Errors Involved in Radiological Heart Volume Determination by the Ellipsoid-approximation Technique.

The radiological measurement of the heart's volume during life has been attempted by model-making (Palmieri, 1929), tomographic (Fuchs and Bayer, 1953), and simple formula techniques. The latter depend upon the assumption that the shape of the heart approximates to that of some regular figure, estimates of its volume then being obtainable from measurements of the heart's shadows in orthodiagraphic or teleradiographic projections. The simplest such assumption is that which considers the heart as a sphere, and on this basis volume estimates can be made from measurement of a single projection of the heart (Geigel, 1914). These estimates are-unsatisfactory, however, chiefly because the heart's depth bears no constant relation to the dimensions of its postero-anterior projection (Ludwig, 1939). Two orthodiagraphic projections at right angles to each other (frontal and sagittal) were used by Rohrer (1916/17) and Kahlstorf (1932), who based their volume estimates on planimetric and linear measurements and the assumption that the heart's shape is betwixt that of an ellipsoid and a paraboloid. Subsequently it has been usual to make only linear measurements of teleradiographic projections (Jonsell, 1939; Liljestrand et al., 1939), basing the estimates on the assumption that the heart's configuration is ellipsoidal. The use of teleradiographic projections demands the incorporation in the formula of a correction for magnification. This is conveniently made by the method of triangles, the theoretical objection that the divergent beam does not project the heart at the extremities of its axes being of no practical importance (Blasius, 1938). Comeau and White (1939) thought the Palmieri modelling technique to be the best available for measurement of heart volume during life but, like more recent modifications (Takahashi and Shinozaki, 1954), it involves too much expenditure of time, and too much irradiation of the patient, for routine use. Similar criticisms apply to the tomographic techniques, estimates from which have been compared with estimates provided by several formula methods (Broustet et al., 1960). The ellipsoid-approximation formula method has proved satisfactory during some 30 years of use in Sweden, and has also been found valuable for clinical studies in this country (Domenet, Evans, and Howarth, 1963). The need for estimates of heart volume as part of a research project led to an appraisal of the errors involved in the ellipsoid-approximation method, with a view to determination of the significance of measured differences. The first source of error requiring assessment was the obvious fact that most human hearts are not …