With 1.6 million deaths per year, lung cancer is one of the leading causes of death worldwide. One reason for this high number is the absence of a preventive medical examination method. Many diagnoses occur in a late cancer stage with a low survival rate. An early detection could significantly decrease the mortality. In recent decades, certain substances in human breath have been linked to certain diseases. Different studies show that it is possible to distinguish between lung cancer patients and a healthy control group by analyzing the volatile organic compounds (VOCs) in their breath. We developed a sensor based on photoacoustic spectroscopy for six of the most relevant VOCs linked to lung cancer. As a radiation source, the sensor uses an optical-parametric oscillator (OPO) in a wavelength region from 3.2 µm to 3.5 µm. The limits of detection for a single substance range between 5 ppb and 142 ppb. We also measured high resolution absorption spectra of the biomarkers compared to the data currently available from the National Institute of Standards and Technology (NIST) database, which is the basis of any selective spectroscopic detection. Future lung cancer screening devices could be based on the further development of this sensor.

We applied a multivariate analysis (MVA) to spectroscopic data of gas mixtures in the mid-IR in order to calculate the concentrations of the single components which exhibit strongly overlapping absorption spectra. This is a common challenge in broadband spectroscopy. Photoacoustic (PA) measurements of different volatile organic compounds (VOCs) in the wavelength region of 3250 nm to 3550 nm served as the exemplary detection technique. Partial least squares regression (PLS) was used to calculate concentrations from the PA spectra. After calibration, the PLS model was able to determine concentrations of single VOCs with a relative accuracy of 2.60%.

We evaluate the spectral resolution and the detection thresholds achievable for a photoacoustic spectroscopy (PAS) system in the broadband infrared wavelength region $3270\mathrm{n}\mathrm{m}\lesssim \lambda \lesssim 3530\mathrm{n}\mathrm{m}$ driven by a continuous wave optical parametric oscillator (OPO) with $\overline{P}\approx 1.26\mathrm{W}$ . The absorption spectra, $I_{\mathrm{PAS}}({\lambda }_{\mathrm{i}})$ , for diluted propane, ethane and methane test gases at low concentrations ( $c\sim 100\mathrm{ppm}$ ) were measured for ∼1350 discrete wavelengths ${\lambda }_{\mathrm{i}}$ . The $I_{\mathrm{PAS}}({\lambda }_{\mathrm{i}})$ spectra were then compared to the high resolution cross section data, ${\sigma }_{\mathrm{FTIR}}$ , obtained by Fourier Transform Infrared Spectroscopy published in the HITRAN database. Deviations of 7.1(6)% for propane, 8.7(11)% for ethane and 15.0(14)% for methane with regard to the average uncertainty between $I_{\mathrm{PAS}}({\lambda }_{\mathrm{i}})$ and the expected reference values based on ${\sigma }_{\mathrm{FTIR}}$ were recorded. The characteristic absorption wavelengths ${\lambda }_{\mathrm{res}}$ can be resolved with an average resolution of $\delta {\lambda }_{\mathrm{res}}\sim 0.08\mathrm{nm}$ . Detection limits range between $7.1$ ppb (ethane) to $13.6$ ppb (methane). In an additional step, EUREQA, an artificial intelligence (AI) program, was successfully applied to deconvolute simulated PAS spectra of mixed gas samples at low limits of detection. The results justify a further development of PAS technology to support e.g., biomedical research.

Photoacoustic spectroscopy allows the identification of specific molecules in gases. We evaluate the spectral resolution and detection limits for a PAS system in the broadband infrared wavelength region 3270 nm ≲ λ ≲ 3530 nm driven by a continuous wave optical parametric oscillator with P ≈ 1.26 W by measuring the absorption of diluted propane, ethane and methane test gases at low concentrations c ~ 100 ppm for ~1350 discrete wavelengths λ_i. The resulting spectra I_PAS(λ_i) were compared to the high resolution cross section data σ_FTIR obtained by Fourier Transform Infrared Spectroscopy from the HITRAN database. Deviations as little as 7.1(6)% for propane, 8.7(11)% for ethane and 15.0(14)% for methane with regard to the average uncertainty between I_PAS(λ_i) and the expected reference values based on σ_FTIR were recorded. The wavelengths λ_res of the characteristic absorption lines can be pinpointed with a high relative accuracy <5 × 10⁻⁵ corresponding to a resolution of λ_res ~ 0.16 nm. Detection limits range between 7.1 ppb (ethane) to 13.6 ppb (methane) coinciding with high experimental signal-to-noise ratios. Moreover, using EUREQA, an artificial intelligence program, simulated mixed gas samples at low limits of detection could be deconvoluted. These results justify a further development of PAS technology to support, e.g., biomedical research.

Multivariate analysis (MVA) was introduced in the early 20th century and first applied in psychology under the name of Multiple Factor Analysis (MFA).