Ventricular enlargement in schizophrenia. The validity of computerised tomographic findings

There has been a considerable range in both the prevalence of ventricular enlargement and values for ventricular size of both schizophrenics and controls, in studies using computerised tomography (CT). The CT scans of 19 unrelated chronic schizophrenics, all of monozygotic (MZ) twin birth, and 36 age-and-sex-matched normal twins from 18 MZ pairs were examined by linear, planimetric, and semi-automated computerised methods. All methods distinguished schizophrenics from controls at approximately the same level of significance, but partial volume artefact led to a greater than two-fold variation in apparent ventricular size, and significantly reduced the validity and reliability of mechanical planimetric and linear measures. Measurement error may be an important source of artefact in uncontrolled studies using those techniques, and when comparing absolute values across studies, but the use of computerised methods should significantly reduce this source of artefact and allow more meaningful comparison.