Predicting ovulation from brain connectivity: Dynamic causal modelling of the menstrual cycle.

Abstract

Longitudinal menstrual cycle research allows the assessment of sex hormones effects on brain organization in a natural framework. Here, we used spectral dynamic causal modelling (spDCM) in a triple network model consisting of the default mode, salience and executive central networks (DMN, SN, and ECN), in order to address the changes in effective connectivity across the menstrual cycle. Sixty healthy young women were scanned three times (menses, pre-ovulatory and luteal phase) and spDCM was estimated for a total of 174 scans. Group level analysis using Parametric empirical Bayes showed lateralized and anterior-posterior changes in connectivity patterns depending on the cycle phase and related to the endogenous hormonal milieu. Right before ovulation the left insula recruited the frontoparietal network, while the right middle frontal gyrus decreased its connectivity to the precuneus. In exchange, the precuneus engaged bilateral angular gyrus, decoupling the DMN into anterior/posterior parts. During the luteal phase, bilateral insula engaged to each other decreasing the connectivity to parietal ECN, which in turn engaged the posterior DMN. Remarkably, the specific cycle phase in which a woman was in could be predicted by the connections that showed the strongest changes. These findings further corroborate the plasticity of the female brain in response to acute hormone fluctuations and have important implications for understanding the neuroendocrine interactions underlying cognitive changes along the menstrual cycle.