Skip to Main Content
Magnetic resonance imaging studies of the resting brain have recently revealed the existence of low-frequency fluctuations of the cerebral hemodynamics. It has been suggested that these fluctuations are linked to baseline neural activity, organized in functional networks. This paper presents a novel method for the detection of these resting-state networks from blood-oxygen level dependent signals, through their wavelet representation in the appropriate frequency range. A valley-seeking clustering principle is employed, requiring no a priori knowledge of the number of functional networks. The technique is applied to a data set acquired at rest and is shown to retrieve a number of identifiable functional networks. The method is proposed as an alternative to e.g. independent component analysis and exhibits an enhanced network separation capability and stability against noise.