Genedata Screener "Early Access Program" リリース：
Genedata pioneers the use of artificial intelligence for classifying high content images
September 12, 2017
Genedata, a leading provider of advanced software solutions for R&D, today launched its Early Access Program to Deep Learning for High Content Screening (HCS) image analysis. The proprietary and trailblazing software will enable pharmaceutical R&D organizations to reduce the time it takes – by more than half – to accurately classify HCS images and detect new phenotypes. Developed over the last few years in close collaboration with leading academic institutions and pharmaceutical R&D organizations, the deep learning technology uses convolutional neural networks (CNNs) to automatically analyze microscopy images in high throughput. Simplifying image analysis workflows, the solution significantly accelerates image analysis, provides high-quality classification results, broadens HCS applications, and enables the discovery of new phenotypes that are difficult to detect with classical analysis approaches. With the Early Access Program, Genedata provides the opportunity to experience how deep learning can be applied to a company’s HCS, and explore its cost- and time-saving benefits.
Deep Learning Reduces HCS Image Analysis Time from Months to Days
Using a variety of real-world HCS datasets and input from its industry partners, Genedata has developed and fine-tuned its deep learning-based classification technology. Findings have been extremely positive with regards to the quality of results and total analysis time. In recent case studies, deep learning allowed for a quick and unbiased exploration of the phenotypic space and easily captured complex phenotypes. Deep learning expert Dr. Oliver Dürr from the Zurich University of applied sciences confirms, “With deep learning-based CNNs, we have been able to reduce the classification time by a large margin compared to classical methods while obtaining results quality equal to human experts.”
Genedata focuses on streamlining and accelerating image analysis. Historically, the HCS image analysis setup was a manual and time-consuming process, which often took several months for elaborate phenotypic screens. The Genedata solution accelerates this laborious and protracted process - even for complex screens. The innovative Genedata approach also eliminates the need to adjust analysis protocols and settings, a typically costly process. Unlike classical image analysis, which is usually based on a priori phenotypic knowledge, the Genedata deep learning solution offers unbiased image analysis as the implemented algorithm only searches on what separates images from each other – enabling the discovery of new phenotypes.
Using the Genedata deep learning solution for HCS image analysis creates new research efficiencies that:
- Shorten Projects by replacing classical image analysis development with a simpler workflow, which also shortens image analysis runtime.
- Enable New Phenotypic Assays by detecting unexpected phenotypes without months of fine-tuning the analysis.
- Scale Up HCS to empower scientists, not trained to develop classical image analysis protocols, to independently perform HCS; reduces image analysis development time; increases throughput; and lowers costs.
“Genedata is revolutionizing HCS image analysis with deep learning technology,” says Dr. Othmar Pfannes, CEO of Genedata. “The Early Access Program gives interested partners a cost-effective way to experience the immense value of deep learning and become early adopters of the technology. We are committed to further development of practical deep learning applications in life sciences through our valued industry collaborations.”
Early Access Program
The Early Access Program to Deep Learning is core to the Genedata Screener® solution portfolio. The program is open to Genedata customers and industry and academia partners. For more information on the Genedata deep learning technology for HCS image analysis, email your inquiries to deep-learning(at)genedata.com.
Genedata, among the many industry thought leaders advancing the science of deep learning as it applies to HCS, will be presenting at the following conferences.
SBI2 – San Diego, CA – September 13-15
Presentation: “A Comprehensive Workflow Enables the Practical Use of Deep Learning for Pharma-relevant Research Applications”
Poster: “A Deep Learning Workflow for Quickly Establishing and Performing Image Analysis in Phenotypic Assays”
SLAS HCS – Madrid, Spain – September 19-10
Presentation: “Deep Learning for HCS: A Production Workflow for CNN-based Image Analysis”
Poster: “An Innovative Workflow for Practical Use of Deep Learning in Image-based Research”
ELRIG Drug Discovery – Liverpool, UK – October 2-4