Three-dimensional contour detection of left ventricular myocardium using elastic surfaces.

Since the human heart has a complex anatomy, the two-dimensional analysis of myocardial scintigrams obviously is not satisfactory. Three-dimensional display can be more easily read by clinicians and depicts more accurately tracer accumulation defects. In this study we propose a three-dimensional myocardial contour detection approach using elastic surfaces. After manual reorientation into short-axis slices and transformation into heart coordinates, the myocardial mid-wall surface passing through the radial activity mass points is calculated using a second-order partial differential equation as a mathematical model. Special considerations are implemented to demarcate organs close to the heart. In a total of 1102 myocardial scintigrams the three-dimensional contour detection calculated a reasonable shape of the left ventricular myocardium in all cases. Even in areas with severe myocardial infarction, this approach was able to bridge tracer accumulation defects. Inter- and intra-observer variability tests confirmed a high reproducibility and user independence. In conclusion, in this study the proposed three-dimensional contour detection of left ventricular myocardium using elastic surfaces was found to be fast, user-independent and reliable, leading to realistic three-dimensional parametric images of myocardial scintigrams that even clinicians not experienced in nuclear cardiology can easily read.