Separating composite signals in multi-probe dynamic biomedical imaging

Multi-probe dynamic biomedical imaging promises powerful tools for the visualization and elucidation of complex biological processes. Recent research aims to simultaneously dissect the spatial-temporal distributions of source signals that often represent a composite of multiple biomarkers independent of spatial resolution. We report a hybrid unmixing method for separating non-negative dependent imaging biomarker mixtures. The geodesic-principled algorithm exploits partial-volume modeling, non-negative clustered component analysis, and convex pyramid analysis, aided by a spatial-temporal coordinated information visualization aid. We demonstrate the principle of the approach on dynamic contrast-enhanced magnetic resonance imaging data and observed the expected vascular permeability and perfusion patterns due to tumor-induced angiogenesis and responses to therapy.