The AI4Life consortium collaborated with researchers from the Instituto Gulbenkian de Ciências (IGC) in Portugal and Åbo Akademi University in Finland to develop DL4MicEverywhere, an intuitive interface that enables researchers to train and apply deep learning models on a variety of computing infrastructures, from laptops to high-performance clusters. The study has been published in Nature Methods.
Image Credit: Anatolii Stoiko/Shutterstock.com
Researchers have created a novel platform to enable life scientists to use state-of-the-art deep learning methods for biomedical research. Thanks to DL4MicEverywhere, researchers of any computational background can now analyze microscopy images with advanced artificial intelligence (AI).
The analysis of extensive and intricate microscopy datasets has been revolutionized by deep learning, a branch of artificial intelligence that makes it possible to automatically identify, track, and analyze cells and subcellular structures. Despite these developments, the use of these methods in life sciences research has been constrained by the requirement for computing power and AI knowledge.
DL4MicEverywhere establishes a bridge between AI technological advances and biomedical research. With it, researchers gain access to cutting-edge methods, enabling them to automatically analyze their microscopy data and potentially discover new biological insights.
Ivan Hidalgo-Cenamor, Study First Author and Researcher, Instituto Gulbenkian de Ciências
The new platform will be available as an open-source resource. The researchers think it will enable breakthroughs in various fields, including drug discovery, personalized medicine, and basic cell biology, by reducing the barriers to advanced microscopy image analysis. DL4MicEverywhere introduces multiple significant improvements over the team’s earlier work, ZeroCostDL4Mic.
It encapsulates deep learning workflows in repeatable and shared Docker containers, thus making it easier to train and deploy models across various computational environments. The platform uses an intuitive graphical user interface to increase the number of models available for common microscopy image analysis tasks.
“DL4MicEverywhere aims to democratize AI for microscopy by promoting community contributions and adhering to FAIR principles – making models findable, accessible, interoperable, and reusable,” details Dr. Estibaliz Gómez-de-Mariscal, Researcher, Instituto Gulbenkian de Ciências and Dr. Joanna Pylvänäinen, Researcher, Åbo Akademi University.
The researchers added, “We hope this platform will empower researchers worldwide to harness these powerful techniques in their work, regardless of their resources or expertise. It will allow life scientists without coding experience to use deep learning on large numbers of microscopy images and videos to make discoveries. This will revolutionize how researchers plan their experiments and extract new information from microscopy datasets.”
With significant assistance from the AI4Life consortium, a global team of specialists in computer science, bioimage analysis, and microscopy worked together to develop DL4MicEverywhere. The laboratories of Prof. Guillaume Jacquemet at Åbo Akademi University and Prof. Ricardo Henriques at IGC co-led the project.
Prof. Guillaume Jacquemet and Prof. Ricardo Henriques said, “This work represents an important milestone in making AI more accessible and reusable for the microscopy community. By enabling researchers to share their models and analysis pipelines easily, we can accelerate discoveries and enhance reproducibility in biomedical research. DL4MicEverywhere has the potential to be transformative for the life sciences. It aligns with our vision in AI4Life to develop sustainable AI solutions that empower researchers and drive innovation in healthcare and beyond.”
Journal Reference:
Hidalgo-Cenalmor, I., et al. (2024) DL4MicEverywhere: deep learning for microscopy made flexible, shareable and reproducible. Nature Methods. doi.org/10.1038/s41592-024-02295-6.