Pooling Morphometric Estimates: A Statistical Equivalence Approach.

Changes in hardware or image-processing settings are a common issue for large multicenter studies. To pool MRI data acquired under these changed conditions, it is necessary to demonstrate that the changes do not affect MRI-based measurements. In these circumstances, classical inference testing is inappropriate because it is designed to detect differences, not prove similarity. We used a method known as statistical equivalence testing to address this limitation. Equivalence testing was carried out on 3 datasets: (1) cortical thickness and automated hippocampal volume estimates obtained from healthy individuals imaged using different multichannel head coils; (2) manual hippocampal volumetry obtained using two readers; and (3) corpus callosum area estimates obtained using an automated method with manual cleanup carried out by two readers. Equivalence testing was carried out using the "two one-sided tests" (TOST) approach. Power analyses of the TOST were used to estimate sample sizes required for well-powered equivalence testing analyses. Mean and standard deviation estimates from the automated hippocampal volume dataset were used to carry out an example power analysis. Cortical thickness values were found to be equivalent over 61% of the cortex when different head coils were used (q < .05, false discovery rate correction). Automated hippocampal volume estimates obtained using the same two coils were statistically equivalent (TOST P = 4.28 × 10(-15) ). Manual hippocampal volume estimates obtained using two readers were not statistically equivalent (TOST P = .97). The use of different readers to carry out limited correction of automated corpus callosum segmentations yielded equivalent area estimates (TOST P = 1.28 × 10(-14) ). Power analysis of simulated and automated hippocampal volume data demonstrated that the equivalence margin affects the number of subjects required for well-powered equivalence tests. We have presented a statistical method for determining if morphometric measures obtained under variable conditions can be pooled. The equivalence testing technique is applicable for analyses in which experimental conditions vary over the course of the study.

Quantitative assessment of corpus callosum morphology in periventricular nodular heterotopia.

We investigated systematic differences in corpus callosum morphology in periventricular nodular heterotopia (PVNH). Differences in corpus callosum mid-sagittal area and subregional area changes were measured using an automated software-based method. Heterotopic gray matter deposits were automatically labeled and compared with corpus callosum changes. The spatial pattern of corpus callosum changes were interpreted in the context of the characteristic anterior-posterior development of the corpus callosum in healthy individuals. Individuals with periventricular nodular heterotopia were imaged at the Melbourne Brain Center or as part of the multi-site Epilepsy Phenome Genome project. Whole brain T1 weighted MRI was acquired in cases (n=48) and controls (n=663). The corpus callosum was segmented on the mid-sagittal plane using the software "yuki". Heterotopic gray matter and intracranial brain volume was measured using Freesurfer. Differences in corpus callosum area and subregional areas were assessed, as well as the relationship between corpus callosum area and heterotopic GM volume. The anterior-posterior distribution of corpus callosum changes and heterotopic GM nodules were quantified using a novel metric and compared with each other. Corpus callosum area was reduced by 14% in PVNH (p=1.59×10(-9)). The magnitude of the effect was least in the genu (7% reduction) and greatest in the isthmus and splenium (26% reduction). Individuals with higher heterotopic GM volume had a smaller corpus callosum. Heterotopic GM volume was highest in posterior brain regions, however there was no linear relationship between the anterior-posterior position of corpus callosum changes and PVNH nodules. Reduced corpus callosum area is strongly associated with PVNH, and is probably associated with abnormal brain development in this neurological disorder. The primarily posterior corpus callosum changes may inform our understanding of the etiology of PVNH. Our results suggest that interhemispheric pathways are affected in PVNH.