Automated ventriculargram boundary delineation

Quantitative analysis of the left ventricle wall motion is important for the diagnosis of coronary heart disease. From wall motion, stroke volume, ejection fraction, and velocity of ventricular circumference can be estimated. These are helpful in identifying the state of diseased hearts. This paper describes a left ventricle boundary delineation algorithm which operates on X-Ray image data sets (ventriculargrams) produced in a catheterization laboratory for patient cardiovascular assessment. On a data set of 377 patient studies, the algorithm was able to delineate the boundary for the left ventricle with an average error just over 2.4 mm relative to the gold standard of cardiologist hand traced boundaries.