Statistical Shape Modeling from Gaussian Distributed Incomplete Data for Image Segmentation

Self-learning shape recognition in medical images

“…Aiming at addressing the training data contamination, early attempts are made to investigate imputation method (cf. our publication [114]) where the high frequency data in the probabilistic distribution are considered as outliers and re-estimated by the proposed method. Owing to the great success in corrupted data recovery, the Robust PCA (RPCA) is leveraged for statistical shape modeling which forms a basis of outlier handling for further methodologies.…”

“…Addressing the data contamination in linear distribution, many efforts have been devoted in this study. In our prior contribution [114], an imputation method is developed to cope with outliers in training shapes. The idea is to replace the identified outliers with the mean value derived from the population, where the outliers are determined in terms of data frequency in distribution.…”