[Zurück]

A. dos Santos, R. Heydenreich, C. Derntl, A. Mach-Aigner, R.L. Mach, G. Ramer, B. Lendl:
"Near-field IR detection enables nanoscale computational staining";
Vortrag: SciX 2020, online (eingeladen); 12.10.2020 - 15.10.2020.

Near-field IR detection enables nanoscale computational staining

The photothermal induced resonance technique (PTIR, also called AFM-IR) coupled with broadly tunable mid-IR laser light sources gives access to nanoscale resolved chemical information via well established spectra-structure correlations of mid-IR spectroscopy. By using an atomic force tip for near field detection, this method can provide spatial resolution on the order of 20 nm laterally. This ultra-high spatial resolution has led to an ever increasing range of applications of PTIR, ranging from archeology, over material science (2D materials, material interfaces, polymer science) to biology and biomedical applications. As applications of PTIR and with them the spectra that are recorded grow ever more complex, tools are required to aid spectroscopists in extracting information from them.

Such methods - unsupervised and supervised chemometric methods - are well established for far-field imaging but have seen only tentative use in near-field imaging. To the best of our knowledge, supervised regression methods have not been applied to PTIR data before, mainly due to the challenge of gathering reference data as required for supervised methods.

Here, we describe our approach to the challenging task of establishing a multivariate calibration for quantitation of a specific group of proteins inside an indivual microorganism. This allows the determination of protein abundance (i.e. computational staining) at the spatial resolution of PTIR via hyperspectral nanoscale imaging.

PTIR; infrared; nano; nanoscale; fungi; machine learning