for SLAS High-Content Screening Conference 2017
View analytic
Wednesday, September 20 • 9:00am - 11:30am
Session 3: Data Analysis
The analysis of the large amount of data generated high content screening experiments represents a significant challenge and is currently a bottleneck in many small molecules and genetic screening projects.  This section is an exciting opportunity for reviewing the main challenges in interpreting complex high-content screening data, including key informatics and data analysis approaches such as image descriptors computations and classification algorithms.

9:00 - 9:30 
Computational methods for fluorescence microscopy and quantitative bioimaging
Charles Kervrann, Senior Researcher, Inria Rennes - Bretagne AtlantiqueSEPICO Project-Team 

During the past two decades, biological imaging has undergone a revolution in the development of new microscopy techniques that allow visualization of tissues, cells, proteins and macromolecular structures at all levels of resolution. Thanks to recent advances in optics, digital sensors and labeling probes, one can now visualize sub-cellular components and organelles at the scale of a few dozens nanometers to several hundreds of nanometers. As a result, fluorescent microscopy and multimodal imaging has become the workhorse of modern biology. All these technological advances in microscopy, created new challenges for researchers in quantitative image processing and analysis. Therefore, dedicated efforts are necessary to develop and integrate cutting-edge approaches in image processing and optical technologies to push the limits of the instrumentation and to analyze the large amount of data being produced.In this talk, we present image processing methods, mathematical models, and algorithms to build an integrated imaging approach that bridges the resolution gaps between the molecule and the whole cell, in space and time. The presented methods are dedicated to the analysis of proteins in motion inside the cell, with a special focus on Rab protein trafficking observed in time-lapse confocal microscopy or total internal reflection fluorescence microscopy. Nevertheless, the proposed image processing methods and algorithms are flexible in most cases, with a minimal number of control parameters to be tuned. They can be applied to a large range of problems in cell imaging and can be integrated in generic image-based workflows, including for high content screening applications.

9:30 - 10:00 
Big data approaches for computational phenotyping
Thomas Walter, MINES ParisTech 

In High Content Screening (HCS) we dispose of the technological tools to perform imaging experiments at an unprecedented scale and thus to generate extremely large and complex image data sets, that can be readily qualified as “big data”. In this context, methods from computer vision and machine learning have been instrumental to deal with the generated large-scale image data sets. After a short review of computer vision techniques for computational phenotyping for live cell imaging data, I will present recent developments in the field of spatial transcriptomics, where we aim at understanding the spatial aspects of gene expression at a large scale. This poses interesting and challenging questions for the analysis of such data: single RNAs can be automatically detected and their spatial distribution analyzed with newly designed features and machine learning methods. Importantly, as there is little prior knowledge available for this type of data, we have built a virtual cell environment in order to simulate RNA localization patterns inside cells to validate the developed methods. More generally, simulation frameworks are becoming increasingly important for computational phenotyping. Finally, I will present new approaches to analyze and computationally phenotype cells in their tissular context. Segmentation is one of the major bottlenecks in computational phenotyping of histopathology data. Here, I present a new technique for the segmentation and classification of nuclei based on deep learning. This method allows us to analyze large cohorts of patient data with respect to their cellular and tissular phenotypes and to relate these descriptors to clinical variables.

10:00 - 10:30
Presentation Title Forthcoming
Marco Prunotto, Roche

10:30 - 11:00
Presentation Title Forthcoming
Lucas Pelkmans, University of Zurich, Institute of Molecular Life Sciences

11:00 - 11:30
Exploring Toxicity Profiling for Drug Screening using Deep Learning Strategies
Daniel Jimenez, Centro Nacional de Investigaciones Cardiovasculares (CNIC)

Toxicity is a major factor of failure in drug development, causing costly withdrawals of drugs from the market. Efficienct drug discovery could be greatly improved if compounds with cytotoxic characteristics were identified during primary screening campaigns. Although apoptotic & necrotic processes involve dramatic changes in cell morphology, these have never been exploited for systematic and quantitative assessment of cellular toxicity due to lack of methods to identify them in a reproducible manner. The development of high-content imaging platforms has allowed the incorporation of complex toxicity counter-screens. However, this requires expensive and lengthy labelling of cytotoxicity specific reporters. There is thus an urgent need to develop cost-effective methods for toxicity assessment qualified for high throughput screening (HTS). We hypothesize that toxicity could be detected without toxicity specific labelling by using state of the art computer vision methods, thus reducing significantly the cost of drug screens. In this work we develop a framework to determine cell cytotoxicity status as a HTS readout based on the analysis of nuclear fluorescence microscopy images by using deep learning strategies. Multiple cell-based assays including concentration-response curves for drugs with different cytotoxic effects are used for training and validation of the system. Preliminary results support our hypothesis, reporting high correlations between toxicity predictions, dose-response curves, and well-established cytotoxicity reporters; suggesting that the automatic pattern recognition performed by deep neural networks is able to detect differential structural and nuclear-based features related with the cytotoxic state of the cells. Therefore, this framework has the potential to become a new scientific breackthrough for toxicity assessment in drug screening.

avatar for Charles Kervrann

Charles Kervrann

Inria Senior Researcher, Inria Rennes - Bretagne Atlantique
Charles Kervrann received the M.Sc. (1992), the PhD (1995) and the HDR (2010) in Signal Processing and Telecommunications from the University of Rennes 1, France. From 1997 to 2010, he was researcher at the INRA Applied Mathematics and Informatics Department (1997-2003) and he... Read More →
avatar for Thomas Walter

Thomas Walter

Thomas Walter has been working in the field of biomedical image analysis for more than 15 years. His most visible scientific contributions have been in the field of High Content Screening (HCS). He has pioneered methods in the field of cellular phenotyping for live cell imaging d... Read More →

Wednesday September 20, 2017 9:00am - 11:30am
Rosales I Courtyard by Marriot Madrid Princesa Hotel
Feedback form isn't open yet.