Model-free Group Analysis Shows Altered BOLD Networks in Dementia
Rombouts S, Scheltens P, Damoiseaux J, Barkhof F, Smith S, Beckmann C
VU University Medical Center
Tensorial probabilistic independent component analysis (T-PICA) is a newly developed, exploratory, model free technique that can be used to analyze group FMRI data. T-PICA was applied to analyze on/off FMRI data (face encoding versus fixation) in 18 patients with Alzheimers disease, 28 with mild cognitive impairment, and 41 healthy elderly controls. This showed decreased reactivity in patients in activated networks of cognitive, motor, and visual processing. Reactivity was also decreased in the deactivated default mode network. Furthermore, T-PICA appeared more sensitive to detect group differences than the commonly applied model-based technique.