Bio-Raman non-negative matrix factorization: its practical methodology

Abstract

Research on live cells using a Raman microscope (bio-Raman research) has been attractive due to its versatility; but informative bio-Raman data has been complicated and largely sized. Non-negative matrix factorization (NMF) is expected to be an effective method to disentangle it; but the problem is that NMF does not give the unique decomposition, depending on different initial settings. That is, NMF causes cross-talks among factorized signals that disturb the quantitative analysis. To exemplify the problem, Raman imaging of a cross section of a rice grain was analyzed. To solve the problem, a practical methodology of bio-Raman NMF was described.