Independent Component Analysis Applied to Pharmacological Magnetic Resonance Imaging (phMRI): New Insights Into the Functional Networks Underlying Panic Attacks as Induced by CCK-4

A.   C.   Dieler; P.   G.   Samann; G.      Leicht; D.      Eser; V.      Kirsch; T.   C.   Baghai; S.      Karch; C.      Schule; O.      Pogarell; M.      Czisch; R.      Rupprecht; C.      Mulert

doi:10.2174/138161208786848801

Abstract

Pharmacological magnetic resonance imaging (phMRI) is a method to study effects of psychopharmacological agents on neural activation. Changes of the blood oxygen level dependent (BOLD), the basis of functional MRI (fMRI), are typically obtained at relatively high sampling frequencies. This has more recently been exploited in the field of fMRI by applying independent component analysis (ICA), an explorative data analysis method decomposing activation into distinct neural networks. While already successfully used to investigate resting network and task-induced activity, its use in phMRI is new. Further extension of this method to tensorial probabilistic ICA (tensor PICA) allows to group similar brain activation across the anatomical, temporal, subject or session domain. This approach is useful for pharmacological experiments when no pharmacokinetic model exists. We exemplify this method using data from a placebo-controlled cholecystokinine- 4 (CCK-4) injection experiment performed on 16 neuropsychiatrically and medically healthy males (age 25.6 ± 4.2 years). Tensor PICA identified strong increases in activity in 12 networks. Comparison with results gained from the standard approach (voxelwise regression analysis) revealed good reproduction of areas previously associated with CCK-4 action, such as the anterior cingulate, orbitofrontal cortex, cerebellum, temporolateral, left parietal and insular areas, striatum, and precuneus. Several other components such as the dorsal anterior cingulate and medial prefrontal cortex were identified, suggesting higher sensitivity of the method. Exploration of the time courses of each activated network revealed differences, that might be lost when a fixed time course is modeled, e. g. neuronal responses to an acoustic warning signal prior to injection. Comparison of placebo and CCK-4 runs further showed that a proportion of networks are newly elicited by CCK-4 whereas other components are significantly active in the placebo conditions but further enhanced by CCK-4. In conclusion, group ICA is a promising tool for phMRI studies that allows quantifying and visualizing the modulation of neural networks by pharmacological interventions.

Keywords: Pharmacological magnetic resonance imaging (phMRI), independent component analysis (ICA), probabilistic ICA (PICA), tensor PICA, panic attacks (PA), panic disorder (PD), cholecystokinine-4 (CCK-4)

« Previous Next »