A comparison of traditional wall motion assessment and quantitative shape analysis: a new method for characterizing left ventricular function in humans.

To forego the need to arbitrarily choose coordinate, reference, and indexing systems and to make other assumptions mandated by traditional methods of measuring wall motion, a technique of regional function analysis based on shape characteristics and pattern recognition was developed. The method is based on curvature analysis, a fundamental shape parameter, and is adaptive to the complex geometry of cineangiographic ventricular images. Quantitative shape parameters were compared to a standard method of regional function analysis (center-line method) in 130 patients. Quantitative shape and wall motion indexes showed a positive correlation over a broad range of normal and abnormal function (r = 0.748, p less than 0.001). Overall sensitivity and specificity for categorization of regional function were not statistically different for either technique. Within regions, however, shape criteria were more specific in categorizing inferior zones than anterior zones and were more often abnormal in the presence of mild regional abnormalities that were not located in the apical region. In conclusion, shape analysis and pattern recognition techniques can be used to forego dependence on the numerous assumptions and approximations required by traditional wall motion techniques, while providing performance characteristics that are similar to, and in some instances better than, traditional approaches. Incorporation of shape information in assessments of regional function provides a more comprehensive evaluation that includes the important visual cues used by experienced observers or "experts."