A Comparative Evaluation of Cortical Thickness Measurement Techniques

In vivo measurements of cortical thickness from MR images have potentially widespread utility in the characterisation of normal brain development and maturation as well as in diagnosing and measuring the progress of a number of cortical pathologies. The literature describes several approaches to this problem, which may be divided into two groups: those relying on deformable models of the inner and outer cortical surfaces, and those relying on image intensities alone. Results from the former may be largely model driven at points deep within the sulci, where no apparent channel of cerebrospinal fluid can be seen at the resolution of typical MR images, potentially introducing bias. We present a comparative evaluation of cortical thickness measurement techniques, which demonstrates that approaches based on edge detection can provide cortical thickness measurements as accurate as those from model-based approaches, using less processor time, and without the possibility of bias from a model.