Double vision: 2D and 3D mosquito trajectories can be as valuable for behaviour analysis via machine learning.

BACKGROUND: Mosquitoes are carriers of tropical diseases, thus demanding a comprehensive understanding of their behaviour to devise effective disease control strategies. In this article we show that machine learning can provide a performance assessment of 2D and 3D machine vision techniques and thereby guide entomologists towards appropriate experimental approaches for behaviour assessment. Behaviours are best characterised via tracking-giving a full time series of information. However, tracking systems vary in complexity. Single-camera imaging yields two-component position data which generally are a function of all three orthogonal components due to perspective; however, a telecentric imaging setup gives constant magnification with respect to depth and thereby measures two orthogonal position components. Multi-camera or holographic techniques quantify all three components. METHODS: In this study a 3D mosquito mating swarm dataset was used to generate equivalent 2D data via telecentric imaging and a single camera at various imaging distances. The performance of the tracking systems was assessed through an established machine learning classifier that differentiates male and non-male mosquito tracks. SHAPs analysis has been used to explore the trajectory feature values for each model. RESULTS: The results reveal that both telecentric and single-camera models, when placed at large distances from the flying mosquitoes, can produce equivalent accuracy from a classifier as well as preserve characteristic features without resorting to more complex 3D tracking techniques. CONCLUSIONS: Caution should be exercised when employing a single camera at short distances as classifier balanced accuracy is reduced compared to that from 3D or telecentric imaging; the trajectory features also deviate compared to those from the other datasets. It is postulated that measurement of two orthogonal motion components is necessary to optimise the accuracy of machine learning classifiers based on trajectory data. The study increases the evidence base for using machine learning to determine behaviours from insect trajectory data.

Insecticidal roof barriers mounted on untreated bed nets can be as effective against Anopheles gambiae as regular insecticide-treated bed nets.

Barrier bednets (BBnets), regular bednets with a vertical insecticidal panel to target mosquitoes above the bednet roof, where they are most active, have the potential to improve existing Insecticidal Treated Bednets (ITNs), by reducing the quantity of insecticide required per net, reducing the toxic risks to those using the net, potentially increasing insecticide choice. We evaluated the performance of PermaNet 3.0 (P3) and untreated (Ut) bed nets with and without pyrethroid and piperonyl butoxide roof barriers in killing pyrethroid-resistant and susceptible Anopheles gambiae, simultaneously video-recording mosquito flight tracks. Bioassay results showed that treated roof barriers, particularly the longitudinal P3 barrier (P3L) could be an effective addition to a bed net: P3 + P3L were consistently significantly more effective than the reference P3 bednet while performance of untreated nets could be raised to equal that of the reference P3 following the addition of a P3 barrier. The BBnet's potential to augment existing bednets and enhance their performance is considered.

Impacts of dual active-ingredient bed nets on the behavioural responses of pyrethroid resistant Anopheles gambiae determined by room-scale infrared video tracking.

BACKGROUND: The success of insecticide treated bed nets (ITNs) for malaria vector control in Africa relies on the behaviour of various species of Anopheles. Previous research has described mosquito behavioural alterations resulting from widespread ITN coverage, which could result in a decrease in net efficacy. Here, behaviours were compared including timings of net contact, willingness to refeed and longevity post-exposure to two next-generation nets, PermaNet® 3.0 (P3 net) and Interceptor® G2 (IG2 net) in comparison with a standard pyrethroid-only net (Olyset Net™ (OL net)) and an untreated net. METHODS: Susceptible and resistant Anopheles gambiae mosquitoes were exposed to the nets with a human volunteer host in a room-scale assay. Mosquito movements were tracked for 2 h using an infrared video system, collecting flight trajectory, spatial position and net contact data. Post-assay, mosquitoes were monitored for a range of sublethal insecticide effects. RESULTS: Mosquito net contact was focused predominantly on the roof for all four bed nets. A steep decay in activity was observed for both susceptible strains when P3 net and OL net were present and with IG2 net for one of the two susceptible strains. Total mosquito activity was higher around untreated nets than ITNs. There was no difference in total activity, the number, or duration, of net contact, between any mosquito strain, with similar behaviours recorded in susceptible and resistant strains at all ITNs. OL net, P3 net and IG2 net all killed over 90% of susceptible mosquitoes 24 h after exposure, but this effect was not seen with resistant mosquitoes where mortality ranged from 16 to 72%. All treated nets reduced the willingness of resistant strains to re-feed when offered blood 1-h post-exposure, with a more pronounced effect seen with P3 net and OL net than IG2 net. CONCLUSION: These are the first results to provide an in-depth description of the behaviour of susceptible and resistant Anopheles gambiae strains around next-generation bed nets using a room-scale tracking system to capture multiple behaviours. These results indicate that there is no major difference in behavioural responses between mosquito strains of differing pyrethroid susceptibility when exposed to these new ITNs under the experimental conditions used.

Finding a Husband: Using Explainable AI to Define Male Mosquito Flight Differences.

Mosquito-borne diseases account for around one million deaths annually. There is a constant need for novel intervention mechanisms to mitigate transmission, especially as current insecticidal methods become less effective with the rise of insecticide resistance among mosquito populations. Previously, we used a near infra-red tracking system to describe the behaviour of mosquitoes at a human-occupied bed net, work that eventually led to an entirely novel bed net design. Advancing that approach, here we report on the use of trajectory analysis of a mosquito flight, using machine learning methods. This largely unexplored application has significant potential for providing useful insights into the behaviour of mosquitoes and other insects. In this work, a novel methodology applies anomaly detection to distinguish male mosquito tracks from females and couples. The proposed pipeline uses new feature engineering techniques and splits each track into segments such that detailed flight behaviour differences influence the classifier rather than the experimental constraints such as the field of view of the tracking system. Each segment is individually classified and the outcomes are combined to classify whole tracks. By interpreting the model using SHAP values, the features of flight that contribute to the differences between sexes are found and are explained by expert opinion. This methodology was tested using 3D tracks generated from mosquito mating swarms in the field and obtained a balanced accuracy of 64.5% and an ROC AUC score of 68.4%. Such a system can be used in a wide variety of trajectory domains to detect and analyse the behaviours of different classes, e.g., sex, strain, and species. The results of this study can support genetic mosquito control interventions for which mating represents a key event for their success.

Improved three-dimensional localization of multiple small objects in close proximity in digital holography.

Using intensity gradient- or sparsity-based focus metrics, the ability to accurately localize the three-dimensional (3D) position of a small object in a digital holographic reconstruction of a large field of view is hindered in the presence of multiple nearby objects. A more accurate alternative method for 3D localization, based on evaluation of the complex reconstructed volume, is proposed. Simulations and experimental data demonstrate a reduction in depth positional error for single objects and a notably improved axial resolution of multiple objects in close proximity.

Diffuse retro-reflective imaging for improved video tracking of mosquitoes at human baited bednets.

Robust imaging techniques for tracking insects have been essential tools in numerous laboratory and field studies on pests, beneficial insects and model systems. Recent innovations in optical imaging systems and associated signal processing have enabled detailed characterization of nocturnal mosquito behaviour around bednets and improvements in bednet design, a global essential for protecting populations against malaria. Nonetheless, there remain challenges around ease of use for large-scale in situ recordings and extracting data reliably in the critical areas of the bednet where the optical signal is attenuated. Here, we introduce a retro-reflective screen at the back of the measurement volume, which can simultaneously provide diffuse illumination, and remove optical alignment issues while requiring only one-sided access to the measurement space. The illumination becomes significantly more uniform, although noise removal algorithms are needed to reduce the effects of shot noise, particularly across low-intensity bednet regions. By systematically introducing mosquitoes in front of and behind the bednet in laboratory experiments, we are able to demonstrate robust tracking in these challenging areas. Overall, the retro-reflective imaging set-up delivers mosquito segmentation rates in excess of 90% compared to less than 70% with backlit systems.

Quantifying late-stage host-seeking behaviour of Anopheles gambiae at the insecticidal net interface using a baited-box bioassay.

BACKGROUND: Insecticide-treated nets (ITNs) are losing efficacy against pyrethroid-resistant malaria vector populations throughout Africa. Safeguarding bed net efficacy, vital for effective malaria control, requires greater knowledge of mosquito-ITN interactions and how this impacts on the mosquito. METHODS: A purpose-built benchtop apparatus with a closed 10 cm cubic chamber (the 'Baited-box') was used to video record behaviour of individual free-flying female Anopheles gambiae during approach and blood-feeding on a human hand through untreated nets and ITNs at close range. Time and duration of defined behavioural events, and knockdown and mortality at 1- and 24-h post-exposure respectively, were recorded for pyrethroid susceptible and resistant mosquitoes. RESULTS: Using three human volunteers differing in relative attractiveness to mosquitoes, 328 mosquitoes were individually tested. There were no significant differences between response rates to ITNs and untreated nets (P > 0.1) or between resistant (Tiassalé) and susceptible (Kisumu) mosquito strains, at untreated nets (P = 0.39) or PermaNet 2.0 (P = 1). The sequence of behavioural events from host-seeking to completion of blood-feeding was consistent in all tests but duration and start time of events involving net contact were reduced or delayed respectively with ITNs. Blood-feeding durations at untreated nets (means from 4.25 to 8.47 min (95% confidence interval (CI) = 3.39-9.89) at 3 human volunteers) were reduced by 37-50% at PermaNet 2.0, in susceptible (mean 2.59-4.72 min, 95% CI = 1.54-5.5, P = < 0.05) and resistant (mean 4.20 min, 95% CI = 3.42-4.97, P = 0.01) strains. Total accumulated net contact was approximately 50% lower at PermaNet and Olyset ITNs (P < 0.0001) in susceptible (two of the three volunteers) and resistant mosquitoes. Times prior to first net contact were similar at untreated nets and ITNs (P > 0.2), and neither ITN type showed detectable spatial repellency. After initial contact, blood-feeding commenced later at Olyset (mean 2.76 min, 95% CI = 1.74-3.76, P = 0.0009) and PermaNet (mean 2.4 min, 95% CI = 1.52-3.33, P = 0.0058) than untreated netting (mean 0.68 min, 95% CI = 0.42-0.94). CONCLUSIONS: The baited box offers a simple method for detailed characterization of mosquito behavioural responses to insecticidal nets, for comparing entomological modes of action between nets and for defining the behavioural responses of particular mosquito strains or populations. The device has potential as a screening assay in the search for novel net treatments and for investigations into behavioural resistance mechanisms.

Barrier bednets target malaria vectors and expand the range of usable insecticides.

Transmission of Plasmodium falciparum malaria parasites occurs when nocturnal Anopheles mosquito vectors feed on human blood. In Africa, where malaria burden is highest, bednets treated with pyrethroid insecticide were highly effective in preventing mosquito bites and reducing transmission, and essential to achieving unprecedented reductions in malaria until 2015 (ref. 1). Since then, progress has stalled2, and with insecticidal bednets losing efficacy against pyrethroid-resistant Anopheles vectors3,4, methods that restore performance are urgently needed to eliminate any risk of malaria returning to the levels seen before their widespread use throughout sub-Saharan Africa5. Here, we show that the primary malaria vector Anopheles gambiae is targeted and killed by small insecticidal net barriers positioned above a standard bednet in a spatial region of high mosquito activity but zero contact with sleepers, opening the way for deploying many more insecticides on bednets than is currently possible. Tested against wild pyrethroid-resistant A. gambiae in Burkina Faso, pyrethroid bednets with organophosphate barriers achieved significantly higher killing rates than bednets alone. Treated barriers on untreated bednets were equally effective, without significant loss of personal protection. Mathematical modelling of transmission dynamics predicted reductions in clinical malaria incidence with barrier bednets that matched those of 'next-generation' nets recommended by the World Health Organization against resistant vectors. Mathematical models of mosquito-barrier interactions identified alternative barrier designs to increase performance. Barrier bednets that overcome insecticide resistance are feasible using existing insecticides and production technology, and early implementation of affordable vector control tools is a realistic prospect.

Host-seeking activity of a Tanzanian population of Anopheles arabiensis at an insecticide treated bed net.

BACKGROUND: Understanding how mosquitoes respond to long lasting insecticide treated nets (LLINs) is fundamental to sustaining the effectiveness of this essential control tool. We report on studies with a tracking system to investigate behaviour of wild anophelines at an LLIN, in an experimental hut at a rural site in Mwanza, Tanzania. METHODS: Groups of adult female mosquitoes (n = 10 per replicate) reared from larvae of a local population, identified as predominantly (95%) Anopheles arabiensis, were released in the hut. An infrared video tracking system recorded flight and net contact activity over 1 h as the mosquitoes attempted to reach a supine human volunteer within a bed net (either a deltamethrin-treated LLIN or an untreated control net). A range of activities, including flight path, position in relation to the bed net and duration of net contact, were quantified and compared between treatments. RESULTS: The total time that female An. arabiensis spent in flight around LLINs was significantly lower than at untreated nets [F(1,10) = 9.26, p = 0.012], primarily due to a substantial reduction in the time mosquitoes spent in persistent 'bouncing' flight [F(1,10) = 18.48, p = 0.002]. Most activity occurred at the net roof but significantly less so with LLINs (56.8% of total) than untreated nets [85.0%; Χ2 (15) = 234.69, p < 0.001]. Activity levels at the bed net directly above the host torso were significantly higher with untreated nets (74.2%) than LLINs [38.4%; Χ2 (15) = 33.54, p = 0.004]. 'Visiting' and 'bouncing' rates were highest above the volunteer's chest in untreated nets (39.9 and 50.4%, respectively) and LLINs [29.9 and 42.4%; Χ2 (13) = 89.91, p < 0.001; Χ2 (9) = 45.73, p < 0.001]. Highest resting rates were above the torso in untreated nets [77%; Χ2 (9) = 63.12, p < 0.001], but in LLINs only 33.2% of resting occurred here [Χ2 (9) = 27.59, p = 0.001], with resting times spread between the short vertical side of the net adjacent to the volunteer's head (21.8%) and feet (16.2%). Duration of net contact by a single mosquito was estimated at 204-290 s on untreated nets and 46-82 s on LLINs. While latency to net contact was similar in both treatments, the reduction in activity over 60 min was significantly more rapid for LLINs [F(1,10) = 6.81, p = 0.026], reiterating an 'attract and kill' rather than a repellent mode of action. CONCLUSIONS: The study has demonstrated the potential for detailed investigations of behaviour of wild mosquito populations under field conditions. The results validate the findings of earlier laboratory studies on mosquito activity at LLINs, and reinforce the key role of multiple brief contacts at the net roof as the critical LLIN mode of action.

High-Precision Registration of Point Clouds Based on Sphere Feature Constraints.

Point cloud registration is a key process in multi-view 3D measurements. Its precision affects the measurement precision directly. However, in the case of the point clouds with non-overlapping areas or curvature invariant surface, it is difficult to achieve a high precision. A high precision registration method based on sphere feature constraint is presented to overcome the difficulty in the paper. Some known sphere features with constraints are used to construct virtual overlapping areas. The virtual overlapping areas provide more accurate corresponding point pairs and reduce the influence of noise. Then the transformation parameters between the registered point clouds are solved by an optimization method with weight function. In that case, the impact of large noise in point clouds can be reduced and a high precision registration is achieved. Simulation and experiments validate the proposed method.

Infrared video tracking of Anopheles gambiae at insecticide-treated bed nets reveals rapid decisive impact after brief localised net contact.

Long-lasting insecticidal bed nets (LLINs) protect humans from malaria transmission and are fundamental to malaria control worldwide, but little is known of how mosquitoes interact with nets. Elucidating LLIN mode of action is essential to maintain or improve efficacy, an urgent need as emerging insecticide resistance threatens their future. Tracking multiple free-flying Anopheles gambiae responding to human-occupied bed nets in a novel large-scale system, we characterised key behaviours and events. Four behavioural modes with different levels of net contact were defined: swooping, visiting, bouncing and resting. Approximately 75% of all activity occurred at the bed net roof where multiple brief contacts were focussed above the occupant's torso. Total flight and net contact times were lower at LLINs than untreated nets but the essential character of the response was unaltered. LLINs did not repel mosquitoes but impacted rapidly: LLIN contact of less than 1 minute per mosquito during the first ten minutes reduced subsequent activity; after thirty minutes, activity at LLINs was negligible. Velocity measurements showed that mosquitoes detected nets, including unbaited untreated nets, prior to contact. This is the most complete characterisation of mosquito-LLIN interactions to date, and reveals many aspects of LLIN mode of action, important for developing the next generation of LLINs.

Interactive effects of trait and state affect on top-down control of attention.

Few studies have investigated how attentional control is affected by transient affective states while taking individual differences in affective traits into consideration. In this study, participants completed a color-word Stroop task immediately after undergoing a positive, neutral or negative affective context manipulation (ACM). Behavioral performance was unaffected by any ACM considered in isolation. For individuals high in trait negative affect (NA), performance was impaired by the negative but not the positive or neutral ACM. Neuroimaging results indicate that activity in primarily top-down control regions of the brain (inferior frontal gyrus and dorsal anterior cingulate cortex) was suppressed in the presence of emotional arousal (both negative and positive ACMs). This effect appears to have been exacerbated or offset by co-occurring activity in other top-down control regions (parietal) and emotion processing regions (orbitofrontal cortex, amygdala and nucleus accumbens) as a function of the valence of state affect (positive or negative) and trait affect (trait NA or trait PA). Neuroimaging results are consistent with behavioral findings. In combination, they indicate both additive and interactive influences of trait and state affect on top-down control of attention.

Algebraic solution for phase unwrapping problems in multiwavelength interferometry.

Recent advances in multiwavelength interferometry techniques [Appl. Opt.52, 5758 (2013)] give new insights to phase unwrapping problems and allow the fringe order information contained in the measured phase to be extracted with low computational effort. This work introduces an algebraic solution to the phase unwrapping problem that allows the direct calculation of the unknown integer fringe order. The procedure resembles beat-wavelength approaches, but provides greater flexibility in choosing the measurement wavelengths, a larger measurement range, and a higher robustness against noise, due to the ability to correct for errors during the calculation.

Resting and task-elicited prefrontal EEG alpha asymmetry in depression: support for the capability model.

The capability model of frontal electroencephalographic (EEG) asymmetry suggests that brain activity during emotional challenge will be a more powerful indicator of predispositions toward psychopathology than activity observed at rest. EEG data were assessed during a resting baseline and a facial emotion task, wherein individuals with (n = 143) and without (n = 163) lifetime major depressive disorder (MDD) made approach (angry and happy) and withdrawal (afraid and sad) facial expressions. EEG asymmetry during emotional challenge was a more powerful indicator of MDD status than resting asymmetry for average, Cz, and linked mastoid references, results in support of the capability model. However, current-source-density (CSD) transformed asymmetry was indicative of lifetime MDD status under resting and task-elicited conditions. Findings suggest that CSD-transformed data may be more robust indicators of trait frontal EEG asymmetry.

Method of excess fractions with application to absolute distance metrology: analytical solution.

Multiwavelength interferometry provides a solution to a number of applications in metrology for the measurement of optical path differences longer than the source wavelength. To this day, the method of excess fractions (EF) has proved to provide very long, unambiguous measurement ranges with the highest reliability for a given set of wavelengths and level of phase noise. This is achieved because EF combines the individual phase values in an equivalent least-square problem and evaluates the correspondence for all possible solutions. However, this procedure can be slow for a number of applications. In this paper, an analytical solution for EF is presented that allows the direct calculation of the unknown integer fringe order. It is shown that this solution is consistent with the other phase unwrapping approaches as beat wavelength or Chinese remainder theorem-based solutions, but moreover, it can be understood as a unified representation and solution of the fringe order problem.

Method of excess fractions with application to absolute distance metrology: wavelength selection and the effects of common error sources.

Multiwavelength interferometry (MWI) is a well established technique in the field of optical metrology. Previously, we have reported a theoretical analysis of the method of excess fractions that describes the mutual dependence of unambiguous measurement range, reliability, and the measurement wavelengths. In this paper wavelength, selection strategies are introduced that are built on the theoretical description and maximize the reliability in the calculated fringe order for a given measurement range, number of wavelengths, and level of phase noise. Practical implementation issues for an MWI interferometer are analyzed theoretically. It is shown that dispersion compensation is best implemented by use of reference measurements around absolute zero in the interferometer. Furthermore, the effects of wavelength uncertainty allow the ultimate performance of an MWI interferometer to be estimated.

Should it matter when we record? Time of year and time of day as factors influencing frontal EEG asymmetry.

Resting frontal encephalographic (EEG) asymmetry, often conceptualized as a trait marker for depression, is influenced by occasion-specific factors, including time of year and the time of day of the recording session as demonstrated recently (Peterson and Harmon-Jones, 2009). The current study examined the influence of seasonal and chronological variables on resting frontal asymmetry, and also assessed whether different reference montages or surface transformations were equally susceptible to these influences. In a direct replication attempt, contrary to previous findings, no simple time of year by time of day interaction was found. Time awake at recording, however, was an important moderating variable of the relationship between photoperiod and time of day. EEG asymmetry scores based on current-source density (CSD) transformed data, however, appeared less vulnerable to these influences, providing further evidence to suggest that the CSD transform may be advantageous for examining stable trait estimates of frontal EEG asymmetry.

Method of excess fractions with application to absolute distance metrology: theoretical analysis.

The method of excess fractions (EF) is well established to resolve the fringe order ambiguity generated in interferometric detection. Despite this background, multiwavelength interferometric absolute long distance measurements have only been reported with varying degrees of success. In this paper we present a theoretical model that can predict the unambiguous measurement range in EF based on the selected measurement wavelengths and phase noise. It is shown that beat wavelength solutions are a subset of this theoretical model. The performance of EF, for a given phase noise, is shown to be equivalent to beat techniques but offers many alternative sets of measurement wavelengths and therefore EF offer significantly greater flexibility in experimental design.

Depression and anxious apprehension distinguish frontocingulate cortical activity during top-down attentional control.

A network consisting of left dorsolateral prefrontal cortex (LDLPFC) and dorsal anterior cingulate cortex (dACC) has been implicated in top-down attentional control. Few studies have systematically investigated how this network is altered in psychopathology, despite evidence that depression and anxiety are associated with attentional control impairments. Functional MRI and dense-array event-related brain potential (ERP) data were collected in separate sessions from 100 participants during a color-word Stroop task. Functional MRI results guided ERP source modeling to characterize the time course of activity in LDLPFC (300-440 ms) and dACC (520-680 ms). At low levels of depression, LDLPFC activity was indirectly related to Stroop interference and only via dACC activity. In contrast, at high levels of depression, dACC did not play an intervening role, and increased LDLPFC activity was directly related to decreased Stroop interference. Specific to high levels of anxious apprehension, higher dACC activity was related to more Stroop interference. Results indicate that depression and anxious apprehension modulate temporally and functionally distinct aspects of the frontocingulate network involved in top-down attention control.