Detecting abnormal cell behaviors from dry mass time series.

The prediction of pathological changes on single cell behaviour is a challenging task for deep learning models. Indeed, in self-supervised learning methods, no prior labels are used for the training and all of the information for event predictions are extracted from the data themselves. We present here a novel self-supervised learning model for the detection of anomalies in a given cell population, StArDusTS. Cells are monitored over time, and analysed to extract time-series of dry mass values. We assessed its performances on different cell lines, showing a precision of 96% in the automatic detection of anomalies. Additionally, anomaly detection was also associated with cell measurement errors inherent to the acquisition or analysis pipelines, leading to an improvement of the upstream methods for feature extraction. Our results pave the way to novel architectures for the continuous monitoring of cell cultures in applied research or bioproduction applications, and for the prediction of pathological cellular changes.

The nocturnal life of the great scallops (Pecten maximus, L.): First description of their natural daily valve opening cycle.

Valvometry techniques used to monitor bivalve gaping activity have elucidated numerous relationships with environmental fluctuations, along with biological rhythms ranging from sub-daily to seasonal. Thus, a precise understanding of the natural activity of bivalves (i.e., not exposed to stressful environmental variations) is necessary as a baseline for detecting abnormal behaviors (deviations). This knowledge is also needed to reliably interpret observations of bivalve gaping behavior and associated biological processes (e.g., respiration, nutrition) acquired over time-limited periods. With this in mind, we investigated the natural daily gaping activity of the great scallop (Pecten maximus) by continuously monitoring 35 individuals in several individual tanks and in situ (Bay of Saint-Brieuc, Brittany, France) using fully autonomous Hall effect sensors. Our results revealed a circadian cycle (τ = 24.0h) in scallop gaping activity. Despite significant inter-individual variability in mean opening and cycle amplitude, almost all individuals (87.5%) exhibited nocturnal activity, with valves more open at night than during the day. A shift in light regime in the tanks triggered an instantaneous change in opening pattern, indicating that light levels strongly determine scallop activity. Based on the opening status of scallops, we also identified several gaping behaviors deviating from the regular daily pattern (lack of rhythmicity, high daytime opening), potentially reflecting physiological weakness. While further long-term studies are required to fully understand the natural activity of scallops, these findings pave the way for studies focused on the scallop response to external factors and introduce further research into the detection of abnormal behaviors. Coupling observations of diel valve gaping cycles with other daily variations in organismal and environmental parameters could help explain mechanisms driving the growth patterns of scallops observed in their shell striations. From a technical perspective, our field-based monitoring demonstrates the suitability of autonomous valvometry sensors for studying mobile subtidal bivalve activity in remote offshore environments.

Multiplane deconvolution in underwater acoustics: Simultaneous estimations of source level and position.

In many acoustic imaging applications, conventional beamforming (CBF) cannot provide both accurate position and source level estimates simultaneously. Also, the CBF acoustic maps suffer from many artifacts due to the spreading of large point-spread-functions. An original CLEAN deconvolution procedure, including an additional plane containing out-of-plane interfering sources, is proposed here to achieve simultaneous localization, source level estimation, and de-noising. The approach is illustrated using experimental data mimicking a challenging deep-sea mining configuration: an underwater acoustic source of interest is located 700 m below the sea surface, tens of meters from a 3 m-length array, with boat noise as the disturbing source.

Surface perturbation inverted from angle variations of eigenbeams in an ultrasonic waveguide.

Ocean acoustic tomography is traditionally performed using the travel-time variations of an acoustic path between a source and a receiver. In the context of shallow-water tomography and multipath propagation, the different acoustic paths can be correctly identified if the source and the receiver are arrays of transducers. Here, a double-beamforming algorithm can be applied to extract a collection of eigenbeams from the raw acoustic dataset. In this study, four observables can be measured for each eigenbeam: the travel-time, the amplitude, and the emitting and receiving angles. In this study, the sensitivity kernel (SK) formulation is used to establish a quantitative relation between a perturbation of the surface of an ultrasonic waveguide and the emitting and receiving angles of each eigenbeam. This theoretical relation is experimentally demonstrated using a forward model experiment designed to measure the SK. The SK formulation is then used in a second experiment to quantitatively and dynamically image the propagation of a surface wave traveling across the surface of the waveguide. The inversion results show that the quality of the joint inversion of the emitting and receiving angles is higher than previous results based on amplitude or travel-time observables.

Shells of the bivalve Astarte moerchi give new evidence of a strong pelagic-benthic coupling shift occurring since the late 1970s in the North Water polynya.

Climate changes in the Arctic may weaken the currently tight pelagic-benthic coupling. In response to decreasing sea ice cover, arctic marine systems are expected to shift from a 'sea-ice algae-benthos' to a 'phytoplankton-zooplankton' dominance. We used mollusc shells as bioarchives and fatty acid trophic markers to estimate the effects of the reduction of sea ice cover on the food exported to the seafloor. Bathyal bivalve Astarte moerchi living at 600 m depth in northern Baffin Bay reveals a clear shift in growth variations and Ba/Ca ratios since the late 1970s, which we relate to a change in food availability. Tissue fatty acid compositions show that this species feeds mainly on microalgae exported from the euphotic zone to the seabed. We, therefore, suggest that changes in pelagic-benthic coupling are likely due either to local changes in sea ice dynamics, mediated through bottom-up regulation exerted by sea ice on phytoplankton production, or to a mismatch between phytoplankton bloom and zooplankton grazing due to phenological change. Both possibilities allow a more regular and increased transfer of food to the seabed. This article is part of the theme issue 'The changing Arctic Ocean: consequences for biological communities, biogeochemical processes and ecosystem functioning'.

Real-Time Dual-Wavelength Time-Resolved Diffuse Optical Tomography System for Functional Brain Imaging Based on Probe-Hosted Silicon Photomultipliers.

Near-infrared diffuse optical tomography is a non-invasive photonics-based imaging technology suited to functional brain imaging applications. Recent developments have proved that it is possible to build a compact time-domain diffuse optical tomography system based on silicon photomultipliers (SiPM) detectors. The system presented in this paper was equipped with the same eight SiPM probe-hosted detectors, but was upgraded with six injection fibers to shine the sample at several points. Moreover, an automatic switch was included enabling a complete measurement to be performed in less than one second. Further, the system was provided with a dual-wavelength ( 670 n m and 820 n m ) light source to quantify the oxy- and deoxy-hemoglobin concentration evolution in the tissue. This novel system was challenged against a solid phantom experiment, and two in-vivo tests, namely arm occlusion and motor cortex brain activation. The results show that the tomographic system makes it possible to follow the evolution of brain activation over time with a 1 s -resolution.

Dynamic imaging of a capillary-gravity wave in shallow water using amplitude variations of eigenbeams.

Dynamic acoustic imaging of a surface wave propagating at an air-water interface is a complex task that is investigated here at the laboratory scale through an ultrasonic experiment in a shallow water waveguide. Using a double beamforming algorithm between two source-receiver arrays, the authors isolate and identify each multi-reverberated eigenbeam that interacts with the air-water and bottom interfaces. The waveguide transfer matrix is recorded 100 times per second while a low-amplitude gravity wave is generated by laser-induced breakdown at the middle of the waveguide, just above the water surface. The controlled, and therefore repeatable, breakdown results in a blast wave that interacts with the air-water interface, which creates ripples at the surface that propagate in both directions. The amplitude perturbations of each ultrasonic eigenbeam are measured during the propagation of the gravity-capillary wave. Inversion of the surface deformation is performed from the amplitude variations of the eigenbeams using a diffraction-based sensitivity kernel approach. The accurate ultrasonic imaging of the displacement of the air-water interface is compared to simultaneous measurements with an optical camera, which provides independent validation.

Automatic fish sounds classification.

The work presented in this paper focuses on the use of acoustic systems for passive acoustic monitoring of ocean vitality for fish populations. Specifically, it focuses on the use of acoustic systems for passive acoustic monitoring of ocean vitality for fish populations. To this end, various indicators can be used to monitor marine areas such as both the geographical and temporal evolution of fish populations. A discriminative model is built using supervised machine learning (random-forest and support-vector machines). Each acquisition is represented in a feature space, in which the patterns belonging to different semantic classes are as separable as possible. The set of features proposed for describing the acquisitions come from an extensive state of the art in various domains in which classification of acoustic signals is performed, including speech, music, and environmental acoustics. Furthermore, this study proposes to extract features from three representations of the data (time, frequency, and cepstral domains). The proposed classification scheme is tested on real fish sounds recorded on several areas, and achieves 96.9% correct classification compared to 72.5% when using reference state of the art features as descriptors. The classification scheme is also validated on continuous underwater recordings, thereby illustrating that it can be used to both detect and classify fish sounds in operational scenarios.

Target-depth estimation in active sonar: Cramer-Rao bounds for a bilinear sound-speed profile.

This paper develops a localization method to estimate the depth of a target in the context of active sonar, at long ranges. The target depth is tactical information for both strategy and classification purposes. The Cramer-Rao lower bounds for the target position as range and depth are derived for a bilinear profile. The influence of sonar parameters on the standard deviations of the target range and depth are studied. A localization method based on ray back-propagation with a probabilistic approach is then investigated. Monte-Carlo simulations applied to a summer Mediterranean sound-speed profile are performed to evaluate the efficiency of the estimator. This method is finally validated on data in an experimental tank.

Whistle source levels of free-ranging beluga whales in Saguenay-St. Lawrence marine park.

Wild beluga whistle source levels (SLs) are estimated from 52 three-dimensional (3D) localized calls using a 4-hydrophone array. The probability distribution functions of the root-mean-square (rms) SL in the time domain, and the peak, the strongest 3-dB, and 10-dB SLs from the spectrogram, were non-Gaussian. The average rms SL was 143.8 ± 6.7 dB re 1 µPa at 1 m. SL spectral metrics were, respectively, 145.8 ± 8 dB, 143.2 ± 7.1 dB, and 138.5 ± 6.9 dB re 1 µPa(2)·Hz(-1) at 1 m.

Arctic underwater noise transients from sea ice deformation: Characteristics, annual time series, and forcing in Beaufort Sea.

A 13-month time series of Arctic Ocean noise from the marginal ice zone of the Eastern Beaufort Sea is analyzed to detect under-ice acoustic transients isolated from ambient noise with a dedicated algorithm. Noise transients due to ice cracking, fracturing, shearing, and ridging are sorted out into three categories: broadband impulses, frequency modulated (FM) tones, and high-frequency broadband noise. Their temporal and acoustic characteristics over the 8-month ice covered period, from November 2005 to mid-June 2006, are presented and their generation mechanisms are discussed. Correlations analyses showed that the occurrence of these ice transients responded to large-scale ice motion and deformation rates forced by meteorological events, often leading to opening of large-scale leads at main discontinuities in the ice cover. Such a sequence, resulting in the opening of a large lead, hundreds by tens of kilometers in size, along the margin of landfast ice and multiyear ice plume in the Beaufort-Chukchi seas is detailed. These ice transients largely contribute to the soundscape properties of the Arctic Ocean, for both its ambient and total noise components. Some FM tonal transients can be confounded with marine mammal songs, especially when they are repeated, with periods similar to wind generated waves.

Laser line illumination scheme allowing the reduction of background signal and the correction of absorption heterogeneities effects for fluorescence reflectance imaging.

Intraoperative fluorescence imaging in reflectance geometry is an attractive imaging modality as it allows to noninvasively monitor the fluorescence targeted tumors located below the tissue surface. Some drawbacks of this technique are the background fluorescence decreasing the contrast and absorption heterogeneities leading to misinterpretations concerning fluorescence concentrations. We propose a correction technique based on a laser line scanning illumination scheme. We scan the medium with the laser line and acquire, at each position of the line, both fluorescence and excitation images. We then use the finding that there is a relationship between the excitation intensity profile and the background fluorescence one to predict the amount of signal to subtract from the fluorescence images to get a better contrast. As the light absorption information is contained both in fluorescence and excitation images, this method also permits us to correct the effects of absorption heterogeneities. This technique has been validated on simulations and experimentally. Fluorescent inclusions are observed in several configurations at depths ranging from 1 mm to 1 cm. Results obtained with this technique are compared with those obtained with a classical wide-field detection scheme for contrast enhancement and with the fluorescence by an excitation ratio approach for absorption correction.

Color Sparse Representations for Image Processing: Review, Models, and Prospects.

Sparse representations have been extended to deal with color images composed of three channels. A review of dictionary-learning-based sparse representations for color images is made here, detailing the differences between the models, and comparing their results on the real and simulated data. These models are considered in a unifying framework that is based on the degrees of freedom of the linear filtering/transformation of the color channels. Moreover, this allows it to be shown that the scalar quaternionic linear model is equivalent to constrained matrix-based color filtering, which highlights the filtering implicitly applied through this model. Based on this reformulation, the new color filtering model is introduced, using unconstrained filters. In this model, spatial morphologies of color images are encoded by atoms, and colors are encoded by color filters. Color variability is no longer captured in increasing the dictionary size, but with color filters, this gives an efficient color representation.

Laser line scanning for fluorescence reflectance imaging: a phantom study and in vivo validation of the enhancement of contrast and resolution.

Intraoperative fluorescence imaging in reflectance geometry is an attractive imaging modality to noninvasively monitor fluorescence-targeted tumors. In some situations, this kind of imaging suffers from poor resolution due to the diffusive nature of photons in tissue. The objective of the proposed technique is to tackle this limitation. It relies on the scanning of the medium with a laser line illumination and the acquisition of images at each position of excitation. The detection scheme proposed takes advantage of the stack of images acquired to enhance the resolution and the contrast of the final image. The experimental protocol is described to fully understand why we overpass the classical limits and validate the scheme on tissue-like phantoms and in vivo with a preliminary testing. The results are compared with those obtained with a classical wide-field illumination.

Shallow-water acoustic tomography from angle measurements instead of travel-time measurements.

For shallow-water waveguides and mid-frequency broadband acoustic signals, ocean acoustic tomography (OAT) is based on the multi-path aspect of wave propagation. Using arrays in emission and reception and advanced array processing, every acoustic arrival can be isolated and matched to an eigenray that is defined not only by its travel time but also by its launch and reception angles. Classically, OAT uses travel-time variations to retrieve sound-speed perturbations; this assumes very accurate source-to-receiver clock synchronization. This letter uses numerical simulations to demonstrate that launch-and-reception-angle tomography gives similar results to travel-time tomography without the same requirement for high-precision synchronization.

Automated segmentation of linear time-frequency representations of marine-mammal sounds.

Many marine mammals produce highly nonlinear frequency modulations. Determining the time-frequency support of these sounds offers various applications, which include recognition, localization, and density estimation. This study introduces a low parameterized automated spectrogram segmentation method that is based on a theoretical probabilistic framework. In the first step, the background noise in the spectrogram is fitted with a Chi-squared distribution and thresholded using a Neyman-Pearson approach. In the second step, the number of false detections in time-frequency regions is modeled as a binomial distribution, and then through a Neyman-Pearson strategy, the time-frequency bins are gathered into regions of interest. The proposed method is validated on real data of large sequences of whistles from common dolphins, collected in the Bay of Biscay (France). The proposed method is also compared with two alternative approaches: the first is smoothing and thresholding of the spectrogram; the second is thresholding of the spectrogram followed by the use of morphological operators to gather the time-frequency bins and to remove false positives. This method is shown to increase the probability of detection for the same probability of false alarms.

Under-ice ambient noise in Eastern Beaufort Sea, Canadian Arctic, and its relation to environmental forcing.

This paper analyzes an 8-month time series (November 2005 to June 2006) of underwater noise recorded at the mouth of the Amundsen Gulf in the marginal ice zone of the western Canadian Arctic when the area was >90% ice covered. The time-series of the ambient noise component was computed using an algorithm that filtered out transient acoustic events from 7-min hourly recordings of total ocean noise over a [0-4.1] kHz frequency band. Under-ice ambient noise did not respond to thermal changes, but showed consistent correlations with large-scale regional ice drift, wind speed, and measured currents in upper water column. The correlation of ambient noise with ice drift peaked for locations at ranges of ~300 km off the mouth of the Amundsen Gulf. These locations are within the multi-year ice plume that extends westerly along the coast in the Eastern Beaufort Sea due to the large Beaufort Gyre circulation. These results reveal that ambient noise in Eastern Beaufort Sea in winter is mainly controlled by the same meteorological and oceanographic forcing processes that drive the ice drift and the large-scale circulation in this part of the Arctic Ocean.

Time-angle sensitivity kernels for sound-speed perturbations in a shallow ocean.

Acoustic waves traveling in a shallow-water waveguide produce a set of multiple paths that can be characterized as a geometric approximation by their travel time (TT), direction of arrival (DOA), and direction of departure (DOD). This study introduces the use of the DOA and DOD as additional observables that can be combined to the classical TT to track sound-speed perturbations in an oceanic waveguide. To model the TT, DOA, and DOD variations induced by sound-speed perturbations, the three following steps are used: (1) In the first-order Born approximation, the Fréchet kernel provides a linear link between the signal fluctuations and the sound-speed perturbations; (2) a double-beamforming algorithm is used to transform the signal fluctuations received on two source-receiver arrays in the time, receiver-depth, and source-depth domain into the eigenray equivalent measured in the time, reception-angle and launch angle domain; and finally (3) the TT, DOA, and DOD variations are extracted from the double-beamformed signal variations through a first-order Taylor development. As a result, time-angle sensitivity kernels are defined and used to build a linear relationship between the observable variations and the sound-speed perturbations. This approach is validated with parabolic-equation simulations in a shallow-water ocean context.

Travel-time tomography in shallow water: experimental demonstration at an ultrasonic scale.

Acoustic tomography in a shallow ultrasonic waveguide is demonstrated at the laboratory scale between two source-receiver arrays. At a 1/1,000 scale, the waveguide represents a 1.1-km-long, 52-m-deep ocean acoustic channel in the kilohertz frequency range. Two coplanar arrays record the transfer matrix in the time domain of the waveguide between each pair of source-receiver transducers. A time-domain, double-beamforming algorithm is simultaneously performed on the source and receiver arrays that projects the multi-reflected acoustic echoes into an equivalent set of eigenrays, which are characterized by their travel times and their launch and arrival angles. Travel-time differences are measured for each eigenray every 0.1 s when a thermal plume is generated at a given location in the waveguide. Travel-time tomography inversion is then performed using two forward models based either on ray theory or on the diffraction-based sensitivity kernel. The spatially resolved range and depth inversion data confirm the feasibility of acoustic tomography in shallow water. Comparisons are made between inversion results at 1 and 3 MHz with the inversion procedure using ray theory or the finite-frequency approach. The influence of surface fluctuations at the air-water interface is shown and discussed in the framework of shallow-water ocean tomography.

In vivo fluorescence spectra unmixing and autofluorescence removal by sparse nonnegative matrix factorization.

Fluorescence imaging locates fluorescent markers that specifically bind to targets; like tumors, markers are injected to a patient, optimally excited with near-infrared light, and located thanks to backward-emitted fluorescence analysis. To investigate thick and diffusive media, as the fluorescence signal decreases exponentially with the light travel distance, the autofluorescence of biological tissues comes to be a limiting factor. To remove autofluorescence and isolate specific fluorescence, a spectroscopic approach, based on nonnegative matrix factorization (NMF), is explored. To improve results on spatially sparse markers detection, we suggest a new constrained NMF algorithm that takes sparsity constraints into account. A comparative study between both algorithms is proposed on simulated and in vivo data.