With modern digital cardiac systems the image data are digitized on-line and in real-time, allowing the replay and subsequent interpretation and analysis during or directly after the cardiac catheterization procedure. In this study we have evaluated the advantages and limitations of a manual tracing technique for left ventricular digital angiograms on the Phillips DCI system. Thirty-three patients who were catheterized for suspected coronary artery disease were studied. The manual tracings were performed by a senior cardiologist and an experienced function-analyst. It was found that the short- and long-term intraobserver variabilities in the assessment of the global ejection fraction were very small; short-term mean difference +/- standard deviation (correlation coefficient): 0.5 +/- 2.7 (r = 0.97) global EF%-units; long term; 0.7 +/- 2.7 (r = 0.96) EF%-units. The interobserver variabilities (5.1 +/- 4.8 (r = 0.93) EF%-units) were slightly higher than the intraobserver variabilities. A decrease by 25% in the amount of contrast medium administered did not significantly influence the variabilities in the contour tracings, which would suggest the use of smaller doses. At the average, the cardiologist and the function-analyst required 6 and 11 min of analysis time for a left ventricular study, respectively, emphasizing the need for further developments towards automated contour detection. Finally, an excellent correlation was found with a standard off-line cinefilm analysis procedure. Thus, it may be concluded that quantitative digital left ventricular angiography based on manual tracing of the outlines performed immediately following the cardiac catheterization (post-processing) is feasible as a routine procedure for the assessment of left ventricular function.

No abstract