Brain Age Estimation Using Multiple Regression Analysis in Brain MR Images

Saadia Binte Alam, Ryosuke Nakano, Syoji Kobashi
2016 International Journal of Innovative Computing, Information and Control  
Evaluating the morphological changes in human brain and comparing it to normal data allow the risks of brain deformation related diseases to be assessed and the prevention process to be started earlier. Physiological age estimation based on human brain MR images has been an interesting research field over the past years. However, it is difficult to evaluate brain disorders based on morphological analysis as the brain deformation progress is different from person to person, and age range. To
more » ... uate brain deformation, this paper proposes an estimation method for both neonatal and adult brain age using manifold learning, principal component analysis, followed by multiple regression models. The regression model is automatically trained from a diverse set of subjects exhibiting significant variation, used to discover anatomical structure related to age and deformation, and fit new subjects to estimate age. The proposed method has been evaluated using 15 neonatal subjects (Revised age: 0-120 days; Mean ± SD: 34.13±42.15 days) and 150 adult subjects (Real age 18-96 years; Mean ± SD: 51.72 ± 22.16 years). Experiments demonstrate effective outcomes for each dataset with a distinctive remark of increased accuracy for common and age-specific methods.
doi:10.24507/ijicic.12.04.1385 fatcat:2takzmkvvzhwhh4fiv35h2gpjm