Left ventricular regression equations from single plane cine and digital X-ray ventriculograms revisited.

For the assessment of left ventricular volume from X-ray ventriculograms, widely known regression equations are used to correct for the irregular shape of the left ventricular lumen and the presence of the papillary muscles and trabeculations. These regression equations were derived in the late nineteen sixties and seventies. With all the changes in X-ray technology that have taken place over the past 20-30 years, the question was raised whether these regression equations were still valid. Therefore, 23 left ventricular casts of known volume were imaged in RAO20, RAO30 and RAO40 angiographic views and recorded on 35 mm cinefilm as well as in digital format. All the frames were traced manually by two observers and the volumes calculated by the Area Length and Simpson Rule approaches. The following conclusions could be drawn: inter- and intra-observer variations were small (systematic differences < 1.5 ml; random differences < 2.9 ml) and statistically not significant; the regression equations are virtually the same for the RAO20, RAO30 and RAO40 views under the different circumstances; the Area Length method was associated with slightly smaller values for the standard-error-of-the-estimate (SEE) suggesting a slight preference for this approach versus the Simpson Rule; significant differences were found between the cinefilm and digital regression equations; and the following new regression equations are proposed, which indeed differ significantly from the earliest proposals and less from the monoplane formulas proposed by Kennedy & Lange in the 1970s: [table: see text]