The data for the first example comprise outlines of the corpus callosum in nearly midsagittal parasagittal sections extracted from ordinary clinical-quality magnetic resonance (MR) images of 13 schizophrenic patients and 12 normal adults (the patients' psychiatrists, actually). Original data collection was not at all elegant; it entailed manually outlining the callosum on these MR images one at a time. In this first example, slices were thick (5 mm), and so the rostrum was illegible. The group to which the image pertained (house staff or schizophrenic patient) was not known to the person carrying out the tracing. The full data set of 25 outlines is shown in Figure 4.1; solid lines are from doctors, dashed lines are from patients. There are 26 points in each outline. The groups are group-matched for age and sex; cases and most nearly midsagittal slices were selected by my colleague John DeQuardo, M.D.
Figure 4.1. A typical data set for studies of callosal shape: tracings of 25 callosal outlines from thick midsagittal MR images. Solid lines: normals; dotted lines: schizophrenics. The
In the jargon of morphometrics (see Slice et al., 1995), these data points are not landmark points having Latin-ate names and clearly punctate anatomical-histological definitions specimen by specimen. Instead, as is typically the case for studies of higher brain structures, they are semilandmarks that represent curving structures (in two dimensions, outlines like these) rather than discrete locations. The particular 26 points here were produced by a method of "sliding" that took into account the presence of some ancillary landmark points as explained and diagrammed by Bookstein (1997a, 1997c). Each of the 26 points is located along the callosal boundary in a manner that is optimal (in a sense to be explained below) given the locations of the other 25 points for its case and the sample Procrustes average of all the forms from both groups.
Was this article helpful?