Comparison of different quantitative coronary analysis systems: ARTREK,CAAS, and CMS

It has been known that the first generation quantitative coronary analysis systems overestimate small vessel sizes. In the 2nd generation the contour detection algorithms, e.g., of the new Cardiovascular Measurement System (CMS), were modified to correct for the limited resolution of the X-ray imaging chain. This study validated and compared the CMS with the well-known Coronary Angiography Analysis System (CAAS) and the vessel tracking program ARTREK in a phantom study and a clinical study. In addition, the influence of different acquisition media (cinefilm vs. digitally acquired angiograms) on the accuracy of quantitative analysis was examined. The phantom study comprised 19 stenotic or non-stenotic glass tubes with a diameter range from 0.54 mm to 4.9 mm. In the clinical study the mean diameters of 322 coronary segments were analysed and the results of the systems were compared among each other. The results of the phantom study were presented in terms of the mean difference (accuracy) between true and measured values. In the phantom study the overall accuracy of the CMS was −6 μm (ARTREK: 85 μm; CAAS: 35 μm) with an overestimation of small vessels of only −11 μm (ARTREK: 97 μm: CAAS: 51 μm). The clinical study showed that the CMS corrected the usually occurring overestimation of small coronary arteries and that the influence on the accuracy of different acquisition media is minor. Due to the modified algorithms the new CMS is able to measure coronary diameters down to 0.5 mm accurately. Therefore, the CMS seems to provide more precise measurements in quantitative analysis of small coronary diameters than CAAS and ARTREK. © 1996 Wiley-Liss, Inc.     

Radiological quality of coronary guiding catheters: A quantitative analysis

Quantitative coronary angiography (QCA) is a validated and widely accepted method to investigate changes in arterial dimension over time. Calibration of measurements is enabled by the use of the coronary catheter as a scaling device. The dimensions and laminar composition of coronary catheters, however, have changed significantly over recent years and the suitability of the current generation of coronary catheters for calibration purposes has not been validated. We therefore recorded 57 coronary guiding catheters on cinefilm, and compared their automated quantitative measurements (Cardiovascular Angiography Analysis System, CAAS) with their true values (precision micrometer). We found an overall underestimation of quantitatively derived dimensions, ranging from ?8.9 to +4% for water-filled catheters and from ?15.5 to ?3.9% for contrast-filled catheters. In conclusion, while the current generation of coronary guiding catheters shows a wide variety in radiological quality, it can be clearly detected by the CAAS system, and is suitable for calibration of QCA measurements (with the exception of the DVI atherectomy catheter), provided that calibration is done on contrast-empty catheters. © Wiley-Liss, Inc.     

Elimination of variable vasomotor tone in studies with repeated quantitative coronary angiography

Summary In quantitative coronary angiographic studies, unintentional changes of coronary vasomotor tone may have a significant influence on the coronary artery diameters, thereby increasing the variability in the measurements. To obtain objective data on these measurement variabilities, two protocols were designed to assess the influences of ionic and nonionic radiographic contrast media on the mean diameters of angiographically normal coronary arteries. The vessel sizes were determined with the CAAS using automated edge detection techniques. In 21 patients (study no. I), coronary angiograms were taken in identical angiographic projections before (control), and immediately following several (at average 7) subsequent diagnostic dye injections administered over a period of about 7 min. The ionic contrast agent diatrizoate 76% induced a coronary dilation of 19 ± 7% (mean ± s.d., p<0.001; n=10); the nonionic agent iopromide 370 increased the coronary artery diameters by only 6 ± 4% (p<0.01; n=11). In another 11 patients (study no. II) coronary angiograms were obtained using the nonionic contrast medium iopamidol 300 at 5, 8, 10 and 11 min after the control acquisition; this protocol was repeated in the same patients with diatrizoate 76%. With iopamidol, coronary diameter changes were not significant at any time; with diatrizoate, however, coronary dilation was measured at 10 min (2 ± 2%; p<0.01) and at 11 min (10 ± 3%; p<0.001).In a third study it was tested, whether standardization of coronary vasomotor tone (e.g. in coronary angiographic follow-up studies) is possible by the induction of a reproducible maximum coronary dilation with nitrocompounds. In 12 patients, the mean diameters of angiographically normal coronary segments were analyzed before and at various times after i.v. administration (over 4 min) of 0.025 mg SIN-1/kg bodyweight. Coronary dilation was maximal at 10 or 15 min after the onset of the SIN-1-infusion (29 ± 5%; p<0.001). 0.8 mg nitroglycerin given s.l. at 15 min did not further dilate the coronary arteries (28 ± 7%). One hour after SIN-1, coronary dilation still amounted to an average of 24 ± 8% (p<0.001) and became maximal again, when 0.8 mg nitroglycerin was again administered sublingually (28 ± 8%; p<0.001).In conclusion, short-term variability of coronary vasomotor tone induced by ionic radiographic contrast media can be eliminated by the use of nonionic contrast agents and observation of injection intervals of at least 2 min. In quantitative coronary angiographic follow-up studies, as well as during acute interventions (e.g. PTCA), identical baseline vasomotor tone can be achieved by induction of the maximal coronary dilation using nitrocompounds.Dedicated to Prof. Paul R. Lichtlen on the occasion of his 60st birthday.     

In vivo validation of an experimental adaptive quantitative coronary angiography algorithm to circumvent overestimation of small luminal diameters

The reliability of quantitative coronary angiography (QCA) measurements is of fundamental importance for the study and practice of interventional cardiology. In vivo validation results have consistently reported a tendency for QCA systems to overestimate small luminal diameters. Such a systematic error may result in the underestimation of luminal gain during intracoronary procedures and in the underestimation of progression of coronary artery disease during longitudinal studies. We report the in vivo validation results of an experimental adaptive edge-detection algorithm that was developed to reduce overestimation of small luminal diameters by incorporating a dynamic function of variable kernel size of the derivative operator and variable weighting of the first and second derivatives of the brightness profile. The results of the experimental algorithm were compared to those of the conventional parent edge detection algorithm with fixed parameters. Dynamic adjustment of the edge-detection algorithm parameters was found to improve measurements of small (lt;0.8-mm) luminal diameters as evidenced by an intercept of +.07 mm for the algorithm with variable weighting compared to +0.21 mm for the parent algorithm with fixed weighting. A slope of <1 was found for both the parent and experimental algorithms with subsequent underestimation of large luminal diameters. Systematic errors in a QCA system can be identified and corrected by the execution of objective in vivo validation studies and the consequent refinement of edge-detection algorithms. The overestimation of small luminal diameters may be overcome by the incorporation of a dynamic edge-detection algorithm. Further refinements in edge-detection algorithms will be required to address the issue of underestimation of large luminal diameters before the absolute values derived from QCA measurements can be considered accurate over the full range of clinically encountered luminal diameters. © 1995 Wiley-Liss, Inc.     

Videodensitometric quantitative angiography after coronary balloon angioplasty, compared to edge-detection quantitative angiography and intracoronary ultrasound imaging.

AIMS: To assess the value of videodensitometric quantification of the coronary lumen after angioplasty by comparison to two other techniques of coronary artery lumen quantification. METHODS AND RESULTS: Videodensitometric quantitative angiography, edge detection quantitative angiography and 30 MHz intracoronary ultrasound imaging were performed after successful balloon angioplasty in 161 patients. Lumen cross-sectional areas were mean (SD) 2.82 (1.15) mm(2)for edge detection quantitative angiography, 3.67 (1.5) mm(2)for videodensitometric quantitative angiography and 5.32 (1.75) mm(2)for intracoronary ultrasound imaging (P<0.001). The correlation between intracoronary ultrasound imaging and videodensitometric quantitative angiography (r=0.44) was almost similar to that of intracoronary ultrasound imaging and edge detection quantitative angiography (r=0. 47). The correlation between the three techniques was not significantly influenced by the presence of ruptures and dissections on intracoronary ultrasound imaging. The absence of calcifications improved the correlation between videodensitometry and intracoronary ultrasound imaging. CONCLUSIONS: The luminal dimensions as measured by videodensitometric quantitative angiography matched intracoronary ultrasound imaging derived dimensions more closely than edge detection quantitative angiography. Videodensitometric quantitative angiography represents an on-line alternative to intracoronary ultrasound imaging for quantitative analysis regardless of the degree of vessel damage.     

Reproducibility of quantitative coronary analysis

Because of limited storage capacity for digital images, angiographic laboratories without cinefilm are dependent on locally performed quantitative coronary angiography (QCA) in clinical studies. In the present study the intra-and interobserver variability, as well as variability between different laboratories and variability due to frame selection was analyzed. A total of 20 coronary lesions were studied in two different digital laboratories 12±8 days apart. Images were analyzed on-line and after being transferred to a Cardiac Work Station (CWS). There was no significant difference between the measurement situations. For minimal luminal diameter (MLD) precision (SD of signed errors) ranged from 0.12 mm to 0.20 mm, for reference diameter (RD) from 0.15 mm to 0.28 mm, and for percent diameter stenosis (DS) from 4.2% to 5.8%. Overall relative precision was obtained by normalizing the QCA parameters, and was 11.9% for MLD, 7.0% for RD and 8.5% for DS (p<0.001, RD and DS compared to MLD). The overall variability in the interobserver and in the interlaboratory comparisons was 11.2% and 10.4%, respectively (n.s.) (n.s.). Thus the variability of QCA performed in cinefilmless, digital laboratories is small, and within a range making it an useful tool for clinical practice and group comparisons in clinical studies. However, the error range of QCA measurements must be taken into consideration when judging results from individual patients.     

Evolution of coronary atherosclerosis in patients with mild coronary artery disease studied by serial quantitative coronary angiography at 2 and 4 years follow-up

AIMS: Angiographic studies on the natural course of both focal anddiffuse coronary atherosclerosis have not been performed before,but can both be assessed by quantitative coronary angiography.The objective of this study was to describe the natural courseof focal and diffuse coronary atherosclerosis over time. METHODS AND RESULTS: In 129 patients with mild coronary artery disease, but not onlipid-lowering medication, three coronary angiograms were madeeach 2 years apart. Nine hundred and sixty five angiographicallydiseased and non-diseased segments were analysed by quantitativecoronary angiography. Mean lumen diameter and minimal lumendiameter were used as measures of diffuse and focal coronaryatherosclerosis. Mean lumen diameter and minimum lumen diameter decreased by0·02 and 0·03 mm per year. The rate of progresssionwas similar in the angiographically non-diseased, as in themildly and moderately diseased segments. Progression of diffusecoronary atherosclerosis was largest in severely stenosed lesions(percentage diameter stenosis 50%) and in the right coronaryartery with a loss of 0·19 mm and 0·16 mm in meanlumen diameter. Progression of focal disease was most prominentin new and mild lesions and the right coronary artery, witha decrease in minimum lumen diameter of 0·34 mm and 0·22mm. In most subgroups, progression occurred gradually over time.On a per segment level, progression and the occurrence of newlesions occurred in 4·4% and 4·2%. Regressionand disappearance of a lesions was found in 2·3% and1·9%. On a per patient level, 36% were progressors, 12%had a mixed response, 36% were stable, and 16% were regressors. CONCLUSION: Diffuse and focal coronary atherosclerosis progressed at thesame rate in the first and second 2 years in stenosed and non-stenosedsegments. The rate of coronary atherosclerosis progression wassmall, but was higher for focal than for diffuse disease. Aminority of lesions progressed and spontaneous regression wasrare.     

The impact of vessel and catheter position on the measurement accuracy in catheter-based quantitative coronary angiography

Background: The calculation of absolute artery dimensions in quantitative coronary angiography is usually carried out by catheter calibration. It is based on the proportional comparison of the dimension of the imaged artery segment to the dimension of the imaged angiographic catheter of known size. This calibration method presumes an identical radiographic magnification between angiographic catheter and artery segment of interest. However, due to the different intrathoracic location of both objects the radiographic magnification or calibration factor is often not identical for a given angiographic projection. The aim of this study was to quantify the magnification error (out-of-plane magnification error) for the major coronary artery segments imaged in frequently used angiographic projections. Methods The intrathoracic spatial location of 468 coronary segments (RCA 196, LAD 156, LCX 116) and their respective coronary catheters were established with biplane angiography and known imaging geometry data. The error in the radiographic magnification or calibration factor was then calculated for all 936 monoplane projections using the spatial coordinates and imaging geometry data. Results The mean magnitude of magnification error was 4% within all 936 measurements. The magnitude and direction of error varied with the lesion localization and the angiographic projection angle (range –12.6% to +10.6%). The error characteristics could be described with six typical error groups by stratifying the data according to the three main coronaries and two angiographic planes. In 24% of measurements, the magnification error exceeded the 5.2% error limit acceptable for reference vessel sizing. Measurements of left coronary arteries were mainly affected by it. Conclusion: The magnification error contributes to the calibration error in measuring arterial dimensions by quantitative angiography. This error may affect the reliability of clinical studies and the proper sizing of interventional devices. These findings could be used to improve current error correction algorithms in order to reduce the effect of the magnification error in measuring arterial dimensions.     

定量多层螺旋CT冠状动脉成像与定量冠状动脉造影的对比研究

目的通过与同期定量冠状动脉造影(QCA)作对照,评价定量冠状动脉多层螺旋CT(QCT)成像对冠状动脉疾病的诊断价值。方法选择冠心病患者78例,利用QCA和QCT测量同一病变的最小管腔直径、最小管腔面积、参考直径、参考面积、靶病变长度、直径狭窄率、面积狭窄率等指标,比较2种方法结果的相关性及一致性。结果 78例患者中,存在明显冠状动脉病变62例(79.49%),QCT和QCA一致认为重度狭窄(>75%)的节段共为47段,其中右冠状动脉1 5段,前降支27段,回旋支5段。QCT和QCA测量的最小管腔面积、面积狭窄率比较,差异有统计学意义(P<0.05),2种测量最小管腔直径、参考直径、参考面积、靶病变长度和直径狭窄率的比较,差并均无统计学意义(P>0.05)。QCT和QCA测量最小管腔直径,靶病变长度,参考直径,参考面积,直径狭窄率一致性较好(0.5≤r_c≤0.85),最小管腔面积、面积狭窄率则一致性不佳(r_c<0.5)。结论 QCT检查能良好评价冠状动脉病变.可用于术前即可指导对经皮冠状动脉介入治疗中球囊和支架的选择。     

Clinical assessment of a new real time 3D quantitative coronary angiography system: evaluation in stented vessel segments.

BACKGROUND: Foreshortening is a recognized problem that is present in angiography and results from views that are not perpendicular to coronary lesions. This limits visual coronary analysis as well as 2D quantitative coronary angiography systems (QCA). The CardiOp-B System is a 3D image acquisition and processing software system designed as an add-on to conventional X-ray angiography system. CardiOp-B's features include real time and off line analysis with comprehensive 3D reconstruction integrating all of the available information of two 2D vessel angiographies into one 3D image. It was the aim of the study to analyze the accuracy of this new 3D QCA system. METHODS: 3D QCA was performed in 50 patients (age 64 +/- 10.9; 84% male; LV-EF 63 +/- 16%) measuring 61 stents during high-pressure inflation (diameter: 2.25-4 mm; length: 8-32 mm). The obtained values (proximal and distal stent diameter, stent length) were correlated with the predefined size of the stents at the used inflation pressure. RESULTS: The linear correlation for the proximal stent diameter was Stent(prox)= 0.03 + 0.93 x real stent size (r(2) = 0.85). The linear correlation for the distal stent diameter was Stent(distal)= -0.03 + 0.89 x real stent size (r(2) = 0.81). The linear correlation for the stent length was Stent(length)= -0.61 + 1.02 x real stent length (r(2) = 0.98). CONCLUSIONS: The CardiOp-B System(R) is a new 3D QCA system with a high linear correlation between the real vessel size and the obtained vessel dimension. It provides real time or off line accurate and comprehensive diagnostic information to the interventional cardiologist without changing the basic coronary angiography procedure.     

Validation of coronary artery saphenous vein bypass graft diameter measurements using quantitative angiography

The accepted value for reproducibility (true change) is two standard deviations (SD) of the differences between repeat measurements. It has been well established for coronary artery measurements using several different quantitative coronary angiography (QCA) systems, but it has not been well documented for saphenous vein grafts (SVG). The purpose of this study was to assess, using the Cardiovascular Measurement System (CMS), the measurement reproducibility of 24 vein grafts from 24 patients who had symptom-directed control angiography. Three equal graft segments were studied separately. Focal narrowings expressed in percent stenosis varied from 5 to 80% (mean 20.8±15.9%). The average minimum lumen diameter (MLD) was 3.07±0.81 mm and the average interpolated reference diameter (Ref.D) was 3.87±0.58 mm. We assessed the reproducibility of measurements obtained from two separate imagings of the graft in the same view but at least 20 minutes apart, near the beginning and at the end of the angiographic procedure (simulating baseline and end-trial examinations). The SD for differences in measurements (variability) was 0.183 mm for the MLD, 0.193 mm for the Ref.D, 0.184 mm for the mean diameter (Mean D) and 3.72% for the percent diameter stenosis (PDS).A reasonable true change cut-off for SVG measurements in our laboratory is 0.4 mm for the minimum and mean lumen diameters, and 10% for the PDS, when QCA is obtained with the QCA-CMS analytical software package.     

A new intracoronary measurement catheter, MetriCath, compared to intravascular ultrasound and quantitative coronary angiography in a stented porcine coronary model.

The purpose of this study was to compare measurements by MetriCath to intravascular ultrasound (IVUS) and quantitative coronary angiography (QCA). The MetriCath system consists of a low-pressure (200 mm Hg) balloon catheter connected to a pressure transducer and infusion pump linked to a computer that records pressure-volume curves. Cross-sectional area of blood vessels is obtained directly from the unrestrained and in-stent pressure-volume measurements. We compared stent cross-sectional area measurements by MetriCath, IVUS, and QCA in a porcine stented coronary artery model. Comparison of area measurements in 14 stents showed no significant differences between the three methods (P = 0.66). On average, values differed 0.37 +/- 0.60 mm(2) between MetriCath and QCA, 0.13 +/- 0.55 mm(2) between MetriCath and IVUS, and 0.22 +/- 0.80 mm(2) between IVUS and QCA. This corresponds to 6.2% +/- 10%, 3.0% +/- 9.0%, and 3.1% +/- 12.9% relative difference from the average of two corresponding measurements. Linear regression analysis showed excellent correlation between measurements (r = 0.99 for all comparisons). The differences in in-stent area measurements between MetriCath and both QCA and IVUS were small. Considering the ease and rapidity of obtaining MetriCath results, this technique may form an alternative to the others in evaluating stent expansion. Based on these findings, clinical evaluation seems warranted.     

Comparison of two different methods of quantitative coronary angiography in patients with acute coronary syndromes.

The minimal cost algorithm (MCA) commonly used for quantitative coronary arteriography has limitations in definition of complex lesion morphology. A gradient field transform (GFT) algorithm has been designed for the better analysis of complex lesions. We compared MCA with GFT in angiograms of 125 patients in the Myocardial Infarction with Novastan and t-PA (MINT) trial. Lesion border definition was rated as one (poor), two (good), or three (very good). While MCA- and GFT-derived reference diameters (RDs) were similar, GFT yielded smaller minimal lumen diameter (MLD) than MCA by 0.22 +/- 0.31 mm (P < 0.01), and the difference between GFT- and MCA-derived MLDs increased with decreasing MLD. Mean percent diameter stenosis (% DS) was 9.1% +/- 11.1% greater by GFT (P < 0.001). Lesion border definition in simple lesions was similar (not significantly different). However, in complex lesions GFT performed better (2.49 +/- 0.61 vs. 2.11 +/- 0.74; P < 0.05). Thus, GFT appears to improve analysis of complex lesions compared to MCA. GFTs role in angiographic trials and clinical practice deserves further study.     

Utility of three-dimensional reconstruction of coronary angiography to guide percutaneous coronary intervention.

OBJECTIVE: The goal of this study was to determine whether three-dimensional (3D) reconstruction of traditional coronary angiography could optimize the choice of drug-eluting stent (DES) length and number during percutaneous coronary intervention (PCI). BACKGROUND: Coronary angiography is subject to significant foreshortening artifact that limits the ability of the operator to accurately determine lesion length. METHODS: The angiographic images of the target vessels of consecutive PCI procedures were postprocessed using a 3D reconstruction algorithm. The appropriate length and optimal number of DES to span each target lesion were calculated and compared with the number and length of DES actually chosen by the operator. RESULTS: A total of 42 target vessels were analyzed, and 3D reconstruction was successful in 38/42 (90.5%) of cases. The results of 3D analysis would have changed operator decision making in six cases (16%): in four cases, the stent chosen by the operator was too short requiring an additional DES; in two cases, the chosen DES was too long and exchanged for a shorter one. In each of these six cases, 3D analysis would have determined the correct stent length prior to stent selection. The optimal stent number derived by 3D reconstruction was significantly less than the actual number of stents per lesion used by the operator (1.31 +/- 0.47 versus 1.54 +/- 0.68, P = 0.01), and the optimal stent length trended less than the actual stented length (27.5 +/- 12.8 mm versus 28.7 +/- 14.7 mm, P = 0.23). CONCLUSIONS: 3D reconstruction algorithm of standard coronary angiography is a promising technique to improve DES utilization during PCI.