Anomaly Detection in Noisy Images

Finding rare events in multidimensional data is an important detection problem that has applications in many fields, such as risk estimation in insurance industry, finance, flood prediction, medical diagnosis, quality assurance, security, or safety in transportation. The occurrence of such anomalies is so infrequent that there is usually not enough training data to learn an accurate statistical model of the anomaly class. In some cases, such events may have never been observed, so the only information that is available is a set of normal samples and an assumed pairwise similarity function. Such metric may only be known up to a certain number of unspecified parameters, which would either need to be learned from training data, or fixed by a domain expert. Sometimes, the anomalous condition may be formulated algebraically, such as a measure exceeding a predefined threshold, but nuisance variables may complicate the estimation of such a measure. Change detection methods used in time series analysis are not easily extendable to the multidimensional case, where discontinuities are not localized to a single point. On the other hand, in higher dimensions, data exhibits more complex interdependencies, and there is redundancy that could be exploited to adaptively model the normal data. In the first part of this dissertation, we review the theoretical framework for anomaly detection in images and previous anomaly detection work done in the context of crack detection and detection of anomalous components in railway tracks. In the second part, we propose new anomaly detection algorithms. The fact that curvilinear discontinuities in images are sparse with respect to the frame of shearlets, allows us to pose this anomaly detection problem as basis pursuit optimization. Therefore, we pose the problem of detecting curvilinear anomalies in noisy textured images as a blind source separation problem under sparsity constraints, and propose an iterative shrinkage algorithm to solve it. Taking advantage of the parallel nature of this algorithm, we describe how this method can be accelerated using graphical processing units (GPU). Then, we propose a new method for finding defective components on railway tracks using cameras mounted on a train. We describe how to extract features and use a combination of classifiers to solve this problem. Then, we scale anomaly detection to bigger datasets with complex interdependencies. We show that the anomaly detection problem naturally fits in the multitask learning framework. The first task consists of learning a compact representation of the good samples, while the second task consists of learning the anomaly detector. Using deep convolutional neural networks, we show that it is possible to train a deep model with a limited number of anomalous examples. In sequential detection problems, the presence of time-variant nuisance parameters affect the detection performance. In the last part of this dissertation, we present a method for adaptively estimating the threshold of sequential detectors using Extreme Value Theory on a Bayesian framework. Finally, conclusions on the results obtained are provided, followed by a discussion of possible future work.

Lesion detection performance: comparative analysis of low-dose CT data of the chest on two hybrid imaging systems

Incidental findings on low-dose CT images obtained during hybrid imaging are an increasing phenomenon as CT technology advances. Understanding the diagnostic value of incidental findings along with the technical limitations is important when reporting image results and recommending follow-up, which may result in an additional radiation dose from further diagnostic imaging and an increase in patient anxiety. This study assessed lesions incidentally detected on CT images acquired for attenuation correction on two SPECT/CT systems. Methods: An anthropomorphic chest phantom containing simulated lesions of varying size and density was imaged on an Infinia Hawkeye 4 and a Symbia T6 using the low-dose CT settings applied for attenuation correction acquisitions in myocardial perfusion imaging. Twenty-two interpreters assessed 46 images from each SPECT/CT system (15 normal images and 31 abnormal images; 41 lesions). Data were evaluated using a jackknife alternative free-response receiver-operating-characteristic analysis (JAFROC). Results: JAFROC analysis showed a significant difference (P < 0.0001) in lesion detection, with the figures of merit being 0.599 (95% confidence interval, 0.568, 0.631) and 0.810 (95% confidence interval, 0.781, 0.839) for the Infinia Hawkeye 4 and Symbia T6, respectively. Lesion detection on the Infinia Hawkeye 4 was generally limited to larger, higher-density lesions. The Symbia T6 allowed improved detection rates for midsized lesions and some lower-density lesions. However, interpreters struggled to detect small (5 mm) lesions on both image sets, irrespective of density. Conclusion: Lesion detection is more reliable on low-dose CT images from the Symbia T6 than from the Infinia Hawkeye 4. This phantom-based study gives an indication of potential lesion detection in the clinical context as shown by two commonly used SPECT/CT systems, which may assist the clinician in determining whether further diagnostic imaging is justified.

Second-order cyclostationarity-based detection and classification of LTE SC-FDMA signals for cognitive radio

Cognitive radio (CR) was developed for utilizing the spectrum bands efficiently. Spectrum sensing and awareness represent main tasks of a CR, providing the possibility of exploiting the unused bands. In this thesis, we investigate the detection and classification of Long Term Evolution (LTE) single carrier-frequency division multiple access (SC-FDMA) signals, which are used in uplink LTE, with applications to cognitive radio. We explore the second-order cyclostationarity of the LTE SC-FDMA signals, and apply results obtained for the cyclic autocorrelation function to signal detection and classification (in other words, to spectrum sensing and awareness). The proposed detection and classification algorithms provide a very good performance under various channel conditions, with a short observation time and at low signal-to-noise ratios, with reduced complexity. The validity of the proposed algorithms is verified using signals generated and acquired by laboratory instrumentation, and the experimental results show a good match with computer simulation results.

Microcantilever sensors for biochemical detection and analysis

Biochemical agents, including bacteria and toxins, are potentially dangerous and responsible for a wide variety of diseases. Reliable detection and characterization of small samples is necessary in order to reduce and eliminate their harmful consequences. Microcantilever sensors offer a potential alternative to the state of the art due to their small size, fast response time, and the ability to operate in air and liquid environments. At present, there are several technology limitations that inhibit application of microcantilever to biochemical detection and analysis, including difficulties in conducting temperature-sensitive experiments, material inadequacy resulting in insufficient cell capture, and poor selectivity of multiple analytes. This work aims to address several of these issues by introducing microcantilevers having integrated thermal functionality and by introducing nanocrystalline diamond as new material for microcantilevers. Microcantilevers are designed, fabricated, characterized, and used for capture and detection of cells and bacteria. The first microcantilever type described in this work is a silicon cantilever having highly uniform in-plane temperature distribution. The goal is to have 100 μm square uniformly heated area that can be used for thermal characterization of films as well as to conduct chemical reactions with small amounts of material. Fabricated cantilevers can reach above 300C while maintaining temperature uniformity of 2−4%. This is an improvement of over one order of magnitude over currently available cantilevers. The second microcantilever type is a doped single crystal silicon cantilever having a thin coating of ultrananocrystalline diamond (UNCD). The primary application of such a device is in biological testing, where diamond acts as a stable, electrically isolated reaction surface while silicon layer provides controlled heating with minimum variations in temperature. This work shows that composite cantilevers of this kind are an effective platform for temperature-sensitive biological experiments, such as heat lysing and polymerase chain reaction. The rapid heat-transfer of Si-UNCD cantilever compromised the membrane of NIH 3T3 fibroblast and lysed the cell nucleus within 30 seconds. Bacteria cells, Listeria monocytogenes V7, were shown to be captured with biotinylated heat-shock protein on UNCD surface and 90% of all viable cells exhibit membrane porosity due to high heat in 15 seconds. Lastly, a sensor made solely from UNCD diamond is fabricated with the intention of being used to detect the presence of biological species by means of an integrated piezoresistor or through frequency change monitoring. Since UNCD diamond has not been previously used in piezoresistive applications, temperature-denpendent piezoresistive coefficients and gage factors are determined first. The doped UNCD exhibits a significant piezoresistive effect with gauge factor of 7.53±0.32 and a piezoresistive coefficient of 8.12×10^−12 Pa^−1 at room temperature. The piezoresistive properties of UNCD are constant over the temperature range of 25−200C. 300 μm long cantilevers have the highest sensitivity of 0.186 m-Ohm/Ohm per μm of cantilever end deflection, which is approximately half that of similarly sized silicon cantilevers. UNCD cantilever arrays were fabricated consisting of four sixteen-cantilever arrays of length 20–90 μm in addition to an eight-cantilever array of length 120 μm. Laser doppler vibrometry (LDV) measured the cantilever resonant frequency, which ranged as 218 kHz−5.14 MHz in air and 73 kHz−3.68 MHz in water. The quality factor of the cantilever was 47−151 in air and 18−45 in water. The ability to measure frequencies of the cantilever arrays opens the possibility for detection of individual bacteria by monitoring frequency shift after cell capture.

Detection of mesoscale thermal fronts from 4km data using smoothing techniques: Gradient-based fronts classification and basin scale application

In order to optimize frontal detection in sea surface temperature fields at 4 km resolution, a combined statistical and expert-based approach is applied to test different spatial smoothing of the data prior to the detection process. Fronts are usually detected at 1 km resolution using the histogram-based, single image edge detection (SIED) algorithm developed by Cayula and Cornillon in 1992, with a standard preliminary smoothing using a median filter and a 3 × 3 pixel kernel. Here, detections are performed in three study regions (off Morocco, the Mozambique Channel, and north-western Australia) and across the Indian Ocean basin using the combination of multiple windows (CMW) method developed by Nieto, Demarcq and McClatchie in 2012 which improves on the original Cayula and Cornillon algorithm. Detections at 4 km and 1 km of resolution are compared. Fronts are divided in two intensity classes (“weak” and “strong”) according to their thermal gradient. A preliminary smoothing is applied prior to the detection using different convolutions: three type of filters (median, average and Gaussian) combined with four kernel sizes (3 × 3, 5 × 5, 7 × 7, and 9 × 9 pixels) and three detection window sizes (16 × 16, 24 × 24 and 32 × 32 pixels) to test the effect of these smoothing combinations on reducing the background noise of the data and therefore on improving the frontal detection. The performance of the combinations on 4 km data are evaluated using two criteria: detection efficiency and front length. We find that the optimal combination of preliminary smoothing parameters in enhancing detection efficiency and preserving front length includes a median filter, a 16 × 16 pixel window size, and a 5 × 5 pixel kernel for strong fronts and a 7 × 7 pixel kernel for weak fronts. Results show an improvement in detection performance (from largest to smallest window size) of 71% for strong fronts and 120% for weak fronts. Despite the small window used (16 × 16 pixels), the length of the fronts has been preserved relative to that found with 1 km data. This optimal preliminary smoothing and the CMW detection algorithm on 4 km sea surface temperature data are then used to describe the spatial distribution of the monthly frequencies of occurrence for both strong and weak fronts across the Indian Ocean basin. In general strong fronts are observed in coastal areas whereas weak fronts, with some seasonal exceptions, are mainly located in the open ocean. This study shows that adequate noise reduction done by a preliminary smoothing of the data considerably improves the frontal detection efficiency as well as the global quality of the results. Consequently, the use of 4 km data enables frontal detections similar to 1 km data (using a standard median 3 × 3 convolution) in terms of detectability, length and location. This method, using 4 km data is easily applicable to large regions or at the global scale with far less constraints of data manipulation and processing time relative to 1 km data.

Identification and Characterization of New Small Molecule Inhibitors of Picornavirus Replication

The Picornaviridae family consists of positive-strand RNA viruses that are the causative agents of a variety of diseases in humans and animals. Few drugs targeting picornaviruses are available, making the discovery of new antivirals a high priority. Here, we identified and characterized three compounds from a library of kinase inhibitors that block replication of poliovirus, coxsackievirus B3, and encephalomyocarditis virus. The antiviral effect of these compounds is not likely related to their known cellular targets because other inhibitors targeting the same pathways did not inhibit viral replication. Using an in vitro translation-replication system, we showed that these drugs inhibit different stages of the poliovirus life cycle. A4(1) inhibited the formation of a functional replication complex, while E5(1) and E7(2) affected replication after the replication complex had formed. A4(1) demonstrated partial protection from paralysis in a murine model of poliomyelitis. Poliovirus resistant to E7(2) had a single mutation in the 3A protein. This mutation was previously found to confer resistance to enviroxime-like compounds, which target either PI4KIIIβ (major enviroxime-like compounds) or OSBP (minor enviroxime-like compounds), cellular factors involved in lipid metabolism and shown to be important for replication of diverse positive-strand RNA viruses. We classified E7(2) as a minor enviroxime-like compound, because the localization of OSBP changed in the presence of this inhibitor. Interestingly, both E7(2) and major enviroxime-like compound GW5074 interfered with the viral polyprotein processing. Multiple attempts to isolate resistant mutants in the presence of A4(1) or E5(1) were unsuccessful, showing that effective broad-spectrum antivirals could be developed on the basis of these compounds. Studies with these compounds shed light on pathways shared by diverse picornaviruses that could be potential targets for the development of broad-spectrum antiviral drugs.

Soybean Stem Fly, Melanagromyza sojae (Diptera: Agromyzidae), in the New World: detection of high genetic diversity from soybean fields in Brazil

Soybean Stem Fly (SSF), Melanagromyza sojae (Zehntner), belongs to the family Agromyzidae and is highly polyphagous, attacking many plant species of the family Fabaceae, including soybean and other beans. SSF is regarded as one of the most important pests in soybean fields of Asia (e.g., China, India), North East Africa (e.g., Egypt), parts of Russia, and South East Asia. Despite reports of Agromyzidae flies infesting soybean fields in Rio Grande do Sul State (Brazil) in 1983 and 2009 and periodic interceptions of SSF since the 1940s by the USA quarantine authorities, SSF has not been officially reported to have successfully established in the North and South Americas. In South America, M. sojae was recently confirmed using morphology and its complete mitochondrial DNA (mtDNA) was characterized. In the present study, we surveyed the genetic diversity of M. sojae, collected directly from soybean host plants, using partial mtDNA cytochrome oxidase I (COI) gene, and provide evidence of multiple (>10) maternal lineages in SSF populations in South America, potentially representing multiple incursion events. However, a single incursion involving multiple-female founders could not be ruled out. We identified a haplotype that was common in the fields of two Brazilian states and the individuals collected from Australia in 2013. The implications of SSF incursions in southern Brazil are discussed in relation to the current soybean agricultural practices, highlighting an urgent need for better understanding of SSF population movements in the New World, which is necessary for developing effective management options for this significant soybean pest. © FUNPEC-RP.

Measurement of finite-frequency current statistics in a single-electron transistor

Electron transport in nanoscale structures is strongly influenced by the Coulomb interaction that gives rise to correlations in the stream of charges and leaves clear fingerprints in the fluctuations of the electrical current. A complete understanding of the underlying physical processes requires measurements of the electrical fluctuations on all time and frequency scales, but experiments have so far been restricted to fixed frequency ranges, as broadband detection of current fluctuations is an inherently difficult experimental procedure. Here we demonstrate that the electrical fluctuations in a single-electron transistor can be accurately measured on all relevant frequencies using a nearby quantum point contact for on-chip real-time detection of the current pulses in the single-electron device. We have directly measured the frequency-dependent current statistics and, hereby, fully characterized the fundamental tunnelling processes in the single-electron transistor. Our experiment paves the way for future investigations of interaction and coherence-induced correlation effects in quantum transport.

Dempster-Shafer for Anomaly Detection

In this paper, we implement an anomaly detection system using the Dempster-Shafer method. Using two standard benchmark problems we show that by combining multiple signals it is possible to achieve better results than by using a single signal. We further show that by applying this approach to a real-world email dataset the algorithm works for email worm detection. Dempster-Shafer can be a promising method for anomaly detection problems with multiple features (data sources), and two or more classes.

High-resolution electrophoretic procedures for the identification of five Eimeria species from chickens, and detection of population variation

To overcome limitations of conventional approaches for the identification of Eimeria species of chickens, we have established high resolution electrophoretic procedures using genetic markers in ribosomal DNA. The first and second internal transcribed spacer (ITS-1 and ITS-2) regions of ribosomal DNA were amplified by polymerase chain reaction (PCR) from genomic DNA samples representing five species of Eimeria (E. acervulina, E. brunetti, E. maxima, E. necatrix and E. tenella), denatured and then subjected to denaturing polyacrylamide gel electrophoresis (D-PAGE) or single-strand conformation polymorphism (SSCP) analysis. Differences in D-PAGE profiles for both the ITS-1 and ITS-2 fragments (combined with an apparent lack of variation within individual species) enabled the unequivocal identification of the five species, and SSCP allowed the detection of population variation between some isolates representing E. acervulina, which remained undetected by D-PAGE. The establishment of these approaches has important implications for controlling the purity of laboratory lines of Eimeria, for diagnosis and for studying the epidemiology of coccidiosis.

The SOPHIE search for northern extrasolar planets: X. Detection and characterization of giant planets by the dozen

We present new radial velocity measurements of eight stars that were secured with the spectrograph SOPHIE at the 193 cm telescope of the Haute-Provence Observatory. The measurements allow detecting and characterizing new giant extrasolar planets. The host stars are dwarfs of spectral types between F5 and K0 and magnitudes of between 6.7 and 9.6; the planets have minimum masses Mp sin i of between 0.4 to 3.8 MJup and orbitalperiods of several days to several months. The data allow only single planets to be discovered around the first six stars (HD 143105, HIP 109600, HD 35759, HIP 109384, HD 220842, and HD 12484), but one of them shows the signature of an additional substellar companion in the system. The seventh star, HIP 65407, allows the discovery of two giant planets that orbit just outside the 12:5 resonance in weak mutual interaction. The last star, HD 141399, was already known to host a four-planet system; our additional data and analyses allow new constraints to be set on it. We present Keplerian orbits of all systems, together with dynamical analyses of the two multi-planet systems. HD 143105 is one of the brightest stars known to host a hot Jupiter, which could allow numerous follow-up studies to be conducted even though this is not a transiting system. The giant planets HIP 109600b, HIP 109384b, and HD 141399c are located in the habitable zone of their host star.

Chilean IPNV isolates: Robustness analysis of PCR detection

Background: The genomes of several infectious pancreatic necrosis viruses (IPNVs) isolated in Chile were sequenced with a single amplification approach for both segments A and B. The resulting sequences were then used to determine the conservation of the primer-binding regions used in polymerase chain reaction (PCR)-based diagnostic methods proposed in the literature. Thus, the robustness of each technique was studied, particularly the eventual effect of further mutations within the primer-binding sites. Results: On analysis, most methods currently used to detect Chilean IPNV varieties were deemed adequate. However, the primers were designed to be genogroup specific, implying that most detection methods pose some risk of detecting all strains prevalent in the country, due to the coexistence of genogroups 1 and 5. Conclusions: Negative resultsmust be interpreted carefully given the high genomic variability of IPNVs. Detection techniques (quantitative reverse transcription (qRT)-PCR) based on degenerate primers can be used to minimize the possibilities of false-negative detections.

Extraction and detection of mycotoxins in medicinal and aromatic plants: a case study with Aloysia citrodora P.

Plants frequently suffer contaminations by toxigenic fungi, and their mycotoxins can be produced throughout growth, harvest, drying and storage periods. The objective of this work was to validate a method for detection of toxins in medicinal and aromatic plants, through a fast and highly sensitive method, optimizing the joint co-extraction of aflatoxins (AF: AFB1, AFB2, AFG1 and AFG2) and ochratoxin A (OTA) by using Aloysia citrodora P. (lemon verbena) as a case study. For optimization purposes, samples were spiked (n=3) with standard solutions of a mix of the four AFs and OTA at 10 ng/g for AFB1, AFG1 and OTA, and at 6 ng/g of AFB2 and AFG2. Several extraction procedures were tested: i) ultrasound-assisted extraction in sodium chloride and methanol/water (80:20, v/v) [(OTA+AFs)1]; ii) maceration in methanol/1% NaHCO3 (70:30, v/v) [(OTA+AFs)2]; iii) maceration in methanol/1% NaHCO3 (70:30, v/v) (OTA1); and iv) maceration in sodium chloride and methanol/water (80:20, v/v) (AF1). AF and OTA were purified using the mycotoxin-specific immunoaffinity columns AflaTest WB and OchraTest WB (VICAM), respectively. Separation was performed with a Merck Chromolith Performance C18 column (100 x 4.6 mm) by reverse-phase HPLC coupled to a fluorescence detector (FLD) and a photochemical derivatization system (for AF). The recoveries obtained from the spiked samples showed that the single-extraction methods (OTA1 and AF1) performed better than co-extraction methods. For in-house validation of the selected methods OTA1 and AF1, recovery and precision were determined (n=6). The recovery of OTA for method OTA1 was 81%, and intermediate precision (RSDint) was 1.1%. The recoveries of AFB1, AFB2, AFG1 and AFG2 ranged from 64% to 110% for method AF1, with RSDint lower than 5%. Methods OTA1 and AF1 showed precision and recoveries within the legislated values and were found to be suitable for the extraction of OTA and AF for the matrix under study.

DCA for bot detection

Ensuring the security of computers is a non-trivial task, with many techniques used by malicious users to compromise these systems. In recent years a new threat has emerged in the form of networks of hijacked zombie machines used to perform complex distributed attacks such as denial of service and to obtain sensitive data such as password information. These zombie machines are said to be infected with a dasiahotpsila - a malicious piece of software which is installed on a host machine and is controlled by a remote attacker, termed the dasiabotmaster of a botnetpsila. In this work, we use the biologically inspired dendritic cell algorithm (DCA) to detect the existence of a single hot on a compromised host machine. The DCA is an immune-inspired algorithm based on an abstract model of the behaviour of the dendritic cells of the human body. The basis of anomaly detection performed by the DCA is facilitated using the correlation of behavioural attributes such as keylogging and packet flooding behaviour. The results of the application of the DCA to the detection of a single hot show that the algorithm is a successful technique for the detection of such malicious software without responding to normally running programs.

Synthetic Circuits for Feedback and Detection in Bacteria

Synthetic biology, by co-opting molecular machinery from existing organisms, can be used as a tool for building new genetic systems from scratch, for understanding natural networks through perturbation, or for hybrid circuits that piggy-back on existing cellular infrastructure. Although the toolbox for genetic circuits has greatly expanded in recent years, it is still difficult to separate the circuit function from its specific molecular implementation. In this thesis, we discuss the function-driven design of two synthetic circuit modules, and use mathematical models to understand the fundamental limits of circuit topology versus operating regimes as determined by the specific molecular implementation. First, we describe a protein concentration tracker circuit that sets the concentration of an output protein relative to the concentration of a reference protein. The functionality of this circuit relies on a single negative feedback loop that is implemented via small programmable protein scaffold domains. We build a mass-action model to understand the relevant timescales of the tracking behavior and how the input/output ratios and circuit gain might be tuned with circuit components. Second, we design an event detector circuit with permanent genetic memory that can record order and timing between two chemical events. This circuit was implemented using bacteriophage integrases that recombine specific segments of DNA in response to chemical inputs. We simulate expected population-level outcomes using a stochastic Markov-chain model, and investigate how inferences on past events can be made from differences between single-cell and population-level responses. Additionally, we present some preliminary investigations on spatial patterning using the event detector circuit as well as the design of stationary phase promoters for growth-phase dependent activation. These results advance our understanding of synthetic gene circuits, and contribute towards the use of circuit modules as building blocks for larger and more complex synthetic networks.