Ultra-high field GRE-MRI phase images present great promise for structural brain studies. Multi-echo GRE-MRI data has been shown to contain signal compartments, which may eventually be used to characterise brain microstructure. Existing studies considered three signal compartments, however it remains unclear how compartments co-localise throughout the brain. We compartmentalised the signal via frequency shift signatures in a mixture of grey-white matter brain regions and implemented quality of fit measures to select the most parsimonious model for each region. We utilised k-means cluster analyses to investigate signal compartment commonalities across different brain regions and found four dominant frequency shift signatures.