Abstract
Although conventional two-dimensional and Doppler blood-flow
echocardiography are the standard imaging approaches in the assessment of heart
disease they do not provide anatomic reconstructions in a form that resembles the
cardiac morphology as visualized by the surgeon.
The work presented in this thesis has explored the hypotheses that threedimensional echocardiography facilitates spatial recognition of intracardiac
structures and therefore enhances the diagnostic confidence of echocardiography in
congenital and acquired heart disease. The accuracy of three-dimensional
reconstructions has been validated in vitro using two different phantoms and in vivo
comparing the results with other established diagnostic techniques or surgical
findings. Additionally, as the main limitation of transthoracic three-dimensional
echocardiography is poor image quality in a substantial proportion of adult patients,
Doppler myocardial imaging has been tested as a potentially superior method to
conventional grey-scale imaging for transthoracic three-dimensional image
acquisition.
In vitro, using a virtual computer-generated phantom and a dynamic tissuemimicking phantom, the accuracy of both linear measurements and volume
computation obtained from three-dimensional images was established. For both
grey-scale and Doppler myocardial imaging, a detail of 1.0 mm dimension and two
details separated from each other by a distance of 1.0 mm were the smallest
structures and distances identified from a three-dimensional image. When testing the
accuracy of volume measurements it appeared that both techniques marginally
underestimated the true phantom volume (by approximately 1.0 ml for Doppler
myocardial imaging and 4.0 ml for grey-scale imaging), but the systematic error was
smaller and more constant in the case of Doppler myocardial imaging over the range
of different true volumes.
In vivo, the study was designed to compare the accuracy of grey-scale and
Doppler myocardial imaging three-dimensional left ventricular volume
measurements and cineventriculography. The differences were significantly smaller
for the Doppler technique during both end-diastole and end-systole. A series of
congenital heart lesions has also been studied. It has been shown that dynamic
surgical reconstruction of the secundum atrial septal defect is feasible from the
transthoracic approach in all patients. However, in adults, Doppler myocardial
imaging proved more effective than grey-scale imaging in the accuracy of threedimensional defect reconstruction. In patients with sinus venosus atrial septal defect,
transthoracic three-dimensional echocardiography was more accurate than standard
echocardiography in diagnosing the defect including a detailed description of the
abnormal pulmonary venous drainage. Finally, in children with atrio-ventricular
septal defects, the 'unroofed' atrial reconstruction of the common valve accurately
displayed dynamic valve morphology en face and the mechanism of valve reflux.
