Statistical thinking in functional and structural magnetic resonance neuroimaging.

Burning issues abound in the emerging field of biostatistics for imaging studies of the living human brain. This paper employs two examples to discuss the roles of informed empirical models and modern statistical methods in the analysis of functional and structural magnetic resonance neuroimages. The first example derives from an analysis of a functional baseline study in which simulated signals are embedded in actual brain 'noise' generated by a resting subject lying passively in the imaging device. The second example is taken from a structural volumetric analysis of human brain structure size differences in a sample of healthy adults.