Electronic noses combine an array of cross-selective gas sensors with a pattern recognition engine to identify odors. Pattern recognition of multivariate gas sensor response is usually performed using existing statistical and chemometric techniques. An alternative solution involves developing novel algorithms inspired by information processing in the biological olfactory system. The objective of this dissertation is to develop a neuromorphic architecture for pattern recognition for a chemosensor array inspired by key signal processing mechanisms in the olfactory system. Our approach can be summarized as follows. First, a high-dimensional odor signal is generated from a chemical sensor array. Three approaches have been proposed to generate this combinatorial and high dimensional odor signal: temperature-modulation of a metal-oxide chemoresistor, a large population of optical microbead sensors, and infrared spectroscopy. The resulting high-dimensional odor signals are subject to dimensionality reduction using a self-organizing model of chemotopic convergence. This convergence transforms the initial combinatorial high-dimensional code into an organized spatial pattern (i.e., an odor image), which decouples odor identity from intensity. Two lateral inhibitory circuits subsequently process the highly overlapping odor images obtained after convergence. The first shunting lateral inhibition circuits perform gain control enabling identification of the odorant across a wide range of concentration. This shunting lateral inhibition is followed by an additive lateral inhibition circuit with center-surround connections. These circuits improve contrast between odor images leading to more sparse and orthogonal patterns than the one available at the input. The sharpened odor image is stored in a neurodynamic model of a cortex. Finally, anti-Hebbian/ Hebbian inhibitory feedback from the cortical circuits to the contrast enhancement circuits performs mixture segmentation and weaker odor/background suppression, respectively. We validate the models using experimental datasets and show our results are consistent with recent neurobiological findings.

Electronic noses combine an array of cross-selective gas sensors with a pattern recognition engine to identify odors. Pattern recognition of multivariate gas sensor response is usually performed using existing statistical and chemometric techniques. An alternative solution involves developing novel algorithms inspired by information processing in the biological olfactory system. The objective of this dissertation is to develop a neuromorphic architecture for pattern recognition for a chemosensor array inspired by key signal processing mechanisms in the olfactory system. Our approach can be summarized as follows. First, a high-dimensional odor signal is generated from a chemical sensor array. Three approaches have been proposed to generate this combinatorial and high dimensional odor signal: temperature-modulation of a metal-oxide chemoresistor, a large population of optical microbead sensors, and infrared spectroscopy. The resulting high-dimensional odor signals are subject to dimensionality reduction using a self-organizing model of chemotopic convergence. This convergence transforms the initial combinatorial high-dimensional code into an organized spatial pattern (i.e., an odor image), which decouples odor identity from intensity. Two lateral inhibitory circuits subsequently process the highly overlapping odor images obtained after convergence. The first shunting lateral inhibition circuits perform gain control enabling identification of the odorant across a wide range of concentration. This shunting lateral inhibition is followed by an additive lateral inhibition circuit with center-surround connections. These circuits improve contrast between odor images leading to more sparse and orthogonal patterns than the one available at the input. The sharpened odor image is stored in a neurodynamic model of a cortex. Finally, anti-Hebbian/ Hebbian inhibitory feedback from the cortical circuits to the contrast enhancement circuits performs mixture segmentation and weaker odor/background suppression, respectively. We validate the models using experimental datasets and show our results are consistent with recent neurobiological findings.

This thesis addresses the problem of detecting changes in the activity of a distant gas source from the response of an array of metal oxide (MOX) gas sensors deployed in an Open Sampling System (OSS). Changes can occur due to gas source activity such as a sudden alteration in concentration or due to exposure to a different compound. Applications such as gas-leak detection in mines or large-scale pollution monitoring can benefit from reliable change detection algorithms, especially where it is impractical to continuously store or transfer sensor readings, or where reliable calibration is difficult to achieve. Here, it is desirable to detect a change point indicating a significant event, e.g. presence of gas or a sudden change in concentration. The main challenges are turbulent dispersion of gas and the slow response and recovery times of MOX sensors. Due to these challenges, the gas sensor response exhibits fluctuations that interfere with the changes of interest. The contributions of this thesis are centred on developing change detection methods using MOX sensor responses. First, we apply the Generalized Likelihood Ratio algorithm (GLR), a commonly used method that does not make any a priori assumption about change events. Next, we propose TREFEX, a novel change point detection algorithm, which models the response of MOX sensors as a piecewise exponential signal and considers the junctions between consecutive exponentials as change points. We also propose the rTREFEX algorithm as an extension of TREFEX. The core idea behind rTREFEX is an attempt to improve the fitted exponentials of TREFEX by minimizing the number of exponentials even further. GLR, TREFEX and rTREFEX are evaluated for various MOX sensors and gas emission profiles. A sensor selection algorithm is then introduced and the change detection algorithms are evaluated with the selected sensor subsets. A comparison between the three proposed algorithms shows clearly superior performance of rTREFEX both in detection performance and in estimating the change time. Further, rTREFEX is evaluated in real-world experiments where data is gathered by a mobile robot. Finally, a gas dispersion simulation was developed which integrates OpenFOAM flow simulation and a filament-based gas propagation model to simulate gas dispersion for compressible flows with a realistic turbulence model.

<p> Volatile organic compounds (VOCs) are byproducts of metabolic processes that can be uniquely dysregulated by various medical conditions and are expressed in biological samples. Therefore, VOCs expressed in breath, urine and other sample types may be utilized for noninvasive, rapid, and accurate diagnostics in a point-of-care setting. Currently, the most common methods for VOC detection include gas chromatography-mass spectrometry (GC-MS) and electronic noses (E-noses) that integrate nanosensors. Both methods present important advantages and challenges that allow their implementation for different applications. While GC-MS can be used to directly identify VOCs in complex matrices, it is a non-portable and high-cost instrument. On the other hand, E-noses are portable and user-friendly VOC detectors, but they do not allow for direct VOC identification or quantification. Among different VOC rich sample types, breath offers the advantage of being a virtually limitless source of endogenous biomarkers that can be implemented for noninvasive VOC detection.</p> <p><br></p> <p>The presented thesis focuses on the optimization of the operating parameters (heater and sensor voltages) of a metal oxide (MOX) sensor and breath sampling techniques (sensor casing, breath fractionation, and exhalation volume) for their implementation in exhaled VOC analysis. In parallel, an in-house feature extraction algorithm was developed and implemented for the optimization of a MOX sensor composed of a tin oxide (SnO2) sensing layer. The optimized sensor parameters (heater voltage equal to 2 V and sensor voltage equal to 0.8 V) and breath sampling protocol (24 L of whole breath analyzed using the in-house sensor casing design) were tested with exhaled breath samples from distinct volunteers which could be successfully separated with 100% accuracy. The sensor response also showed a high degree of intrasubject reproducibility (RSD < 6%). Additionally, the sensor performance was further validated under ambient conditions, and sensor degradation was studied over the course of 3 months. Finally, sensor response to synthetic VOC profiles and individual VOC standards was explored. Optimized SnO2 sensors distinguished between VOC mixtures regardless of variations in relative humidity (RH) levels. Furthermore, the characteristic sensor response to VOC standards indicates that the sensors are most sensitive toward isopropanol by a factor of 1.15 in 45% RH and a factor of 3.58 in 85% RH relative to isoprene. </p> <p><br></p> <p>To translate the potential of MOX sensors to point-of-care biomedical applications, there first exists the need to establish a reference of sensor baseline signals corresponding to exhaled breath samples from healthy individuals. SnO2 sensors and breath sampling methods were implemented for the collection of individual samples from 109 relatively healthy volunteers. 10 of these volunteers provided 9 additional samples over the course of six months. In parallel, exhaled breath samples were also analyzed by GC-MS to comprehensively profile VOCs present in the samples. The results from these experiments not only aid in the identification of the healthy breath signal baseline but also allow the exploration of VOC reproducibility over time. High variation between samples from distinct volunteers was observed, but samples longitudinally collected across volunteers could not be distinguished, alluding to the existence of a universal range of sensor signals that could describe the composition of exhaled breath from healthy subjects. Finally, results were compared with relevant confounding variables to better understand how VOCs are impacted by an array of factors that are not directly correlated to disease diagnosis. Sensor signals were significantly elevated in breath samples from male volunteers compared to samples from female subjects (p-value = 0.044). Interestingly, isoprene signals resulting from the GC-MS analysis were also higher in male subjects relative to females. No other relationships were identified between sensor signals and the confounding variables of interest. </p> <p><br></p> <p>Future work would require a deeper understanding of sensor degradation and life cycle, along with sensor testing using a broader range of individual VOC standards and more complex VOC profiles. Additionally, further comparison between sensor signal and GC-MS signal of relevant VOC biomarkers present in breath would be beneficial. Nonetheless, the presented be leveraged in future investigations aiming to identify biomarkers for different medical conditions. Finally, the findings disclosed in the deposited thesis suggest the ability of a SnO2 nanosensor array to be implemented for breath analysis, providing a noninvasive, easy to use, and reliable diagnostic device in a point-of-care setting. </p>