Analysis of bioimaging and biospectra data has received increasingly attention in recent years.
Here we will present two experimental results based on independent component analysis (ICA):
differentiation of superparamagnetic iron oxide (SPIO) nanoparticles used as contrast agents in magnetic
resonance imaging (MRI), and differentiation of mixed chemical analytes by surface-enhanced Raman
scattering (SERS). The SPIO nanoparticles have been applied extensively as contrast agent in MRI for
tracking of stem cells, targeted detection of cancer, due to its biocompatible and biodegradable features.
For differentiation of SPIO from the background signal (e.g. interface between air and tissues), the signal
voids from multiple sources makes the task very difficult. To solve this problem, we assume that the
number of sensors corresponds to the number of acquisitions with different combinations of MR
parameters, i.e., longitudinal and transverse relaxation times. For detection of chemical and biological
analytes, the SERS approach has drawn more interest because of its high sensitivity. SERS spectra of
mixed analytes were acquired at different locations of a silver nanorod array substrate. Due to the nonuniform
diffusion and adsorption of the analytes, these spectra have been successfully used to identify the
characteristic SERS spectrum of individual analytes. In both the MRI and SERS data, signal source separation (SPIO or mixed chemical analytes from background signal) was performed on a pixel by pixel basis. The ICA was performed by a spatial analysis using the fast ICA method.