Assessment of a procedure for left ventricle volumetry.

A three-step radiographic procedure is described for the determination of left ventricle volumes, based on one lateral and two angled single plane images. The first angled image is taken to yield a minimally foreshortened LV image and the second image, at the same angle, is taken of a calibration sphere. Five sources of operator-related, or 'input', error are analysed, two of which are normally removable. The remaining three input errors are assessed using a commercial digital subtraction angiography system and comparisons with the analysis are given. A summary result is that relative errors in end-systolic and end-diastolic LV volumes, as well as cardiac output, should normally be less than 13% under reasonable operator care. The input errors should not contribute to errors in calculated ejection fractions.