994-96 The Impact of the Vessel Position on the Accuracy of Vessel Measurement in Routine Quantitative Coronary Arteriography 320A ABSTRACTS JACC Febreary 1995 Methods: The effect on visual assessment of stenosis severity of digi- tal coronary angiograms was evaluated on a set of 100 image sequences using lossy (15: 1) Joint Photographic Experts Group (JPEG) compression software. A panel of 3 angiographers reviewed 6200 frames from 10 clini- cal exams both with and without standard JPEG compression. Original and compressed versions of each sequence were viewed in random order by all panel members blinded to compression status. Images were viewed on the same display used for digital acquisition with the Philips Digital Cardiac Imag- ing (DCI) System. Panelists identified and graded severity of 99 stenoses in quartiles with >50% considered to be significant. Results: There was no significant difference in the number of lesions iden- tified with each modality. Overall agreement for severity of all lesions be- tween compressed and non-compressed modalities was 0.60 and Kappa = 0.52. If lesions were dichotomized into "significant" (>50%) or "insignifi- cant" (::::50%). agreement was 0.94 and Kappa = 0.88, suggesting that when disagreement occurred, it tended to be within one severity grade. These agreement statistics are consistent with previously reported intra-observer variability in the review of cine-coronary angiograms. Conclusions: The significant reduction in digital storage and exchange re- quirements provided by lossy JPEG does not result in a decrease in diag- nostic quality of digital coronary angiograms. Variability in visual assessment from original and compressed data formats is comparable to intra-observer variability from identical data formats. Therefore, JPEG compression does not result in loss of diagnostic information and is a valid means of reducing storage and exchange requirements of coronary angiograms. Angiographic projection: tions of the vessel and catheter. The error in converting the vessel's size is dependent on the fractional positions of the catheter and vessel along their illuminating x-ray beams. We studied in 207 daily biplane angiograms the general magnitude of the measuring error, as well as its fractional parts with regard to the artery being studied (LAD n = =5, RCA n = 53, LCX n = 59) and the selected angiographic projection (RAO 30°, LAO 60°). Results: The position difference between the vessel and catheter causes a measuring error of more than 5% in one third of the analyses. The resulting error in the vessel measurements (mean ± std) is: The measuring error for the LAD coronary artery is in LAO 60° projection 45%, for the RCA in RAO 30° projection 65% and for the LCX coronary artery in LAO 30° projection 22% lower, than in the opposite projections. Conclusions: Since the position caused errors might be superimposed by detection errors, the total measuring error, using the coronary catheter for conversion to absolute vessel sizes, might increase markedly. Improve- ment of the measuring accuracy might be achieved by: (i) Selection of an angiographic projection with minimum displacement of the artery from the catheter, (ii) error balancing by emptylfilled catheters. Catheterization labora- tories equipped for biplane angiography should apply biplane angiographic correction procedures for position errors or use analytic calibration, which calculates the angiographic magnification of a vessel directly. 0.12±004 0.21 ± 0.06 LAO 60° 012±0.05 021 ± 0.07 RAO 30° lesion diameter error Imml: normal diameter error [mml: Variability Sources in Quantitative Coronary Arteriography 1994-95 1 Quantitating Pulmonary Capillary Volume Using Digital Parametric Angiographic Analysis Paul A. Robiolio, Vera H. Rigolin, John S. Wilson, Jack 1. Cusma, Thomas M. Bashore, J. Kevin Harrison. Duke University Medical Center, Durham, NC Assessment of the distal pulmonary vasculature in patients with pulmonary hypertension has been limited to qualitative description of pulmonary arteri- ograms. Digital parametric imaging, using contrast density and transit time, has been used to quantitate blood volume and flow in the coronary and re- nal vascular beds. This study was performed to determine whether digital parametric imaging can quantitate vascular volume in the distal pulmonary capillary bed with pulmonary flow intact. Two digital angiograms of the pulmonary vasculature were acquired in 11 patients with varying degrees of pulmonary hypertension. A balloon flotation catheter was advanced distally into the pulmonary artery. The first angiogram (static image) was performed with blood flow occluded by inflation of the catheter balloon. Non-ionic contrast was then hand injected to completely fill the vasculature beyond the balloon occlusion. The second angiogram (flow image) was performed with the balloon deflated and blood flow preserved. A hand injection of a rapid bolus of contrast, 1-2 cc, was given. Digital subtrac- tion image data were obtained at 15 frames/sec at end expiration for both angiograms. Contrast density measurements of the distal pulmonary vascu- lature were determined from the static images in various 2 x 2 mm areas using digital parametric imaging. The maximum density in these same areas was similarly determined from the flow images. The correlation of the density measurements between the static and the flow images in 64 regions of interest was excellent (R = 0.92, regression slope = 0.98). This correlation was similar to that observed for repeated in- jections using the same technique (flow image) (R = 0.97, regression slope = 0.97). Conclusion: Digital parametric measurements of pulmonary capillary vol- ume obtained with blood flow preserved are the same as those obtained with flow occluded and the entire bed replaced by contrast. This method allows quantitation of pulmonary vascular volume and flow using a simple, single contrast injection in the distal pulmonary artery. Uncertainties in radiation risk estimates at low doses «0.1 Gy) include the shape of the dose-response curve, use of a relative or absolute risk model, and the length of the latent cancer induction period. Coronary procedures are often repeated within short in many patients, but neither absorbed doses nor imparted energies are routinely measured. We used LiF thermolumines- cence dosimeters in 15 consecutive diagnostic (D) and 15 PTCA (I) proce- dures, with stent implantation in 1 case, multivessel PTCA in 2, and PTCA of chronic occlusion in 2. A Philips Optimus 2000 DCI was used, with a standard 1994-971 Patients' Radiation Risk During Diagnostic and Interventional Coronary Procedures Flavio Ribichini, Giuseppe Steffenino, Antonio Dellavalle, Veronica Rossetti, Riccardo Cerati, Mario Garbarino 1. Division of CardiOlogy Ospedale S. Croce, Cuneo, Italy: 1 Fiat Sepin. Torino, Italy 1994-981 5.70 590 MEASRATE 4.38 340 923 1520 FRAME Independent variable 23.67 100 STEN Coefficient of variation % contribution to total variance Thomas Linderer, Karl Wegscheider, Werner Wunderlich, Frank Fischer. Div. of Cardiology, KJinikum Benjamin Franklin, Free University Berlin, FRG In a trial of progression/regression of coronary artery disease the results of quantitative coronary arteriography are affected by the following main fac- tors: (i) the frame selected for analysis (FRAME). according to the general agree- ment, that the lesion should be measured at enddiastole. (ii)the frame rate (RATE). To obtain a truelyenddiastolic image of a coronary lesion, a cine frame rate of 25 frames/sec is mandatory up to date. Newer digital equipped systems allow to use a frame rate of 12.5/sec, but there is concern, that one miss a truely representative enddiastolic image. (iii) the measurement variance (MEAS) obtained from repeated measure- ments. We analyzed the impact of these variability sources on the measurements in a study of 29 coronary lesions. The lesions were filmed at 25 and 12.5 frames/sec. The truely enddiastolic frame as well as the frame preceed- ing and following it was analyzed. Each frame was measured twice, using computer-assisted analysis of vessels. A nested multivariate analysis of vari- ance was developed to quantify the effects of the independent variables RATE (12.5 instead of 25/sec). FRAME (enddiastole or a frame deviating from it) and MEAS (measuring the same frame twice) on the "outcome" in the sample the mean % diameter stenosis. The total variance in the sample by considering different stenosis (STEN; 15-75% diameter stenosis) was set to 100%. Results: Multivariate analysis of variance shows the following influence of the various components on the size of % diameter stenosis: The Impact of the Vessel Position on the Accuracy of Vessel Measurement in Routine Quantitative Coronary Arteriography Werner Wunderlich, Beate Rohrig, Frank Fischer, Thomas Linderer. Div. of Cardiology, Klinikum Benjamin Franklin, Free University Berlin, FRG Conclusions: Frame selection is the major source of variability quantifying coronary lesions. Compared to the total variance the variance attributable to frame selection is nearly 3 times higher than the measurement variance and nearly 5 times higherthanthe rate attributable variance. Thus, one hasto take great care of selecting appropriate frames and may use the lower frame rate (12.5/sec) to reduce radiation exposure and facilitate digital image archiving. To convert computer-detected vessel dimensions in digitized angiograms from pixels into millimeters, the coronary catheter is commonly used as a calibration object. The absolute size of the vessel is then inferred from the ratio of the known to detected size of the catheter. However, the reliability of this inference is significantly affected by different angiographic magnifica- 1994-961