Biology subjects

Kervrann, C.

Publications and source records attributed to Kervrann, C..

14 recordsLinked to original sources

bioRxiv2026

Deep Learning-Meets-Active Surfaces: A Unified Framework for Multi-Object Segmentation and Shape Analysis in 3D Fluorescence Microscopy

In this paper, we introduce Nagini3D, a 3D biological image segmentation method that hybridizes two high-performing yet highly complementary segmentation paradigms: deep learning-based approaches and snake-based methods. Our approach leverages both the high efficiency of data-driven techniques and the convenient shape representation provided by active surfaces, which proves highly practical for incorporating a priori knowledge as well as for the post-analysis of segmented objects. By considering a complete 3D volume, our method ensures segmentation consistency across all three dimensions, unlike many approaches that handle volumes in a slice-by-slice manner. Additionally, it theoretically enables the representation of any object with spherical topology, overcoming the limitations of common 3D methods that rely on object convexity assumptions. Furthermore, the continuous representation of objects allows for a resolution-independent description, facilitating the extraction of various geometric features of interest (e.g., local curvature, principal deformations). We demonstrate both the quality of the segmentation produced by our method and the richness of the accessible geometric features through an extensive set of experiments conducted on multiple real and simulated datasets, including objects with challenging geometries.

bioinformatics↗

bioRxiv2024

Deep learning detection of dynamic exocytosis events in fluorescence TIRF microscopy

Segmentation and detection of biological objects in fluorescence microscopy is of paramount importance in cell imaging. Deep learning approaches have recently shown promise to advance, automatize and accelerate analysis. However, most of the interest has been given to the segmentation of static objects of 2D/3D images whereas the segmentation of dynamic processes obtained from time-lapse acquisitions has been less explored. Here we adapted DeepFinder, a U-net originally designed for 3D noisy cryo-electron tomography (cryo-ET) data, for the detection of rare dynamic exocytosis events (termed ExoDeepFinder) observed in temporal series of 2D Total Internal Reflection Fluorescent Microscopy (TIRFM) images. ExoDeepFinder achieved good absolute performances with a relatively small training dataset of 60 cells/[~]12000 events. We rigorously compared deep learning performances with unsupervised conventional methods from the literature. ExoDeepFinder outcompeted the tested methods, but also exhibited a greater plasticity to the experimental conditions when tested under drug treatments and after changes in cell line or imaged reporter. This robustness to unseen experimental conditions did not require re-training demonstrating generalization capability of ExoDeepFinder. ExoDeepFinder, as well as the annotated training datasets, were made transparent and available through an open-source software as well as a Napari plugin and can directly be applied to custom user data. The apparent plasticity and performances of ExoDeepFinder to detect dynamic events open new opportunities for future deep-learning guided analysis of dynamic processes in live-cell imaging.

bioinformatics↗

bioRxiv2024

Template Learning: Deep Learning with Domain Randomization for Particle Picking in Cryo-Electron Tomography

Cryo-electron tomography (cryo-ET) enables the three-dimensional visualization of biomolecules and cellular components in their near-native state. Particle picking, a crucial step in cryo-ET data analysis, is traditionally performed by template matching--a method utilizing cross-correlations with available biomolecular templates. Despite the effectiveness of recent deep learning-based particle picking approaches, their dependence on initial data annotation datasets for supervised training remains a significant limitation. Here, we propose a technique that combines the accuracy of deep learning particle identification with the convenience of the model training on biomolecular templates enabled through a tailored domain randomization approach. Our technique, named Template Learning, automates the simulation of training datasets, incorporating considerations for molecular crowding, structural variabilities, and data acquisition variations. This reduces or even eliminates the dependence of supervised deep learning on annotated experimental datasets. We demonstrate that models trained on simulated datasets, optionally fine-tuned on experimental datasets, outperform those exclusively trained on experimental datasets. Also, we illustrate that Template Learning used as an alternative to template matching, can offer higher precision and better orientational isotropy, especially for picking small non-spherical particles. Template Learning software is open-source, Python-based, and GPU and CPU parallelized.

molecular biology↗

bioRxiv2023

Ensembling Unets, sparse representation and low dimensional visualization for rare chromosomal aberration detection in light microscopy images

AO_SCPLOWBSTRACTC_SCPLOWIn biological dosimetry, a radiation dose is estimated using the average number of chromosomal aberration per peripheral blood lymphocytes. To achieve an adequate precision in the estimation of this average, hundreds of cells must be analyzed in 2D microscopy images. Currently, this analysis is performed manually, as conventional computer vision techniques struggle with the wide variety of shapes showcased by chromosomes. The false discovery rate of current automated detection systems is high and variable, depending on small variations in data quality (chromosome spread, illumination variations ...), which makes using it in a fully automated fashion impossible. Automating chromosomal aberration is needed to reduce diagnosis time. Furthermore, an automated system can process more images, which improves confidence intervals around the estimated radiation dose. We build an object detection model to automate chromosomal aberration detection using recent advances in deep convolutional neural networks and statistical learning. We formulated the problem of rare aberration detection as a heatmap regression problem requiring the minimization of a sparsity-promoting loss to reduce the false alarm rate. Our Unet-based approach is analoguous to a one-stage object detector, and keeps the number of hyperparameters to a minimum. Finally, we demonstrate large performance improvements using an ensemble of checkpoints collected during a single run of training. A PCA-based strategy is used to provide cues for interpretation of our deep neural network-based model. The methodology is demonstrated on real, large, and challenging datasets depicting rare chromosomal aberrations and is favorably compared to a reference dosimetry technique.

bioinformatics↗

bioRxiv2023

DeepCristae, a CNN for the restoration of mitochondria cristae in live microscopy images

Mitochondria play an essential role in the life cycle of eukaryotic cells. However, we still dont know how their ultrastructure, like the cristae of the inner membrane, dynamically evolves to regulate these fundamental functions, in response to external conditions or during interaction with other cell components. Although high-resolution fluorescent microscopy coupled with recently developed innovative probes can reveal this structural organization, their long-term, fast and live 3D imaging remains challenging. To address this problem, we have developed a convolutional neural network, called DeepCristae, to restore mitochondria cristae in low spatial resolution microscopy images. Our network is trained from 2D STED images using a novel loss specifically designed for cristae restoration. To efficiently increase the size of the training set, we also developed a random image patch sampling centered on mitochondrial areas. To evaluate DeepCristae, quantitative assessments were carried out using metrics we derived by focusing on the mitochondria and cristae pixels rather than on the whole image as usual. Depending on the conditions of use indicated, DeepCristae works well on broad microscopy modalities (Stimulated Emission Depletion (STED), Live-SR, AiryScan and Lattice Light Sheet Microscopy). It is ultimately applied in the context of mitochondrial network dynamics during interaction with endo/lysosome membranes.

bioinformatics↗

bioRxiv2022

STracking: a free and open-source python library for particle tracking and analysis

SummaryAnalysis of intra and extra cellular dynamic like vesicles transport involves particle tracking algorithms. Designing a particle tracking pipeline is a routine but tedious task. Therefore, particle dynamics analysis is often performed by combining several pieces of software (filtering, detection, tracking...) requiring a lot of manual operations, and therefore leading to poorly reproducible results. A good synergy between particle detector with a tracker is of paramount importance. In addition, a user-friendly interface to control the quality of estimated trajectories is necessary. To address these issues, we developed STracking a python library that allows to combine algorithms into standardized particle tracking pipelines. Availability and ImplementationSTracking is available as a python library using "pip install" and the source code is publicly available on GitHub (https://github.com/sylvainprigent/stracking). A graphical interface is available using two napari plugins: napari-stracking and napari-tracks-reader. These napari plugins can be installed via the napari plugins menu or using "pip install". The napari plugin source codes are available on GitHub (https://github.com/sylvainprigent/napari-tracks-reader, https://github.com/sylvainprigent/napari-stracking). Contactsylvain.prigent@inria.fr, cesar-augusto.valades-cruz@curie.fr

bioinformatics↗

bioRxiv2022

Towards unsupervised classification of macromolecular complexes in cryo electron tomography: challenges and opportunities

Background and ObjectivesCryo electron tomography visualizes native cells at nanometer resolution, but analysis is challenged by noise and artifacts. Recently, supervised deep learning methods have been applied to decipher the 3D spatial distribution of macromolecules. However, in order to discover unknown objects, unsupervised classification techniques are necessary. In this paper, we provide an overview of unsupervised deep learning techniques, discuss the challenges to analyze cryo-ET data, and provide a proof-of-concept on real data. MethodsWe propose an unsupervised sub-tomogram classification method based on transfer learning. We use a deep neural network to learn a clustering friendly representation able to capture 3D shapes in the presence of noise and artifacts. This representation is learned here from a synthetic data set. ResultsWe show that when applying k-means clustering given a learning-based representation, it becomes possible to satisfyingly classify real sub-tomograms according to structural similarity. It is worth noting that no manual annotation is used for performing classification. ConclusionsWe describe the advantages and limitations of our proof-of-concept and raise several perspectives for improving classification performance.

bioinformatics↗

bioRxiv2022

SPITFIR(e): A supermaneuverable algorithm for restoring 2D-3D fluorescence images and videos,and background subtraction

While fluorescent microscopy imaging has become the spearhead of modern biology as it is able to generate long-term videos depicting 4D nanoscale cell behaviors, it is still limited by the optical aberrations and the photon budget available in the specimen and to some extend to photo-toxicity. A direct consequence is the necessity to develop flexible and "off-road" algorithms in order to recover structural details and improve spatial resolution, which is critical when pushing the illumination to the low levels in order to limit photo-damages. Moreover, as the processing of very large temporal series of images considerably slows down the analysis, special attention must be paid to the feasibility and scalability of the developed restoration algorithms. To address these specifications, we present a very flexible method designed to restore 2D-3D+Time fluorescent images and subtract undesirable out-of-focus background. We assume that the images are sparse and piece-wise smooth, and are corrupted by mixed Poisson-Gaussian noise. To recover the unknown image, we consider a novel convex and non-quadratic regularizer Sparse Hessian Variation) defined as the mixed norms which gathers image intensity and spatial second-order derivatives. This resulting restoration algorithm named SPITFIR(e) (SParse fIT for Fluorescence Image Restoration) utilizes the primal-dual optimization principle for energy minimization and can be used to process large images acquired with varied fluorescence microscopy modalities. It is nearly parameter-free as the practitioner needs only to specify the amount of desired sparsity (weak, moderate, high). Experimental results in lattice light sheet, stimulated emission depletion, multifocus microscopy, spinning disk confocal, and wide-field microscopy demonstrate the generic ability of the SPITFIR(e) algorithm to efficiently reduce noise and blur, and to subtract undesirable fluorescent background, while avoiding the emergence of deconvolution artifacts.

bioinformatics↗

bioRxiv2021

BioImageIT: Open-source framework for integration of image data-management with analysis

Open science and FAIR principles have become major topics in the field of bioimaging. This is due to both new data acquisition technologies that generate large datasets, and new analysis approaches that automate data mining with high accuracy. Nevertheless, data are rarely shared and rigorously annotated because it requires a lot of manual and tedious management tasks and software packaging. We present BioImageIT, an open-source framework for integrating data management according to FAIR principles with data processing.

bioinformatics↗

bioRxiv2021

High-resolution reconstruction and deconvolution of array detector images

Array detector allows a resolution gain for confocal microscopy by combining images sensed by a set of photomultipliers tubes (or sub-detectors). Several methods have been proposed to reconstruct a high resolution image by linearly combining sub-detector images. To overcome the limitations of these techniques, we propose a new reconstruction method that takes the full stack of spatially reassigned detector signals as input. We show on both calibration slides and real data that our deconvolution method allows to achieve a better reconstruction performance in terms of resolution, image contrast, and spatial intensity homogeneity. The tested algorithms are available in an open source software.

bioinformatics↗

bioRxiv2020

Quantitative dSTORM super-resolution microscopy localizes Aurora kinase A/AURKA in the mitochondrial matrix

Mitochondria are dynamic organelles playing essential metabolic and signaling functions in cells. Their ultrastructure has largely been investigated with electron microscopy (EM) techniques, which provided a wide range of information on how mitochondria acquire a tissue-specific shape, how they change during development, and how they are altered in disease conditions. However, quantifying protein-protein proximities using EM is extremely challenging. Super-resolution microscopy techniques as direct stochastic optical reconstruction microscopy (dSTORM) now provide a fluorescent-based alternative to EM with a higher quantitative throughput. Recently, super-resolution microscopy approaches including dSTORM led to valuable advances in our knowledge of mitochondrial ultrastructure, and in linking it with new insights in organelle functions. Nevertheless, dSTORM is currently used to image integral mitochondrial proteins only, and there is little or no information on proteins transiently present at this compartment. The cancer-related Aurora kinase A/AURKA is a protein localized at various subcellular locations, including mitochondria. After performing dSTORM, we here use the Geo-coPositioning System (GcoPS) image analysis method to quantify the degree of colocalization of AURKA with compartment-specific mitochondrial markers. We show that two-color dSTORM provides sufficient spatial resolution to visualize AURKA in the mitochondrial matrix. We conclude by demonstrating that optimizing fixation procedures is a key step to follow AURKA in the matrix. In this light, we show that a methanol-based fixation leads to a better detection of the matrix pool of AURKA than an aldehyde-based fixation. Our results indicate that dSTORM coupled to GcoPS colocalization analysis is a suitable approach to explore the compartmentalization of non-integral mitochondrial proteins as AURKA, in a qualitative and quantitative manner. This method also opens up the possibility of analyzing the proximity between AURKA and its multiple mitochondrial partners with exquisite spatial resolution, thereby allowing novel insights into the mitochondrial functions controlled by AURKA.

cell biology↗

bioRxiv2020

Deep Learning Improves Macromolecules Localization and Identification in 3D Cellular Cryo-Electron Tomograms

Cryo-electron tomography (cryo-ET) visualizes the 3D spatial distribution of macromolecules at nanometer resolution inside native cells. While this label-free cryogenic imaging technology produces data containing rich structural information, automated identification of macromolecules inside cellular tomograms is challenged by noise and reconstruction artifacts, as well as the presence of many molecular species in the crowded volumes. Here, we present a computational procedure that uses artificial neural networks to simultaneously localize with a multi-class strategy several macromolecular species in cellular cryo-electron tomograms. Once trained, the inference stage of DeepFinder is significantly faster than template matching, and performs better than other competitive deep learning methods at identifying macromolecules of various sizes in both synthetic and experimental datasets. On cellular cryo-ET data, DeepFinder localized membrane-bound and cytosolic ribosomes (~3.2 MDa), Rubisco (~540 kDa soluble complex), and photosystem II (~550 kDa membrane complex) with comparable accuracy to expert-supervised ground truth annotations. Furthermore, we show that DeepFinder is flexible and can be combined with template matching to localize the missing macromolecules not found by one or the other method. The DeepFinder algorithm is therefore very promising for the semi-automated analysis of a wide range of molecular targets in cellular tomograms, including macromolecules with weights of 500-600 kDa and membrane proteins.

bioinformatics↗

bioRxiv2019

Two dynamical behaviours of the microtubules at cell cortex reveal pulling and pushing forces that position the spindle in C. elegans embryo.

In the Caenorhabditis elegans zygote, astral microtubules generate forces, pushing against and pulling from the cell periphery. They are essential to position the mitotic spindle. By measuring the dynamics of astral microtubules at the cortex, we revealed the presence of two populations, residing there for 0.4 s and 1.8 s, which correspond to the pulling and pushing events, respectively. Such an experiment offers a unique opportunity to monitor both forces that position the spindle under physiological conditions and study their variations along the anteroposterior axis (space) and the mitotic progression (time). By investigating pulling-force-generating events at the microscopic level, we showed that an anteroposterior asymmetry in dynein on-rate - encoding pulling-force imbalance - is sufficient to cause posterior spindle displacement. The regulation by spindle position - reflecting the number of microtubule contacts in the posterior-most region - reinforces this imbalance only in late-anaphase. Furthermore, we exhibited the first direct proof that the force-generator increasing persistence to pull (processivity) accounts for the temporal control of pulling force throughout mitosis. We thus propose a three-fold control of pulling force, by the polarity, spindle position and mitotic progression. Focusing on pushing force, we discovered a correlation between its density and the stability of the spindle position during metaphase, which strongly suggests that the pushing force contributes to maintaining the spindle at the cell centre. This force remains constant and symmetric along the anteroposterior axis during the division. The pulling one increases in intensity and becomes dominant at anaphase. In conclusion, the two-population study enabled us to decipher the complex regulation of the spindle positioning during cell division.

biophysics↗

bioRxiv2019

Ends and middle: global force balance determines septum location in fission yeast

The fission yeast cell is shaped as a very regular cylinder ending by hemi-spheres at both cell ends. Its conserved phenotypes are often used as read-outs for classifying interacting genes and protein networks. Using Pascal and Young-Laplace laws, we proposed a framework where scaling arguments predicted shapes. Here we probed quantitatively one of these relations which predicts that the division site would be located closer to the cell end with the larger radius of curvature. By combining genetics and quantitative imaging, we tested experimentally whether altered shapes of cell end correlate with a displaced division site, leading to asymmetric cell division. Our results show that the division site position depends on the radii of curvatures of both ends. This new geometrical mechanism for the proper division plane positioning could be essential to achieve even partitioning of cellular material at each cell division.

cell biology↗