Technical note: fourier transform infrared (FTIR) spectroscopy as an optical nose for predicting odor sensation

Quantifying odor is important for objectively assessing the impact of animal production systems on surrounding areas. A possible method that has received little attention is Fourier transform (mid) infrared spectroscopy (FTIR). Gases that contribute to odor have unique infrared spectra, and the advantage of FTIR over electronic nose technology or gas chromatography is that theoretically all these gases can be analyzed instantaneously. To determine the feasibility of FTIR for predicting odor, 71 air samples analyzed by olfactometry were scanned in a spectrometer using an 84-m path-length gas cell. Scans were obtained over a period of about 1 min and from 4,000 to 740 cm(-1) with a resolution of 0.5 cm(-1). Calibrations for predicting odor were developed using partial least squares regression with full cross-validation. Air samples were obtained from experiments with pigs fed diets formulated to alter odor emission or from stored manure. Odor threshold dilution ratios averaged 676+/-491 units, with a range from 120 to 2,161. Using these samples, a prediction error for odor sensation of 344 units (R2 = 0.51) was obtained. Log transformation of the odor data improved the R2 to 0.61. Based on the olfactometry data, it is estimated that the measurement error of olfactometry is 250 units, which limits the R2 of any method to approximately 0.74. Thus, this calibration is very encouraging. In conclusion, FTIR shows promise as a practical means for objectively assessing swine odor.