Particle detection from spatially modulated fluorescence signals

Abstract

Flow cytometry relies on the detection of cells selectively stained with fluorescence markers. Optically they can be detected as fluorescence particles. The use of microfluidics offers a wide range of benefits over traditional flow cytometer designs but when replacing expensive components with inexpensive counterparts the sensitivity of the instrument suffers. To increase the sensitivity of the detection system, spatial modulation has been proposed. Spatial modulation is implemented via fine pitched shadow masks close to the microfluidic channel which generate a signal pattern when a fluorescent particle passes by. Using a fine pitch and long total length for the pattern a high spatial resolution and long total exposure time are combined. Particle detection from local maxima is not directly possible with spatially modulated signals due to the jagged pulse shape. We compare the performance of different approaches for particle detection from local maxima. Matched filtering and the derivative of the correlation signal provide either a good peak-signal-to-noise ratio (PSNR) or a high spatial resolution. But both approaches suffer from low dynamic range due to side maxima. We derive the solution for a minimummean-square-error (MMSE) filter which transforms the modulated pulse shape into a target pulse shape with a single strong maximum. We investigate the performance of the MMSE filter and find that it provides tunable suppression of noise and side maxima along with a high spatial resolution. The use of the MMSE filter therefore is an ideal choice for particle detection from spatially modulated signals.