Biology subjects

Rousseau, D.

Publications and source records attributed to Rousseau, D..

4 recordsLinked to original sources

bioRxiv2025

Cellular porosity in dentin exhibits complex network characteristics with spatio-temporal fluctuations

According to the current hydrodynamic theory, teeth sensitivity is mediated by odontoblast cell processes which can be activated by fluid flow in the pericellular space of bulk dentin. To better understand the possible spatial extent of such phenomena, we investigated the topology and connectivity of dentinal porosity of a healthy human tooth. Using confocal fluorescence microscopy, we modeled the porosity as a spatial graph with edges representing dentinal tubules or lateral branches and nodes defining their connections. A large fraction of porosity channels in crown dentin was found to be interconnected, with 47 % of nodes linked in a single component over a millimetric distance from the dentin-enamel junction (DEJ). However, significant differences in network topology were also observed. A sharp transition in connectivity from 83 % to 43 % occurred at 300 {micro}m from the DEJ, which corresponds to an early stage of tooth formation. This was reflected in all graph metrics investigated, in particular the network resilience which dropped by a factor 2. To test the robustness of our observations, an in-depth analysis of potential remaining biases of the graph extraction was conducted. Most graph metrics considered were found to be within a 10 % precision range from a manually annotated ground truth. However, path metrics, which characterize transport properties, proved very sensitive to network defects. Residual errors were classified in 4 topological classes related to fluorescence staining and confocal detection efficiency, instrumental resolution and image processing. Their relative importance was estimated using statistical and physical graph attack simulations in a broad experimental range. Our modeling thus provides a practical framework to estimate the interpretability of calculated graph metrics for a given experimental microscopy setup and image processing pipeline. Overall, this study shows that dentin porosity exhibits typical characteristics of a complex network and quantitatively emphasize the importance of the smallest lateral branches. Our results could be used to model fluid flow more accurately in order to better understand mechanosensing by odontoblasts in dentin.

biophysics↗

bioRxiv2023

GenoDrawing: An autoencoder framework for image prediction from SNP markers

Advancements in genome sequencing have facilitated whole genome characterization of numerous plant species, providing an abundance of genotypic data for genomic analysis. Genomic selection and neural networks, particularly deep learning, have been developed to predict complex traits from dense genotypic data. Autoencoders, a neural network model to extract features from images in an unsupervised manner, has proven to be useful for plant phenotyping. This study introduces an autoencoder framework, GenoDrawing, for predicting and retrieving apple images from a low-depth single nucleotide polymorphism (SNP) array, potentially useful in predicting traits that are difficult to define. GenoDrawing demonstrated proficiency in its task while using a small dataset of shape-related SNPs, and multiple experiments were conducted to evaluate the impact of SNP selection and shape relation. Results indicated that the correct relationship of SNPs with visual traits had a significant impact on the generated images, consistent with biological interpretation. While using significant SNPs is crucial, incorporating additional, unrelated SNPs results in performance degradation for simple NN architectures that cannot easily identify the most important inputs. The proposed GenoDrawing method is a practical framework for exploring genomic prediction in fruit tree phenotyping, particularly beneficial for small to medium breeding companies to predict economically significant heritable traits. Although GenoDrawing has limitations, it sets the groundwork for future research in image prediction from genomic markers. Future studies should focus on using stronger models for image reproduction, SNP information extraction, and improved dataset balance in terms of shape for more precise outcomes.

plant biology↗

bioRxiv2022

Machine learning-based detection of label-free cancer stem-like cell fate

Most imaging methods rely on labelling biological samples in order to provide specific and easily detectable features. However, label-free imaging is a non-invasive and non-toxic alternative that requires accurate image analysis algorithm based on cell morphology. Such analysis has to deal with a high image variability while fewer features are extractable, so far, fast analysis of label-free brightfield microscopy (LFBM) images remains a challenging task. With the development of microfabricated devices during the last decades, high throughput image generation makes it possible to use machine learning-based algorithms in order to analyse LFBM images. Fast algorithms are also crucially needed to analyze high throughput experiments. In this paper, we provide a data-driven study in order to assess the complexity of LFBM time-lapses monitoring isolated cancer stem-like cells (CSCs) fate in non-adherent conditions. We combined for the first time individual cell fate and cell state temporality analysis in a unique algorithm. Several image analysis algorithms of increasing processing capacities were tested: a classical computer vision algorithm (CCVA), a shallow learning-based algorithm (SLBA) and a deep learning-based algorithm (DLBA). We show that our optimized DLBA has by far the best accuracy compared to CCVA and SLBA, is at least as accurate as other state-of-the-art DLBAs while being faster. With this study, we demonstrate that optimizing our DLBA accordingly to the image analysis problem can overall provide better results than pretrained models. Such a fast and accurate DLBA is therefore compatible with the generation of high throughput data and opens the route for on-the-fly analysis of CSC fate from LFBM time-lapses.

bioinformatics↗

bioRxiv2021

Clearing spheroids for 3D fluorescent microscopy:combining safe and soft chemicals with deep convolutional neural network

In life sciences, there are increasing interest in 3D culture models to better reproduce the 3D environment encountered in-vivo. Imaging of such 3D culture models is instrumental for drug discovery, but face several issues before its use becomes widespread. Extensive microscopic investigation of these 3D cell models faces the challenge of light penetration in depth in opaque biological tissues. To overcome this limit, diverse clearing techniques have emerged over the past decades. However, it is not straightforward to choose the best clearing protocols, and assess quantitatively their clearing efficiency. Focusing on spheroids, we propose a combination of fast and cost-effective clearing procedure for such medium-sized samples. A generic method with local contrast metrics and deep convolutional neural network-based segmentation of nuclei is proposed to quantify the efficiency of clearing. We challenged this method by testing the possibility to transfer segmentation knowledge from a clearing protocol to another. The later results support the pertinence of training deep learning algorithms on cleared samples to further use the segmentation pipeline on non-cleared ones. This second step of the protocol gives access to digital clearing possibilities applicable to live and high-throughput optical imaging.

bioinformatics↗