Two Automatic Training-Based Forced Calibration Algorithms for Left Ventricle Boundary Estimation i - Engineering in Medicine and Biology society, 1997. Proceedings of the 19th Annual International Co

Pixel classification algorithms [l] based on temporal information [3] when used in combination are not sufic,ent f o r boundaries produced by a pixel-based classifier are underestimated in the apical zone with respect to ground truth (GT) boundaries. timation developed at the author’s laboratories. The first mation, edge detection algorithms based on spatial inforboundary estimation of the left ventricle (LV) in Cardiovascular X-ray images. Poor contrast in the LV apex zone [4], Figure presents a two stage system for boundary esstage (upper consists of finding the raw boundaries using three kinds of approaches: first, Or shown in the fuzzy region in the inferior wall due to the overlap of the the Pixel classification approach [I], second, the edge deLV with the diaphragm, the inherent noise, and the variability of the modulation transfer function in x-ray imaging tection approach [3i, and third, the classifier-ed!Y fusion approach* The second stage (lower Or shown in system,T causes great dificulties in LV segmentation. T~ is the refinement stage of the left ventricle boundaries obOver come the above problems, calibration algorithms were developed b y sur i et al. [4/, [SI, [8/, [r/. These algorithms tained from the first stage. called the calibration system or correction syst,em behaving like a regThis stage is are training-based and provides a correction to the based classification or edge detection raw boundaries. This paper presents two training-based forced calibration algaszfiers. W e force the raw L V contour to pass through the ularizer to change the unsmoothed left ventricle curves to smooth left ventricle curves. This correction system is also called left ventricle calibration because it calibrates out the thought Of as a first layer Of networks where the rithms f o r correcting the raw boundaries produced b y claserrors introduced at stage 1. This calibration stage Can be LV apex and then perform the calibration. Over a database of 377 patient studies having end-diastole and end-systole global shape parameters are learned by the training sYskm and then applied on the test data to transform them frames, the mean boundary emor f o r the classifier system is 5 . 2 ~ the two forced calibration algorithms yield an emor of 3.14 mm and 3.04 mm with a standard deviation of 2.73 mm and 2.89 mm. K~~ Words: Boundaries, Force, Bias errors. for clinical purposes. Figure 1 also shows the off-line coefficient generation process which takes two inputs: the ideal boundary coordinates ( x , ~ ) and the raw boundary coordinates (2, y) generated at stage-I. These coefficients are then applied to the test raw boundary data sets. The off-line coefficients could be generated in three ways; from the edge boundary, from the classifier boundary or from the classifier-edge fusion boundary. The off-line coefficients can then be applied to the test edge boundary, or the test classifier boundary or to the test classifier-edge fused boundary. The final estimated boundary of the entire system undergoes equal arc interpolation and spline fitting followed by its performance evaluation. Suri et al. discussed stage-11 in detail in [4], [5] and [g]. classifier, Left Ventricle, L~~ Contrast,