmenting vessel membranes and locating the calcific region in intravascular ultrasound (IVUS) images aid physicians in the diagnosis of atherosclerosis. However, the manual extraction of the media adventitia (MA)/lumen border and calcification location are cumbersome due to the excessive number of IVUS frames. Moreover, most existing (semi-)automatic detection methods cannot achieve both vessel membrane extraction and calcification location simultaneously, and they are unable to detect vessel membranes in IVUS frames from different acquisition systems.
A fully automatic approach is proposed based on extremal regions and a flexible selection strategy to extract vessel membranes in different IVUS frames and locate the calcific region in high-frequency ones. Three main steps are included in the algorithm. First, a region detector is employed to extract extremal regions from an IVUS image. Then, according to the selection strategy, a part of the extracted regions is selected. At the same time, the calcification is located according to its special acoustic properties. Next, approximate MA and lumen border segmentation is achieved based on the selected extremal regions and the located calcification in polar coordinates. Finally, the final segmentation results are obtained by smoothing the approximate values.
To demonstrate the feasibility of the method, it was evaluated based on a standard public dataset. Furthermore, to quantitatively evaluate the segmentation performance, the Hausdorff distance (HD), Jaccard measure (JM) and percentage of area difference (PAD) were used. The results show that a mean HD of 1.13/1.21 mm, a mean JM of 0.83/0.77 and a mean PAD of 0.11/0.23 are achieved for MA/lumen border detection in 77 40-MHz IVUS images. For MA/lumen border extraction in 435 20-MHz IVUS frames, the average HD, JM and PAD values are 0.47/0.28 mm, 0.84/0.89 and 0.13/0.10, respectively. In addition, the approach successfully achieves calcification location in 40-MHz IVUS frames. In comparison with other published methods, the method proposed in this study is competitive.
According to these results, our strategy can extract MA/lumen borders in different IVUS frames and effectively locate calcification in high-frequency IVUS frames.
Automatic IVUS lumen segmentation using a 3D adaptive helix model
2019, Computers in Biology and Medicine
Citation Excerpt :
Using the framework of IVUS segmentation based on active contours or snakes, several methods using bi-dimensional parametric, geometric, geodesic, and region-based active contours (fast-marching method) have been developed [10–13]. An extension to 3D active contour methods based on local properties of the image gradient and image intensity have also been developed to successfully extract contours in IVUS sequences [14–16]. Likewise, our group has developed a level set approach to detect IVUS-relevant regions based on a mixture of Rayleigh probability distribution [17].
In this paper, we develop a three dimensional (3D) segmentation algorithm of the lumen visualized using intravascular ultrasound (IVUS) imaging. These images are known for their various granular textures (speckles) that make the discrimination of different tissues very difficult, especially as a result of the presence of artifacts and shadows generated by tissue calcification. Our model consists of a helical active contour initialized automatically over the sequence, that evolves based on the analysis of the Rayleigh distribution of gray levels in order to extract the luminal border. This novel algorithm is fast, uses an adaptive simple space curve for 3D extraction of the lumen, and is fully automatic. Consequently, it does not require an initialization close to the lumen border. Segmentation was carried out on 19 IVUS sequences with a total of 8918 images acquired in vivo on nine femoral and ten coronary arteries using a 20 MHz probe. These sequences showed many difficulties, such as severe stenosis, bifurcations, side vessels, shadows, and other artifacts. The quantitative evaluation of our algorithm compared to the ground truth for the femoral and coronary datasets showed an overlap greater than 89% for the Jaccard index and greater than 94% for the Dice index, yielding an accuracy of more than 98.5%. Several other metrics are also presented that confirm the efficiency of our helix model compared to other recent methods reported in the literature using a similar ultrasound probe.
Calcium detection, its quantification, and grayscale morphology-based risk stratification using machine learning in multimodality big data coronary and carotid scans: A review
2018, Computers in Biology and Medicine
Atherosclerosis is the leading cause of cardiovascular disease (CVD) and stroke. Typically, atherosclerotic calcium is found during the mature stage of the atherosclerosis disease. It is therefore often a challenge to identify and quantify the calcium. This is due to the presence of multiple components of plaque buildup in the arterial walls. The American College of Cardiology/American Heart Association guidelines point to the importance of calcium in the coronary and carotid arteries and further recommend its quantification for the prevention of heart disease. It is therefore essential to stratify the CVD risk of the patient into low- and high-risk bins.
Calcium formation in the artery walls is multifocal in nature with sizes at the micrometer level. Thus, its detection requires high-resolution imaging. Clinical experience has shown that even though optical coherence tomography offers better resolution, intravascular ultrasound still remains an important imaging modality for coronary wall imaging. For a computer-based analysis system to be complete, it must be scientifically and clinically validated. This study presents a state-of-the-art review (condensation of 152 publications after examining 200 articles) covering the methods for calcium detection and its quantification for coronary and carotid arteries, the pros and cons of these methods, and the risk stratification strategies. The review also presents different kinds of statistical models and gold standard solutions for the evaluation of software systems useful for calcium detection and quantification. Finally, the review concludes with a possible vision for designing the next-generation system for better clinical outcomes.

View all citing articles on Scopus

¹: [email protected]–August2000,addresscorrespondenceto:HarvardMedicalSchool,DepartmentofRadiology,BethIsraelDeaconessMedicalCenter,330BrooklineAvenue,Boston,MA02215,USA

View full text

Original ContributionsThree-dimensional automatic quantitative analysis of intravascular ultrasound images

Abstract

Introduction

Section snippets

Data acquisition

Quantitative results

Discussion

J Am Coll Cardiol

Ultrasound Med Biol

Am J Cardiol

Am Heart J

SnakesActive contour models

Int J Comp Vision

Original Contributions
Three-dimensional automatic quantitative analysis of intravascular ultrasound images