Statistical Methods in Neuroimaging: Brain Changes Induced by Fatherhood

Human brain presents a characteristic folding pattern. With the aid of new imaging technologies, such as magnetic resonance, we are able to study how these folding patterns evolve throughout different processes. Two of the most widely used magnetic resonance modalities are functional and structural. The first one studies which regions are activated under certain stimulus while the second one is used for neuroanatomic quantification. These neuroimaging data constitute a modern big-data problem because of the large amount of information provided by a single image. The common analyses in the field consist on testing each hypothesis, by means of adjusting linear models, in every point of the brain. This univariate approach produces a statistical map that is affected by false positives and multiple testing correction methods must be applied.

In the current work, we have applied a state-of-the-art neuroimaging methodology, that combined along with multiple testing correction strategies, enhances the sensitivity of depicting true positive results. It is the case of Threshold-Free Cluster Enhancement (TFCE), that enhances the detection of true positive results by including contiguous information since adjacent brain regions behave similar. In this project we exemplified these methods on a real neuroimaging study, showing the importance of the correction for false positive results. Specifically, we processed multimodal MRI data, coming from functional and structural modalities, using the above-mentioned statistical methods to obtain brain activations and morphometric traits (cortical volume, thickness and area). The particular neuroimaging case aimed to study how the human brain changes with fatherhood. For that, we compared longitudinally (before and after partner pregnancy) different morphological traits of the brain between a sample of first-time fathers and nulliparous males. We found that fathers exhibited volumetric reductions over time and that these reductions were associated with paternal brain responses (as measured with functional images).