Computational Neuroanatomy of Schizophrenia

This chapter discusses methodology and application of the nascent field of computational anatomy for the quantification of neuroanatomical alterations in schizophrenia, as measured by MRI. The basics of morphological quantification via high-dimensional transformations are discussed and examples from two different studies are shown, indicating extensive reduction of cortical gray matter in frontal, temporal, and limbic regions in patients with schizophrenia. The heterogeneity of this pattern of cortical atrophy is further investigated in relation to the clinical phenotype of positive, negative, and disorganized clinical measures. The pattern of cortical abnormality is then synthesized into indices of diagnostic and predictive values for individual patients using machine learning methods. It is shown that individual patients can be distinguished to a large extent via their neuroanatomical profiles, and promising predictions of clinical progression in at-risk individuals can be obtained by measuring the patterns of brain change over time. Overall, these and other studies have provided strong evidence that computational neuroanatomy and machine learning methods are paving the way for subtler and more precise characterization of brain alterations in schizophrenia, with potential for better prediction of clinical outcome.