Characterization of a robust probabilistic framework for brain magnetic resonance image data distributions

Probabilistic characterisation of image data for accurate prognosis and treatment planning remains a long-standing problem in medical research, especially when the distribution depicts flat-top high-order contact. Such distributions are quite common brain magnetic resonance (MR) data, where density drops sharply beyond flat interval. Intuitively, it would indicate bipartition into positive region containing observations definitely belonging to class boundary with possibly it. The peak also imply that multiple values equally most likely belong class. However, popular probability used such cases unimodal, creating ambiguity about region. In this work, we study statistical properties develop likelihood-based iterative estimation method parameters novel platykurtic normal, called stomped normal distribution, provides more modelling distributions. robustness proposed model has been illustrated six simulated nine real MR volumes. Our analysis shows substantial improvement explaining variety shapes using model.