Texture analysis of cerebral white matter in SIV-infected macaque monkeys

Image texture analysis is used in a wide variety of applications in medical research. Neurovirulent simian immunodeficiency virus (SIV) infection in monkeys is considered a good model for HIV-1 infection in humans and causes neuropathological changes in white matter which can include diffuse myelin pallor, subtle white matter astrocytosis, perivascular macrophage infiltrates, and microglial nodules with multinucleated giant cells. The ability of image texture analysis to quantify these changes was evaluated. Sections of thionin-stained brain tissue from eight male rhesus macaques ranging in age from 42-59 months were used. Four animals served as controls and four animals were infected with neurovirulent SIVmac239/17E-R71 by bone marrow inoculation. Images of cerebral white matter were captured and analyzed by calculating 13 textural features based on statistical analysis of spatial co-occurrence matrices. Statistical analysis of the results included multiple comparisons using the Newman-Keuls multiple range test. The effect of variation in background illumination used at image acquisition was also evaluated. Ten of the 13 textural features used in this study successfully discriminated between tissue from control and SIV-infected animals and were consistent with independent neuropathological assessment. Three textural features were highly sensitive to variation in background illumination and found not useful in this application.