High-resolution magnetic resonance imaging sinc-interpolation-based subvoxel registration and semi-automated quantitative lateral ventricular morphology employing threshold computation and binary image creation in the study of fatty acid interventions in schizophrenia, depression, chronic fatigue syndrome and Huntington's disease.

Serial high-resolution structural magnetic resonance imaging scans of the brain can now be precisely aligned, with six degrees of freedom (three mutually orthogonal translational and three rotational degrees of freedom around three mutually orthogonal axes), using a rigid-body subvoxel registration technique. This is driven by the in-plane point spread function for images acquired in the Fourier domain with data obtained over a bounded region of k-space, namely the sinc interpolation function, where sinc z = (sin z)/z, with z being any complex number (including zero). Computational subtraction of the three-dimensional Cartesian spatial representation matrices of serially acquired scan data allows for the determination of structural cerebral changes with great precision, since voxel signals from unchanged structures are almost completely cancelled. Thus changes readily show up against a background of noise. Furthermore, lateral ventricular changes can now be accurately quantified using a semi-automated method involving contour production, threshold computation, binary image creation and ventricular extraction. These techniques have been applied to the investigation of the effects on cerebral structure of intervention with fatty acids, particularly the long-chain polyunsaturated n-3 fatty acid eicosapentaenoic acid (EPA), in disorders such as schizophrenia, treatment-resistant depression, chronic fatigue syndrome (myalgic encephalomyelitis or ME), and Huntington's disease.