ECG-Gated Three-dimensional Intravascular Ultrasound
Feasibility and Reproducibility of the Automated Analysis of Coronary Lumen and Atherosclerotic Plaque Dimensions in Humans
Background Automated systems for the quantitative analysis of three-dimensional (3D) sets of intravascular ultrasound (IVUS) images have been developed to reduce the time required to perform volumetric analyses; however, 3D image reconstruction by these nongated systems is frequently hampered by cyclic artifacts.
Methods and Results We used an ECG-gated 3D IVUS image acquisition workstation and a dedicated pullback device in atherosclerotic coronary segments of 30 patients to evaluate (1) the feasibility of this approach of image acquisition, (2) the reproducibility of an automated contour detection algorithm in measuring lumen, external elastic membrane, and plaque+media cross-sectional areas (CSAs) and volumes and the cross-sectional and volumetric plaque+media burden, and (3) the agreement between the automated area measurements and the results of manual tracing. The gated image acquisition took 3.9±1.5 minutes. The length of the segments analyzed was 9.6 to 40.0 mm, with 2.3±1.5 side branches per segment. The minimum lumen CSA measured 6.4±1.7 mm2, and the maximum and average CSA plaque+media burden measured 60.5±10.2% and 46.5±9.9%, respectively. The automated contour-detection required 34.3±7.3 minutes per segment. The differences between these measurements and manual tracing did not exceed 1.6% (SD<6.8%). Intraobserver and interobserver differences in area measurements (n=3421; r=.97 to.99) were <1.6% (SD<7.2%); intraobserver and interobserver differences in volumetric measurements (n=30; r=.99) were <0.4% (SD<3.2%).
Conclusions ECG-gated acquisition of 3D IVUS image sets is feasible and permits the application of automated contour detection to provide reproducible measurements of the lumen and atherosclerotic plaque CSA and volume in a relatively short analysis time.
Intravascular ultrasound allows transmural, tomographic imaging of coronary arteries in humans in vivo and provides insights into the pathology of coronary artery disease by defining vessel wall geometry and the major components of the atherosclerotic plaque.1 2 3 4 5 6 7 Although invasive, IVUS is safe8 9 and allows a more comprehensive assessment of the atherosclerotic plaque than the “luminal silhouette” furnished by coronary angiography.10 11 12 13 14 Nevertheless, conventional IVUS analysis is a planar technique. Volumetric analysis of conventionally obtained IVUS images using Simpson’s rule and planar analysis of multiple image slices is possible and may yield additional information, although it is time-consuming. To reduce the time for volumetric analysis15 of IVUS images, automated 3D image reconstruction systems have been developed.16 17 18 19 20 21 22 23 24 25 26 27 However, these systems have limitations, including (1) an inconsistent ability to detect the external arterial boundary and (2) imaging artifacts produced by cyclic changes in vascular dimensions and by movement of the IVUS catheter relative to the vessel.20 22 24
As a consequence, we have developed an analysis system that (1) uses 3D IVUS image sets acquired with an ECG-gated image acquisition workstation and pullback device to limit cyclic artifacts28 and (2) detects both the luminal and external vascular boundaries of atherosclerotic coronary arteries to permit plaque volume measurement.10 29 30 31 We report the feasibility of IVUS image acquisition and the reproducibility of analysis with this methodology.
Between August 1, 1995, and February 29, 1996, we examined 28 patients with ECG-gated 3D IVUS, which represented a consecutive series of patients investigated with this approach. There were 23 men and 5 women who ranged in age from 38 to 72 years (mean, 55.3±8.9 years). All but 3 of them, studied at routine follow-up after previous catheter-based interventions, were symptomatic and/or had revealed signs of myocardial ischemia during noninvasive functional testing. Reasons for cardiac catheterization were either for diagnostic evaluation (n=20) or for follow-up study after a previous angioplasty procedure (n=8). Of the 20 patients examined during diagnostic catheterizations, 6 had one-vessel, 8 had two-vessel, and 1 had three-vessel disease. All patients with one- and two-vessel disease subsequently underwent successful catheter-based interventions (balloon angioplasty, n=3; directional atherectomy, n=2; stenting, n=9). Bypass surgery was performed in the patient with three-vessel disease. Of the 8 patients investigated at follow-up after previous interventions (after balloon angioplasty, n=5; directional atherectomy, n=3), 3 patients showed a significant restenosis and were successfully treated by repeat balloon angioplasty.
Thirty atherosclerotic coronary segments located in the left anterior descending coronary artery (n=15), right coronary artery (n=12), and left circumflex coronary artery (n=3) were analyzed; 13 segments were proximal, 15 mid, and 2 distal. As a condition for inclusion, segments had to be angiographically relatively straight (in at least two angiographic views from opposite projections). An exclusion criterion was calcification encompassing >180° of the arterial circumference over a ≥5-mm-long axial segment. This study was approved by the Local Council on Human Research. All patients signed a written informed consent form approved by the Medical Ethical Committee of the University Hospital Rotterdam-Dijkzigt.
All patients received 250 mg aspirin and 10 000 U heparin IV. If the duration of the entire catheterization procedure exceeded 1 hour, the activated clotting time was measured, and intravenous heparin was administered to maintain an activated clotting time of >300 seconds. After intracoronary injection of 0.2 mg nitroglycerin, the atherosclerotic coronary segment to be reconstructed was examined with a mechanical IVUS system (ClearView, CardioVascular Imaging Systems Inc) and a sheath-based IVUS catheter incorporating a 30-MHz beveled, single-element transducer rotating at 1800 rpm (MicroView, CardioVascular Imaging Systems Inc). This catheter is equipped with a 2.9F 15-cm-long sonolucent distal sheath with a common lumen that alternatively houses the guidewire (during catheter introduction) or the transducer (during imaging after the guidewire has been pulled back), but not both. This design avoids direct contact of the IVUS imaging core with the vessel wall. The IVUS transducer was withdrawn through the stationary imaging sheath by an ECG-triggered pullback device with a stepping motor developed at the Thoraxcenter Rotterdam.28
ECG-Gated 3D IVUS Image Acquisition
The ECG-gated image acquisition and image digitization was performed by a workstation initially designed for the 3D reconstruction of echocardiographic images28 (Echoscan, TomTec). This workstation received input from the IVUS machine (video) and the patient (ECG signal) and on the other hand, controlled the motorized transducer pullback device.
The steering logic of the workstation considered the heart rate variability and checked for the presence of extrasystoles during image acquisition and digitization (Fig 1⇓). First, the RR intervals were measured over a 2-minute period to define the upper and lower limits of the range of acceptable RR intervals (mean value±50 ms). IVUS images were acquired 40 ms after the peak of the R wave. When the length of the RR interval met the preset range, the IVUS image was stored in the computer memory. Consecutively, the IVUS transducer was withdrawn 200 μm to acquire the next image. Although the longitudinal resolution available with this technical setup is 100 μm,28 in the present study only one IVUS image per 200 μm axial arterial length was acquired. Thus, an average of 114 images per segment were digitized and analyzed (range, 48 to 200 images per segment; corresponding segment length, 9.6 to 40.0 mm).
IVUS Analysis Protocol
Each set of digitized IVUS images was analyzed off-line by two independent observers using an automated, computerized contour detection algorithm.29 30 31 These measurements (Ia and II) were compared to study the interobserver variability. Blinded analyses were repeated by the first observer after an interval of at least 6 weeks. These measurements (Ia and Ib) were compared to study the intraobserver variability.
Two hundred planar images were randomly selected for “manual” analysis by a third investigator (MA-III) who was experienced in IVUS image analysis but blinded to the (above) automated contour detection results. This analyst could review the videotape to ensure a maximum accuracy of contour tracing, performed within an average of 4.1 minutes per image. Validation of manual CSA measurements by IVUS has been reported previously.32 33 34 These measurements were compared with the automated contour detection analysis made by observer I.
The CSA measurements included the lumen and EEM CSA. Plaque+media CSA was calculated as EEM minus lumen CSA, and the CSA plaque+media burden was calculated as plaque+media CSA divided by EEM CSA. The EEM CSA (which represents the area within the border between the hypoechoic media and the echoreflective adventitia) has been shown to be a reproducible measure of the total arterial CSA. As in many previous studies using IVUS, plaque+media CSA was used as a measure of atherosclerotic plaque, because ultrasound cannot measure media thickness accurately.35 Lumen, EEM, and plaque+media volumes were calculated as
where H is the thickness of a coronary artery slice, represented by a single tomographic IVUS image, and n is the number of IVUS images in the 3D data set. The volumetric plaque+media burden was calculated as plaque+media volume divided by EEM volume.
Plaque composition was assessed visually to identify lesion calcium. Calcium produced bright echoes (brighter than the reference adventitia), with acoustic shadowing of deeper arterial structures. The largest arc(s) of target lesion calcium was identified and measured in degrees with a protractor centered on the lumen. The overall length (in mm) of lesion calcium was measured by use of the length measurements provided by the 3D reconstruction.
Computerized Contour Detection in ECG-Gated 3D IVUS
Steps Involved in Image Analysis
Two longitudinal sections were constructed, and contours corresponding to the lumen-tissue and media-adventitia interfaces were automatically identified (Fig 1⇑). The necessity to manually edit these contours was significantly reduced, because cyclic “saw-shaped” image artifacts that can hamper the automated detection in nongated image sets were virtually abolished (Fig 2⇓). The sufficiency of the contour detection was visually checked, requiring an average of 5 minutes. If necessary, these longitudinal contours were edited with computer assistance (see below) within <1 minute. The longitudinal contours were transformed to individual edge points on the planar images, defining center and range of the automated boundary search on the planar images.
Subsequently, contour detection of the planar images was performed. The axial location of an individual planar image was indicated by a cursor, which was used to scroll through the entire set of planar images while the detected contours were visually checked. Correct detection of the longitudinal contours minimized the need for computer-assisted editing of the cross-sectional contours. Careful checking and editing of the contours of the planar images was performed within an average of 25 minutes. Finally, the contour data of the planar images were used for the computation of the results.
Minimum-Cost Algorithm and Computer-Assisted Contour Editing
A minimum-cost algorithm was used to detect the luminal and external vessel boundaries.29 Each digitized IVUS image was resampled in a radial format (64 radii per image); a cost matrix representing the edge strength was calculated from the image data. For the boundary between lumen and plaque, the cost value was defined by the spatial first derivative.36 For the external vessel boundary, a cross-correlation pattern matching process was used for the cost calculations. The path with the smallest accumulated value was determined by dynamic programming techniques.29 The computer-assisted editing differed considerably from conventional manual contour tracing. The computer mouse was pointed on the correct boundary to give that site a very low value in the cost matrix, and subsequently the automated detection of the minimum cost path was updated within <1 second. Editing the contour of a single slice caused the entire data set to be updated (dynamic programming).
Handling of Side Branches and Calcification
Side branches with a relatively small ostium were generally ignored by the algorithm as a result of its robustness, which means that the automated contour detection did not follow every abrupt change in the cost path. However, in branches with a large ostium, the contour did follow the lumen and vessel boundaries of the side branch. This was corrected by displaying the side branch in one of the longitudinal sections and interpolating the longitudinal vessel contours as straight lines. As a result, the side branch was outside the region of interest on the planar images. Similarly, small calcific portions of the plaque did not affect the detection of the external vessel boundary because of the robustness of the algorithm. In case of marked vessel wall calcification, the automated approach fails to detect the external vessel boundary. However, the 3D approach of the analysis system allowed interpretation of the external vessel boundary in the longitudinal dimension and facilitated tracing of a straight contour line behind the calcium.
Previous Validation In Vitro and In Vivo
In vitro, the algorithm has been validated in a tubular phantom consisting of several segments. The automated measurements revealed a high correlation with the true phantom areas and volumes (r=.99); mean differences were −0.7% to 3.9% (SD<2.6%) for the areas and 0.3% to 1.7% (SD<3.8%) for the volumes of the various segments.30 A comparison between automated 3D IVUS measurements in 13 atherosclerotic coronary specimen (area plaque+media burden <40%) in vitro and morphometric measurements on the corresponding histological sections revealed good correlations for measurements of lumen, EEM, plaque+media, and plaque+media burden (r=.94,.88,.80, and.88 for areas and.98,.91,.83, and.91 for volumes).31 In vitro, both area and volume measurements by the automated system agreed well with results obtained by manual tracing of IVUS images, showing low (−3.7% to 0.3%) mean between-method differences with SD <6% and high correlation coefficients (r≥.97 for areas and r=.99 for volumes).31 In vivo, using 3D IVUS image sets acquired during nongated continuous pullbacks through 20 diseased coronary segments, intraobserver and interobserver comparisons revealed high correlations (r=.95 to.98 for area and r=.99 for volume)30 and small mean differences (−0.9% to 1.1%), with SD of lumen, EEM, and plaque+media not exceeding 7.3%, 4.5%, and 10.9% for areas and 2.7%, 0.7%, and 2.8% for volumes. The time of (automated) analysis in that study was 69±19 minutes. Importantly, that study did not include segments with more than focal calcification, more than one side branch, or extensive systolic-diastolic movement artifacts in the longitudinally constructed images.
Quantitative data were given as mean±SD; qualitative data were presented as frequencies. According to Bland and Altman,37 the intraobserver and interobserver agreement (reproducibility) of the contour detection method was assessed by determining the mean and SD of the between-observation and between-observer differences, respectively. The results of the repeated contour analyses (Ia versus Ib), the independent contour detection analyses (Ia versus II), and the manual versus the contour analyses (III-MA versus Ia) were compared by the two-tailed Student’s t test for paired data analysis and linear regression analysis; values of P<.05 were considered statistically significant.
Feasibility and Acquisition and Processing Time
The gated IVUS image acquisition required 3.9±1.5 minutes (1.5 to 6.9 minutes) per coronary segment, which corresponds to 2.0±0.1 seconds (1.7 to 2.3 seconds) per image (Table 1⇓). All segments could be analyzed by the computerized contour detection system during an analysis time of 34.3±7.3 minutes per segment (21.3 to 48.4 minutes), corresponding to 0.3±0.1 minutes (0.2 to 0.5 minutes) per computerized IVUS image analysis.
IVUS Segment Characteristics
All but two of the segments (93%) contained at least one side branch (Table 2⇓). The average number of side branches per segment was 2.3±1.5 (range, 0 to 6). Calcification was present in 17 segments (57%), 11 (37%) showed a single calcium deposit, and 6 (20%) contained multiple calcium deposits. The maximum arc of calcium was 114±49° (50° to 190°); in 6 segments, the length of the calcified portion exceeded 1 mm.
The minimal lumen CSA as measured by the contour detection system was 6.4±1.7 mm2 (3.5 to 9.7 mm2). The maximum and average CSA plaque+media burden were 60.5±10.2% (31.7% to 77.7%) and 46.5±9.9% (22.8% to 65.9%).
Manual Tracing Versus Automated Contour Detection
In the 200 randomly selected image slices, the measurements of the lumen, EEM, and plaque+media CSAs and the CSA plaque+media burden obtained with the automated contour detection system (9.37±3.09 mm2, 18.33±6.70 mm2, 8.95±5.16 mm2, and 46.03±13.46%, respectively) were similar to the results obtained by manual tracing (9.35±3.18 mm2, 18.37±6.62 mm2, 9.02±5.08 mm2, and 46.53±13.41%; n=200). Between-method differences were 0.4±4.3%, −0.4±3.6%, −1.6±9.1%, and −1.2± 6.8%, respectively (all P=NS). The correlations between the measurements provided by both methods were high (r≥.98; Fig 3⇓).
Reproducibility of the Contour Detection Analysis
For measurements of lumen, EEM, and plaque+ media CSA and the CSA plaque+media burden (n=3421), both intraobserver (−0.4±2.7%, −0.4± 1.8%, −0.4±5.1%, and −0.0±4.2%) and interobserver (0.4±5.2%, −0.9±2.7%, −1.5±7.2%, and −1.5±6.9%; all P<.001) differences were low. Correlation coefficients were high for repeated measurements by the same observer (r=.99) and measurements by the two observers (r≥.97; Fig 4⇓). For the corresponding volumetric measurements (n=30), the intraobserver (−0.4±1.1%, −0.4±0.6%, −0.3±1.0%, and 0.0±0.4%) and interobserver (0.6±2.9%, −0.8±1.0%, −2.5±3.2%, and 0.8±1.5%; P<.05) differences were also low, and high correlations were found for both intraobserver and interobserver comparisons (r=.99; Fig 5⇓).
The present study demonstrates that (1) ECG-gated acquisition of 3D IVUS images is feasible, (2) there is a good agreement between the results provided by the automated contour detection method and manual border tracing, and (3) the automated contour detection analysis can be performed in a relatively short analysis time with a high degree of reproducibility.
3D reconstruction of IVUS images was first used to visually assess the spatial configuration of plaques, dissections, and stents and to perform basic measurements.16 17 19 More recently, the 3D reconstruction systems have included algorithms for automated quantification of lumen dimensions.16 17 18 19 20 21 25 26 27 The contour detection system used in the present study can be used for the detection of both the tissue-lumen boundary and the media-adventitia (EEM) boundary, and therefore plaque volume can be measured.
Non–ECG-gated image acquisition is frequently marred by cardiac cycle–linked coronary artery vasomotion and IVUS catheter motion, which produce sawtooth artifacts in the reconstructed 3D images that can interfere with automated contour detection (both the ease of use and, presumably, reproducibility). Conversely, in the present ECG-gated image sets, the longitudinal contours were smooth and without such artifacts. Therefore, there was much less need to manually edit the automatically detected longitudinal contours. Moreover, the accuracy of the derived edge information improved the performance of the second automated contour detection step on the planar IVUS images. This reduction in manual editing time on both longitudinal and planar images accounts for the low time of analysis compared with a previous study using nongated image acquisition30 (34 minutes and 69 minutes, respectively). Indeed, this represents a significant reduction in analysis time and as a consequence reduces the cost of the analysis. However, the ECG-gated 3D IVUS acquisition in the present study required a longer acquisition time than conventional motorized pullback (eg, non–ECG-triggered pullback at 0.5 mm/s). On average, only a 6-mm-long coronary segment could be imaged in 1 minute.
Reproducibility of the Contour Detection
In the present study, the measurement of the lumen, EEM, and plaque+media CSA differed little from the results obtained by manual contour tracing of these borders; there were only small interobserver and intraobserver differences in both the planar and volumetric analyses. However, the reproducibility of the plaque+media measurements was lower than for the other measures, which may reflect the combined variability of both the luminal and the EEM contours, confirming previous in vitro31 and in vivo data (nongated patient data)30 and findings of others.38 The reproducibility of the volumetric measurements was higher than for the CSA measurements, which may be a result of an averaging of the differences between the individual CSA measurements.
Although the segments in this ECG-gated contour detection study were nonselected and included calcified segments with some side branches, the reproducibility of the CSA measurements was consistently better than observed in a previous study using nongated contour detection.30 We believe that the key factors explaining the overall high reproducibility of automated contour detection observed in this study are (1) the integrated analyses of the conventional cross-sectional image slices with two longitudinal sections and (2) the facilitated and improved detection as a result of the smoothness of the contours on the ECG-gated longitudinal IVUS sections.
Reproducibility of Alternative Methods of Quantitative 3D IVUS
There is very little information on the reproducibility of 3D IVUS measurements using other measurement systems and algorithms. Matar and colleagues21 reported a Pearson’s correlation coefficient of .98 for an intraobserver study of lumen volume measurement by an automated threshold-based IVUS analysis system, confirming the low variability of the volumetric measurements observed in the present study. Another acoustic quantification system25 performs measurements of lumen CSA and volume, based on the automated detection of the blood pool in single IVUS images acquired at random during the cardiac cycle.21 25 Because the measurements are based on single-frame analysis, ECG-gated image acquisition may not influence the reproducibility of such systems.
Conversely, 3D contour detection–based analysis approaches benefit from an ECG-gated image acquisition.20 Sonka and associates39 40 developed an alternative 3D contour detection system that performs computerized detection of the luminal and external vascular boundaries in 3D sets of planar IVUS images without the additional information provided by the longitudinal contours. In their study,39 the correlation between automated and manually traced CSA measurements was quite good (r=.91 and.83 for lumen and plaque CSA, respectively). Using ECG-gated 3D IVUS, they found significantly improved results (r=.98 and.94 for lumen and plaque+media CSA, respectively),40 underlining the significance of ECG-gated IVUS image acquisition. Most likely, other promising contour detection algorithms41 42 for 3D analyses may also benefit from an ECG-gated image acquisition.
Potential Sources of Error and Study Limitations
Problems related to IVUS in general43 and to 3D reconstruction in particular22 23 may influence the contour detection process. The quality of the basic IVUS images is crucial to both planar and 3D image analysis.22 Incomplete visualization of the vessel wall, for example as caused by acoustic shadowing6 from lesion-associated calcium, hampers conventional planar IVUS analyses; however, 3D IVUS allows interpretation in the axial dimension and estimated contour tracing of the external vascular boundary. Image distortion caused by nonuniform transducer rotation or noncoaxial IVUS catheter position in the lumen may create artifacts both in planar images and in 3D reconstruction.22
Vessel curvatures may cause differences between the movement of the distal transducer tip and the proximal part of the catheter (although the use of sheath-based IVUS catheters reduces the latter problem) and a significant distortion of the 3D image reconstruction.
Most importantly, linear 3D systems such as used in this study can provide only approximate values of the volumetric parameters44 because they do not account for vascular curvatures and the real spatial geometry. In curved vascular segments, this results in an overestimation of plaque volume at the inner side (expansion) and an underestimation of plaque volume at the outer side (compression) of the curve.22 Approaches combining data obtained from angiography and IVUS45 46 47 48 can provide information on the real spatial geometry of the vessel. Unquestionably, the combined approaches have a unique potential, but currently these sophisticated techniques are still laborious, restricted to research applications, and not yet useful for routine off-line analysis of clinical IVUS examinations. In the present study, only relatively straight coronary segments, showing no more than mild vessel curvatures, were included. We felt that this premise was important to limit curve distortion–induced deviation of volumetric measurement,44 because linear 3D analysis systems do not account for vascular curvatures.
Compared with conventional motorized transducer pullback at a uniform speed, ECG-gated image acquisition takes longer, which may limit its use before intervention, especially in patients with very severe coronary stenoses. Therefore, we currently perform ECG-gated IVUS examinations during diagnostic or follow-up catheterizations and at the presumed end point of coronary interventions.
The examination of coronary arteries by IVUS permits the comprehensive assessment of atherosclerosis1 2 3 6 7 10 11 and the evaluation of the instantaneous27 49 and long-term effects of catheter-based interventions on the coronary lumen and plaque. To quantify these changes, anatomic landmarks such as side branches or spots of calcium can be used to define specific anatomic image slices for comparative analysis in serial studies.
The proposed 3D IVUS method, which permits reproducible and reliable contour detection of both lumen and plaque, may facilitate volumetric measurements10 30 31 and obviate the need for laborious analyses based on Simpson’s rule.15 Furthermore, the use of ECG-gated image acquisition28 increases the applicability of the contour detection algorithm by shortening the analysis time49 and increasing the reproducibility of the method. These advantages may be most significant in studies that are expected to show only small changes in plaque and/or lumen over time (eg, in trials evaluating the progression or regression of atherosclerosis during pharmacological therapy10 ). In addition, because the time from the peak of the R wave to image acquisition can be varied, this method can be used to study the cyclic (systole versus diastole) changes in vessel dimensions.
ECG-gated acquisition of 3D IVUS image sets is feasible and permits the application of automated contour detection to provide reproducible measurements of the lumen and atherosclerotic plaque CSA and volume in a relatively short analysis time.
Selected Abbreviations and Acronyms
|EEM||=||external elastic membrane|
Dr von Birgelen is the recipient of a fellowship of the German Research Society (DFG, Bonn, Germany).
- Received January 8, 1997.
- Revision received May 8, 1997.
- Accepted May 28, 1997.
- Copyright © 1997 by American Heart Association
Yock PG, Linker DT. Intravascular ultrasound: looking below the surface of vascular disease. Circulation. 1990;81:1715-1718.
Nissen SE, Gurley JC, Grines CL, Booth DC, McClure R, Berk M, Fischer C, DeMaria AN. Intravascular ultrasound assessment of lumen size and wall morphology in normal subjects and patients with coronary artery disease. Circulation. 1991;84:1087-1099.
Fitzgerald PJ, St Goar FG, Connolly AJ, Pinto FJ, Billingham ME, Popp RL, Yock PG. Intravascular ultrasound imaging of coronary arteries: is three layers the norm? Circulation. 1992;86:154-158.
Ge J, Erbel R, Rupprecht HJ, Koch L, Kearney P, Görge G, Haude M, Meyer J. Comparison of intravascular ultrasound and angiography in the assessment of myocardial bridging. Circulation. 1994;89:1725-1732.
Mintz GS, Painter JA, Pichard AD, Kent KM, Satler LF, Popma JJ, Chuang YC, Bucher TA, Sokolowicz LE, Leon MB. Atherosclerosis in angiographically ‘normal’ coronary artery reference segments: an intravascular ultrasound study with clinical correlations. J Am Coll Cardiol. 1995;25:1479-1485.
Erbel R, Ge J, Bockisch A, Kearney P, Görge G, Haude M, Schürmann D, Zamorano J, Rupprecht HJ, Meyer J. Value of intracoronary ultrasound and Doppler in the differentiation of angiographically normal coronary arteries: a prospective study in patients with angina pectoris. Eur Heart J. 1996;17:880-889.
Pinto FJ, St Goar FG, Gao SZ, Chenzbraun A, Fischell TA, Alderman EL, Schroeder JS, Popp RL. Immediate and one-year safety of intracoronary ultrasonic imaging: evaluation with serial quantitative angiography. Circulation. 1993;88:1709-1714.
Hausmann D, Erbel R, Alibelli-Chemarin MJ, Alibelli-Chemarin MJ, Boksch W, Caracciolo E, Cohn JM, Culp SC, Daniel WG, De Scherder I, Di Mario C, Ferguson JJ III, Fitzgerald PJ, Friedrich G, Ge J, Görge G, Hanrath P, Hodgson J McB, Isner JM, Jain S, Maier-Rudolph W, Mooney M, Moses JW, Mudra H, Pinto FJ, Smalling RW, Talley JD, Tobis JM, Walter PD, Weidinger F, Werner GS, Yeung AC, Yock PG. The safety of intracoronary ultrasound: a multicenter survey of 2207 examinations. Circulation. 1995;91:623-630.
von Birgelen C, Slager CJ, Di Mario C, de Feyter PJ, Serruys PW. Volumetric intracoronary ultrasound: a new maximum confidence approach for the quantitative assessment of progression-regression of atherosclerosis? Atherosclerosis. 1995;118(suppl):S103-S113.
Mintz GS, Popma JJ, Pichard AD, Kent KM, Satler LF, Chuang YC, DeFalco RA, Leon MB. Limitations of angiography in the assessment of plaque distribution in coronary artery disease: a systematic study of target lesion eccentricity in 1446 lesions. Circulation. 1996;93:924-931.
de Feyter PJ, Serruys PW, Davies MJ, Richardson P, Lubsen J, Oliver MF. Quantitative coronary angiography to measure progression and regression of coronary atherosclerosis: value, limitations, and implications for clinical trials. Circulation. 1991;84:412-423.
Serruys PW, de Jaegere P, Kiemeneij F, Macaya C, Rutsch W, Heyndrickx G, Emanuelsson H, Marco J, Legrand V, Materne P, Belardi J, Sigwart U, Colombo A, Goy JJ, van der Heuvel P, Delcan J, Morel AA, for the Benestent Study Group. A comparison of balloon expandable stent implantation with balloon angioplasty in patients with coronary artery disease. N Engl J Med. 1994;331:489-495.
von Birgelen C, Umans V, Di Mario C, Keane D, Gil R, Prati F, de Feyter PJ, Serruys PW. Mechanism of high-speed rotational atherectomy and adjunctive balloon angioplasty revisited by quantitative coronary angiography: edge detection versus videodensitometry. Am Heart J. 1995;130:405-412.
Rosenfield K, Losordo DW, Ramaswamy K, Pastore JO, Langevin RE, Razvi S, Kosowsky BD, Isner JM. Three-dimensional reconstruction of human coronary and peripheral arteries from images recorded during two-dimensional intravascular ultrasound examination. Circulation. 1991;84:1938-1956.
Mintz GS, Pichard AD, Satler LF, Popma JJ, Kent KM, Leon MB. Three-dimensional intravascular ultrasonography: reconstruction of endovascular stents in vitro and in vivo. Clin Ultrasound. 1993;21:609-615.
Roelandt JRTC, Di Mario C, Pandian NG, Li W, Keane D, Slager CJ, de Feyter PJ, Serruys PW. Three-dimensional reconstruction of intracoronary ultrasound images: rationale, approaches, problems, and directions. Circulation. 1994;90:1044-1055.
Di Mario C, von Birgelen C, Prati F, Soni B, Li W, Bruining N, de Jaegere PJ, de Feyter PJ, Serruys PW, Roelandt JRTC. Three-dimensional reconstruction of two-dimensional intracoronary ultrasound: clinical or research tool? Br Heart J. 1995;73(suppl 2):26-32.
von Birgelen C, Kutryk MJB, Gil R, Ozaki Y, Di Mario C, Roelandt JRTC, de Feyter PJ, Serruys PW. Quantification of the minimal luminal cross-sectional area after coronary stenting by two- and three-dimensional intravascular ultrasound versus edge detection and videodensitometry. Am J Cardiol. 1996;78:520-525.
von Birgelen C, Gil R, Ruygrok P, Prati F, Di Mario C, van der Giessen WJ, de Feyter PJ, Serruys PW. Optimized expansion of the Wallstent compared with the Palmaz-Schatz stent: online observations with two- and three-dimensional intracoronary ultrasound after angiographic guidance. Am Heart J. 1996;131:1067-1075.
Bruining N, von Birgelen C, Di Mario C, Prati F, Li W, Den Hood W, Patijn M, de Feyter PJ, Serruys PW, Roelandt JRTC. Dynamic three-dimensional reconstruction of ICUS images based on an ECG-gated pull-back device. In: Computers in Cardiology 1995. Los Alamitos, Calif: IEEE Computer Society Press; 1995:633-636.
Li W, von Birgelen C, Di Mario C, Boersma E, Gussenhoven EJ, van der Putten N, Bom N. Semi-automatic contour detection for volumetric quantification of intracoronary ultrasound. In: Computers in Cardiology 1994. Los Alamitos, Calif: IEEE Computer Society Press; 1994:277-280.
von Birgelen C, Di Mario C, Li W, Schuurbiers JCH, Slager CJ, de Feyter PJ, Roelandt JRTC, Serruys PW. Morphometric analysis in three-dimensional intracoronary ultrasound: an in-vitro and in-vivo study using a novel system for the contour detection of lumen and plaque. Am Heart J. 1996;132:516-527.
von Birgelen C, van der Lugt A, Nicosia A, Mintz GS, Gussenhoven EJ, de Vrey E, Mallus MT, Roelandt JRTC, Serruys PW, de Feyter PJ. Computerized assessment of coronary lumen and atherosclerotic plaque dimensions in three-dimensional intravascular ultrasound correlated with histomorphometry. Am J Cardiol. 1996;78:1202-1209.
Hodgson J McB, Graham SP, Sarakus AD, Dame SG, Stephens DN, Dhillon PS, Brands D, Sheehan H, Eberle MJ. Clinical percutaneous imaging of coronary anatomy using an over-the-wire ultrasound catheter system. Int J Card Imaging. 1989;4:186-193.
Tobis JM, Mallery JA, Gerrert J, Griffith J, Mahon D, Bessen M, Moriuchi M, McLeay L, McRae M, Henry WL. Intravascular ultrasound cross-sectional arterial imaging before and after balloon angioplasty in vitro. Circulation. 1989;80:873-882.
Li W, Bosch JG, Zhong Y, van Urk H, Gussenhoven EJ, Mastik F, van Egmond F, Rijsterborgh H, Reiber JHC, Bom N. Image segmentation and 3D reconstruction of intravascular ultrasound images. In: Wei Y, Gu B, eds. Acoustical Imaging, Vol 20. New York, NY: Plenum Press; 1993:489-496.
Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;2:307-310.
Sonka M, Zhang X, Siebes M, DeJong S, McKay CR, Collins SM. Automated segmentation of coronary wall and plaque from intravascular ultrasound image sequences. In: Computers in Cardiology 1994. Los Alamitos, Calif: IEEE Computer Society Press; 1994:281-284.
Sonka M, Liang W, Zhang X, De Jong S, Collins SM, McKay CR. Three-dimensional automated segmentation of coronary wall and plaque from intravascular ultrasound pullback sequences. In: Computers in Cardiology 1995. Los Alamitos, Calif: IEEE Computer Society Press; 1995:637-640.
Frank RJ, McPherson DD, Chandran KB, Dove EL. Optimal surface detection in intravascular ultrasound using multi-dimensional graph search. In: Computers in Cardiology 1996. Los Alamitos, Calif: IEEE Computer Society Press; 1996:45-48.
Brathwaite PA, Chandran KB, McPherson DD, Dove EL. Lumen detection in human IVUS images using region-growing. In: Computers in Cardiology 1996. Los Alamitos, Calif: IEEE Computer Society Press; 1996:37-40.
ten Hoff H, Gussenhoven EJ, Korbijn A, Mastik F, Lancee CT, Bom N. Mechanical scanning in intravascular ultrasound: artifacts and driving mechanisms. Eur J Ultrasound. 1995;2:227-237.
Waligora MJ, Vonesh MJ, Wiet SP, McPherson DD. Effect of vascular curvatures on three-dimensional reconstruction of intravascular ultrasound images Circulation. 1994;90(suppl I):I-227. Abstract.
Koch L, Kearney P, Erbel R, Roth T, Ge J, Brennecke R, Meyer J. Three dimensional reconstruction of intracoronary ultrasound images: roadmapping with simultaneously digitised coronary angiograms. In: Computers in Cardiology 1993. Los Alamitos, Calif: IEEE Computer Society Press; 1993:89-91.
Laban M, Oomen JA, Slager CJ, Wentzel JJ, Krams R, Schuurbiers JCH, den Boer A, von Birgelen C, Serruys PW, de Feyter PJ. ANGUS: a new approach to three-dimensional reconstruction of coronary vessels by combined use of angiography and intravascular ultrasound. In: Computers in Cardiology 1995. Los Alamitos, Calif: IEEE Computer Society Press; 1995:325-328.
Evans JL, Ng KH, Wiet SG, Vonesh MJ, Burns WB, Radvany MG, Kane BJ, Davidson CJ, Roth SI, Kramer BL, Meyers SN, McPherson DD. Accurate three-dimensional reconstruction of intravascular ultrasound data: spatially correct three-dimensional reconstructions. Circulation. 1996;93:567-576.
von Birgelen C, Mintz GS, Nicosia A, Foley DP, van der Giessen WJ, Bruining N, Airiian SG, Roelandt JRTC, de Feyter PJ, Serruys PW. Electrocardiogram-gated intravascular ultrasound image acquisition after coronary stent deployment facilitates on-line three-dimensional reconstruction and automated lumen quantification. J Am Coll Cardiol.. 1997;30:436-443.