Improved Infant MRI Brain Extraction utilizing Clustering and Morphological Approaches

Accurate brain extraction is a key procedure in neuroimage analyses. This paper aims to solve the intracranial cavity overestimation issue inherent to existing brain extraction methods when applied to infant brains. We applied k-means clustering method and morphological approaches to improve the accuracy of previously published brain extraction techniques. Evaluation of our proposed method on 28 preterm MR images showed more robust and effective infant brain extraction compared to previous methods.