Biosensors for Odor Detection: A Review.

Animals can easily detect hundreds of thousands of odors in the environment with high sensitivity and selectivity. With the progress of biological olfactory research, scientists have extracted multiple biomaterials and integrated them with different transducers thus generating numerous biosensors. Those biosensors inherit the sensing ability of living organisms and present excellent detection performance. In this paper, we mainly introduce odor biosensors based on substances from animal olfactory systems. Several instances of organ/tissue-based, cell-based, and protein-based biosensors are described and compared. Furthermore, we list some other biological materials such as peptide, nanovesicle, enzyme, and aptamer that are also utilized in odor biosensors. In addition, we illustrate the further developments of odor biosensors.

Active Tracking of Temporally Changing Gas-Phase Odor Mixture Using an Array of Cells Expressing Olfactory Receptors.

A cell expressing an olfactory receptor (OR) exhibits excellent odorant detection ability and thus is widely applied in odor biosensors. Most of those biosensors, however, could detect only liquid-phase nonchanging single-component odorants. In this paper, we raised up an odor biosensor for the active tracking of temporally changing gas-phase odor mixture by an array of cells expressing ORs. A thin stable liquid film covered the cell, thus allowing gas-phase odorants to penetrate. The online image processing generated individual cell brightness data which were used to compute the biosensor response. Based on the obtained responses, we adjusted the known odor components to be similar with the unknown odor. The function of our biosensor was validated by tracking the variable single-component odorant or the binary odor mixture. The influence from the sensor drift could be overcome by comparing the adjacent unknown and known odor responses. In the odor mixture quantification, adding the OR label to mixed cells and then quantifying separately (named as the pre-label method) was more efficient, while directly using the cell response pattern (named as the label-free method) was still capable even if the OR odor had cross-sensitivity.

Array of Miniaturized Amperometric Gas Sensors Using Atomic Gold Decorated Pt/PANI Electrodes in Room Temperature Ionic Liquid Films.

Miniaturized sensors possess many advantages, such as rapid response, easy chip integration, a possible lower concentration of target compound detection, etc. However, a major issue reported is a low signal response. In this study, a catalyst, the atomic gold clusters of Aun where n = 2, was decorated at a platinum/polyaniline (Pt/PANI) working electrode to enhance the sensitivity of butanol isomers gas measurement. Isomer quantification is challenging because this compound has the same chemical formula and molar mass. Furthermore, to create a tiny sensor, a microliter of room-temperature ionic liquid was used as an electrolyte. The combination of the Au2 clusters decorated Pt/PANI and room temperature ionic liquid with several fixed electrochemical potentials was explored to obtain a high solubility of each analyte. According to the results, the presence of Au2 clusters increased the current density due to electrocatalytic activity compared to the electrode without Au2 clusters. In addition, the Au2 clusters on the modified electrode had a more linear concentration dependency trend than the modified electrode without atomic gold clusters. Finally, the separation among butanol isomers was enhanced using different combination of room-temperature ionic liquids and fixed potentials.

Extraction of sensing data for desired scent impressions using mass spectra of odorant molecules.

Most of the olfactory perception works focused on forward prediction of odor impression, for example, given an odorant's molecular structure parameters or the sensing data predict its odor impression. So far, mapping of mass spectrum of odorant molecules into the odor perception space (binary or continuous sensory space) has been successfully performed. However, it is difficult to predict odorant's sensing data associated with binary odor descriptors (e.g., minty, peach, vanilla etc.). In this study, we have proposed a method to extract the corresponding sensing data (mass spectrum as sensing data) for a desired scent impression although one-to-one relationships are not usually guaranteed. Our target is to extract the sensing data for a given odor descriptor that will help perfumers to create scent. This study is first report for predicting sensing data for a given binary odor descriptor.

Exploration of sensing data to realize intended odor impression using mass spectrum of odor mixture.

Recently, olfactory information on odorants has been associated with their corresponding molecular features. Such information has been obtained by predicting the sensory test evaluation scores from the molecular structure parameters or the sensing data. On the other hand, we develop a method of the prediction of molecular features corresponding to the odor impression. We utilize a machine-learning-based odor predictive model introduced in our previous research, and we propose a mathematical model for exploring the sensing data space. By using mass spectrum as sensing data in the predictive model, we can represent predicted mass spectrum as those of an odor mixture, and the mixing ratio can be obtained. We show that the mass spectrum of apple flavor with enhanced 'fruit' and 'sweet' impressions can be obtained using 59 and 60 molecules respectively by using our analysis method.

Predicting individual perceptual scent impression from imbalanced dataset using mass spectrum of odorant molecules.

Predicting odor impression is considered an important step towards measuring the quality of scent in the food, perfume, and cosmetic industries. In odor impression identification and classification, the main target is to predict scent impression while identifying non-target odor impressions are less significant. However, the effectiveness of predictive models depends on the quality of data distribution. Since it is difficult to collect large scale sensory data to create an evenly distributed positive (target odor) and negative (non-target odor) samples, a method is necessary to predict the individual characteristics of scent according to the number of positive samples. Moreover, it is required to predict large number of individual odor impressions from such kind of imbalanced dataset. In this study, we used mass spectrum of flavor molecules and their corresponding odor impressions which have a very disproportioned ratio of positive and negative samples. Thus, we used One-class Classification Support Vector Machine (OCSVM) and Cost-Sensitive MLP (CSMLP) to precisely classify target scent impression. Our experimental results show satisfactory performance in terms of AUCROC to detect the olfactory impressions of 89 odor descriptors from the mass spectra of flavor molecules.

Extending lifetime of gas-phase odor biosensor using liquid thickness control and liquid exchange.

In recent few years, researchers utilized cell expressing olfactory receptor for vapor detection under various sensing mechanisms. Those olfactory systems, however, have relatively short lifetime due to the dry out of aqueous solution covering the cell. In this paper, we came up with a feedback control structure composed of an impedance measurement circuit, a microcontroller and two syringe pumps for maintaining thin liquid layer above cell. Cell lifetime was improved from less than 40 min to longer than 75 min when liquid film control was introduced. However, the biosensor lifetime remained similar between with or without liquid thickness control. Then, we added liquid exchange to further extend the lifetime of our odor biosensor. Minimal liquid exchange speed was able to significantly extend the biosensor lifetime. Meanwhile, faster liquid exchange speed resulted in better sensor responses. Furthermore, the enhancement acquired from intermittent liquid exchange was compared with continuous one. In this study, the lifetime of odor biosensor was extended to more than 3 h whereas it was less than half an hour without liquid thickness control. We believe the methodology we established in this paper will facilitate gas phase odor biosensor in continuous monitoring of target substances.

Binary mixture quantification using cell-based odor biosensor system with active sensing.

Organisms perceive odorants in the environment through the use of a large number of olfactory receptors. Various odor biosensors have been researched and developed in order to mimic this olfactory mechanism. This study examines the quantification of odorant concentrations through the use of a sensor array comprised of several types of cell-based odor sensors expressing insect olfactory receptors with nonlinear characteristics. The sensor system utilized an active sensing method in order to compare the responses of a target odorant and a prepared odorant in determining the relative concentration of the target odorant. By combining an active sensing method with a real-time reference method in which the target odorant was measured every time the prepared odorant was measured, the relative concentrations were successfully determined even when the response fluctuation was large or odorant sensor cell responses varied as measurement time increased. For proof of concept purposes, the study primarily focused on quantifying odorant concentrations composed of one or two odorant components. It was confirmed that an algorithm to find the optimal relative odorant concentration among a limited number of odorant concentrations is achievable. Though this study is still in the initial stage of the developing odor sensors and has many challenges, it can provide insight into paving the way towards a new type of odor biosensor with active sensing.

Study of Room Temperature Ionic Liquids as Gas Sensing Materials in Quartz Crystal Microbalances.

Twenty-eight quartz crystal microbalance (QCM) sensors coated with different sensing films were tested and analyzed in this work; twenty-three sensors were coated in different room temperature ionic liquids (RTILs) and five additional QCM sensors were coated with conventional films commonly used as stationary phases in gas chromatography. Four volatile organic compounds (VOCs), in gaseous phase-hexanol, butyl acetate, 2-hexanone, and hexanoic acid-were measured. Two transducer mechanisms were used; resonant frequency shift and resistance shift of a QCM Mason equivalent circuit. The sensors were characterized by their sensitivity to the VOCs and their discrimination power of the four VOCs. The highest separation among VOCs was obtained when frequency and resistance information of both RTIL and conventional films was used, a sensor array composed by two RTILs (1-butyl-1-methylpyrrolidinium bis(trifluoromethanesulfonyl)imide and 1-hexyl-3-methylimidazolium hexafluorophosphate) and two conventional films (tricresyl phosphate and apiezon-L) was found to improve the Wilks lambda separation for the tested gases two orders of magnitude compared to the Wilks lambda using only a conventional films array.

Indirect Sensing of Lower Aliphatic Ester Using Atomic Gold Decorated Polyaniline Electrode.

Novel sensing materials have been formed by decorating polyaniline conducting polymers with atomic gold clusters where the number of atoms is precisely defined. Such materials exhibit unique electrocatalytic properties of electrooxidation to aliphatic alcohols, although analytes with other functional groups have not been studied. This paper reports a study of cyclic voltammetric patterns obtained with bi-atomic gold nanocomposite response to analytes with other functional groups for sensor applications. Principal component analysis shows separation among normal-propanol, iso-propanol and ethyl formate/ethanol groups. Indirect sensing of ethyl formate is demonstrated by electrooxidation of the product upon hydrolysis in alkaline medium. Voltammograms of ethyl formate are studied in gaseous phases.

Predicting human odor perception represented by continuous values from mass spectra of essential oils resembling chemical mixtures.

There have been recent advances in predicting odor characteristics using molecular structure parameters of chemicals. Although the molecular structure parameters are available for each chemical, they cannot be used for chemical mixtures. This study will elucidate a computational method of predicting human odor perception from the mass spectra of chemical mixtures such as essential oils. Furthermore, a method for obtaining similarity among odor descriptors has been proposed although the dataset contains binary values only. When the database indicates a set of odor descriptors for one sample, only binary data are available and the correlation between the similar descriptors disappears. Thus, the prediction performance degrades for not considering the similarity among the odor descriptors. Since mass spectra dataset is highly dimensional, we use auto-encoder to learn the compressed representation from the mass spectra of essential oils in its bottleneck hidden layer and then accomplishes the hierarchical clustering to create odor descriptor groups with similar odor impressions using a matrix of continuous value-based correlation coefficient as well as natural language processing. This work will help to expatiate the process of overcoming binary value problem and find out the similarity among odor descriptors using machine learning with natural language semantic representation of words. To overcome the problem of disproportionate ratio of positive and negative class for both the continuous value-based correlation coefficient and word similarity based models, we use Synthetic Minority Oversampling Technique (SMOTE). This model allows us to predict human odor perception through computer simulations by forming odor descriptors group. Accordingly, this study demonstrates the feasibility of ensembling machine learning with natural language processing and SMOTE approach for predicting odor descriptor group from mass spectra of essential oils.

On-Line Mixture Quantification to Track Temporal Change of Composition Using FAIMS.

This paper reports on-line mixture quantification with FAIMS. Ternary gas mixtures composed of acetone, ethanol, and diethyl ether were used for quantification. We succeeded in an on-line quantification of ppm-level concentration and even sub-ppm-level gases using the gradient descent method. It took 10 minutes to quantify the ternary mixture. However, it was too long, because we aim to track the temporal change of each component concentration in the mixture. Then, an algorithm based on feedback control was introduced to reduce the quantification time. The feedback method successfully tracked concentrations in three cases. The simulation result shows that the proposed method can reduce the quantification time.

Optimization of Modulation Methods for Solenoid Valves to Realize an Odor Generation System.

An artificial olfactory system coupled with an odor generation system is herein reported. The artificial olfactory system is composed of four chemical sensors consisting of quartz crystal microbalances (QCMs) coated with room temperature ionic liquids (RTILs). The sensors are interrogated by four vector network analyzers, which are used to measure the series resonant frequency and motional resistance. The odor generation system can generate eight different odors and mix them in any composition. Solenoid valves are used to switch the path and control the concentration of the different odors before blending. Two algorithms to control the solenoid valves, delta-sigma modulator, and simple pulse width modulation (PWM) are studied, optimized, and compared. Finally, the uncertainty of the odor generating system is calculated.

An Olfactory Sensor Array for Predicting Chemical Odor Characteristics from Mass Spectra with Deep Learning.

Machine learning techniques are useful for applications such as electronic nose (e-nose) systems to classify or identify the target odor. In recent years, deep learning is regarded as one of the most powerful machine learning methods. It enables researchers to extract useful features automatically from high-dimensional raw data and has been widely applied to computer vision, speech recognition, and natural language processing, though little has been reported in the field of olfaction. In this chapter, we describe the procedure to build a deep neural network to predict odor characteristics of chemicals from their mass spectra.

Ternary Gas Mixture Quantification Using Field Asymmetric Ion Mobility Spectrometry (FAIMS).

Gas mixture quantification is essential for the recording and reproducing odors, because an odor consists of multiple chemical compounds. Gas mixture quantification using field asymmetric ion mobility spectrometry (FAIMS) was studied. Acetone, ethanol, and diethyl ether were selected as components of a ternary gas mixture sample as representatives of the ketone, alcohol, and ether chemical classes, respectively. One hundred and twenty-five points with different concentrations were measured. The results were evaluated by error hypersurface, variance, and the coefficient of variation. The error hypersurface showed that it is possible to reach the target composition by following the error-hypersurface gradient. Successful convergence was achieved with the gradient descent method in a simulation based on the measurement data. This result verified the feasibility of the quantification of a gas mixture using FAIMS.

Enhancement Method of Surface Acoustic Wave-Atomizer Efficiency for Olfactory Display.

Since olfaction is an important sense in human interfaces, we have developed an olfactory display using a surface acoustic wave (SAW) atomizer and micro-dispensers. In this olfactory display, the efficiency of atomization is important in order to avoid smell persistence problems often encountered in human olfactory interfaces. Thus, the SAW device is coated with amorphous Teflon film to change the substrate nature from hydrophilic to hydrophobic. It is also necessary to silanize the piezoelectric substrate surface prior to Teflon coating to enhance the adhesion of the film. A dip coating method was adopted to obtain uniform coating on the substrate. The high-speed solenoid valve was used as micro-dispenser to spout a liquid droplet to the SAW device surface since its accuracy and reproducibility were high. Then, the atomization became easier on the hydrophobic substrate. In this study, the amorphous Teflon coating for minimizing the remaining liquid on the substrate after atomization was studied. The goal of the protocol described here is to show the methods for coating a SAW device surface with amorphous Teflon film and generating the smell using the SAW atomizer and a micro-dispenser, followed by a sensory test.

Correction: Predictive modeling for odor character of a chemical using machine learning combined with natural language processing.

[This corrects the article DOI: 10.1371/journal.pone.0198475.].

Predictive modeling for odor character of a chemical using machine learning combined with natural language processing.

Recent studies on machine learning technology have reported successful performances in some visual and auditory recognition tasks, while little has been reported in the field of olfaction. In this paper we report computational methods to predict the odor impression of a chemical from its physicochemical properties. Our predictive model utilizes nonlinear dimensionality reduction on mass spectra data and performs the clustering of descriptors by natural language processing. Sensory evaluation is widely used to measure human impressions to smell or taste by using verbal descriptors, such as "spicy" and "sweet". However, as it requires significant amounts of time and human resources, a large-scale sensory evaluation test is difficult to perform. Our model successfully predicts a group of descriptors for a target chemical through a series of computer simulations. Although the training text data used in the language modeling is not specialized for olfaction, the experimental results show that our method is useful for analyzing sensory datasets. This is the first report to combine machine olfaction with natural language processing for odor character prediction.

Odor Impression Prediction from Mass Spectra.

The sense of smell arises from the perception of odors from chemicals. However, the relationship between the impression of odor and the numerous physicochemical parameters has yet to be understood owing to its complexity. As such, there is no established general method for predicting the impression of odor of a chemical only from its physicochemical properties. In this study, we designed a novel predictive model based on an artificial neural network with a deep structure for predicting odor impression utilizing the mass spectra of chemicals, and we conducted a series of computational analyses to evaluate its performance. Feature vectors extracted from the original high-dimensional space using two autoencoders equipped with both input and output layers in the model are used to build a mapping function from the feature space of mass spectra to the feature space of sensory data. The results of predictions obtained by the proposed new method have notable accuracy (R≅0.76) in comparison with a conventional method (R≅0.61).

[Odor sensing system and olfactory display].

In this review, an odor sensing system and an olfactory display are introduced into people in pharmacy. An odor sensing system consists of an array of sensors with partially overlapping specificities and pattern recognition technique. One of examples of odor sensing systems is a halitosis sensor which quantifies the mixture composition of three volatile sulfide compounds. A halitosis sensor was realized using a preconcentrator to raise sensitivity and an electrochemical sensor array to suppress the influence of humidity. Partial least squares (PLS) method was used to quantify the mixture composition. The experiment reveals that the sufficient accuracy was obtained. Moreover, the olfactory display, which present scents to human noses, is explained. A multi-component olfactory display enables the presentation of a variety of smells. The two types of multi-component olfactory display are described. The first one uses many solenoid valves with high speed switching. The valve ON frequency determines the concentration of the corresponding odor component. The latter one consists of miniaturized liquid pumps and a surface acoustic wave (SAW) atomizer. It enables the wearable olfactory display without smell persistence. Finally, the application of the olfactory display is demonstrated. Virtual ice cream shop with scents was made as a content of interactive art. People can enjoy harmony among vision, audition and olfaction. In conclusion, both odor sensing system and olfactory display can contribute to the field of human health care.