HMM-based brain age interpolation using kriging estimator

Age-related gray matter (GM) volume loss becomes widespread from middle age and the decreasing rate shows linear relationship with real age. These characteristics provide us evidences of brain ageing following certain patterns. Magnetic resonance imaging (MRI) based brain structural hidden Markov model (HMM) is a method to model brain ageing pattern, and further detect the patterns deviating from normal pattern to provide evidence of neurodegenerative disease for clinical diagnosis. It requires a complete candidate MRI database covering all age stages which is usually not easily available. Therefore, we propose a HMM-based brain age interpolation using kriging estimator for the creation of HMM of the uncollected MRI. Preliminary results show that the proposed interpolation scheme can create HMM with the age prediction error of 1.29 years which is better than other current brain age prediction methods.