Recent studies suggest good prospective accuracy for two-dimensional echocardiographic imaging of ventricular septal defects (VSD). We obtained two-dimensional images with high-frequency, high-resolution scanners from 36 patients proved by cardiac catheterization to have perimembranous VSD. In 20 patients, the VSD was an isolated lesion and in 16 it was associated with other forms of heart disease. VSDs were imaged in long-axis, apical four-chamber and subcostal echocardiographic views. The smallest VSD imaged was 2 mm in diameter on echo; the largest, 23 mm. The imaged size of VSDs was larger at end-diastole than at end-systole by paired t test on all views (all p less than 0.005). VSD size also varied between views, with no predictive relationship except between apical and subcostal four-chamber views in diastole (r = 0.71, p less than 0.005). This agreed with qualitative direct observations of an ellipsoid or irregularly shaped VSD in operated patients. Echocardiographically measured VSD size normalized for either aortic root size or for patient weight could be used to separate isolated VSDs with large shunts (Qp/Qs greater than 2:1) from those with small shunts. Review of 250 two-dimensional echocardiographic studies from patients proved not to have a VSD revealed 28 planes of imaging with false-positive VSD. None of the false-positive VSDs was imaged consistently on all views. Additionally, a "T" artifact (broadening of septal edges around a VSD) has been found to be a reliable marker of true VSD imaging. To best quantify VSD size and to avoid false-positive diagnoses, it is necessary to use multiple views and to consider the marked changes in VSD size that occur between diastole and systole.
