Bayesian inversion of multi-mode NEMS mass spectrometry signal

Abstract

Nano ElectroMechanical Systems are a new class of sensors that offers high sensitivity and opens new perspectives in the mass spectrometry field. This acquisition is performed in counting-mode, and the main tasks of associated information processing are to detect the molecules, to quantify their respective mass and to combine this information in order to recover the mass spectrum of the analysed solution. We propose a joint detection-quantification method based on a hierarchical description of the measurement system. Computation is done using a Reversible Jumps Monte-Carlo Markov-Chain algorithm. The approach we are describing in this communication solves the two problems of the joint impulse deconvolution on multiple output signals (multi-mode acquisition) and the non-linear relation between the observed signals and the mass of molecules, including the localization of the molecules on the sensor. We validate our method on both simulated and experimental data.