Greedy algorithm for error correction in automatically produced boundariesfrom low contrast ventriculograms

Non-homogeneous mixing of the dye with the blood in the left ventricle cham ber of the heart causes poor contrast in the ventriculograms. The pixel-bas ed classifiers [1] operating on these ventriculograms yield boundaries whic h are not close to ground truth boundaries as delineated by the cardiologis t. They have a mean boundary error of 6.4 mm and an error of 12.5 mm in the apex zone. These errors have a systematic positional and orientational bia s, the boundary being under-estimated in the apex zone. This paper discusse s two calibration methods: the identical coefficient and the independent co efficient to remove these systematic biases. From these methods, we constit ute a fused algorithm which reduces the boundary error compared to either o f the calibration methods. The algorithm, in a greedy way, computes which a nd how many vertices of the left ventricle boundary can be taken from the c omputed boundary of each method in order to best improve the performance. T he corrected boundaries have a mean error of less than 3.5 mm with a standa rd deviation of 3.4 mm over the approximately 6 x 10(4) vertices in the dat a set of 291 studies. Our method reduces the mean boundary error by 2.9 mm over the boundary produced by the classifier. We also show that the calibra tion algorithm performs better in the apex zone where the dye is unable to propagate. For end diastole, the: algorithm reduces the error in the apex z one by 8.5 mm over the pixel-based classifier boundaries.