Investigation of signal extraction in the frame of LIBS imaging

Abstract Laser-induced breakdown spectroscopy (LIBS)-based imaging techniques have become well known among spatially resolved elemental approaches due to their mature instrumentation and outstanding advantages and applications. Data processing and in particular signal extraction are key in all LIBS-based imaging analyses to provide robust and reliable results. To date, there has not been a statistical evaluation of this issue when processing large and complex LIBS datasets. In this work, we aimed to test the performance of three extraction methods applied to micro-LIBS-based imaging. We also proposed a new conditional data extraction procedure relying on the statistical uncertainty associated with the extracted signal. We built a synthetic spectral dataset with controlled spectral features and tested the linearity, dynamic range and operating speed of different extraction approaches. The results of this study demonstrate the importance of data extraction and provide evidence for its optimization. This procedure is of particular relevance for the extraction of weak line intensities and in cases where the presence or absence of certain elements is critical (i.e., biomedical applications or trace analysis). In addition, the proposed conditional approach offers new insights into the means of providing LIBS imaging results.