Artificial intelligence in the big-data era: an opportunity for FAIR microscopy
David Poger1, Lisa Yen1, Filip Braet2,3 1Microscopy Australia, Sydney, NSW, Australia2Australian Centre for Microscopy & Microanalysis, The University of Sydney, Sydney, NSW, Australia3School of Medical Sciences (Molecular and Cellular Biomedicine), The University of Sydney, Sydney, NSW, Australia
Abstract
The booming development of new hardware and software tools in microscopy has resulted in larger image files acquired at ever-increasing speed and resolution. This big-data revolution brings along numerous challenges in data transfer, computing and management that artificial intelligence (AI) can contribute to addressing. In particular, AI (machine- or deep-learning) is increasingly used in data processing and analysis to extract meaningful information from large datasets, from noise reduction to the facilitation and automation of pattern and shape recognition. In this presentation, I will show that, while the use of AI to assist in image data processing is an efficient and time-effective avenue to automate complex, tedious and time-consuming tasks that would need to be otherwise completed manually, it also presents the unique opportunity to promote consistent and standardised practices through the automation of tasks. For example, using controlled vocabulary in AI-assisted image annotation can favour the adoption of common, community-endorsed, machine-readable metadata standards and interoperability between programs. Machine-readable metadata are essential for seamless discovery, sharing and reusability of data, which is a cornerstone of FAIR data (findable, accessible, interoperable and reusable). In conclusion, the growing use of AI in image data processing and analysis is a chance to put the FAIR principles in practice in microscopy. AI can foster and accelerate the establishment of guidelines and standards for metadata collection for data description and annotation as well as the rationalisation or unification of data formats. This will in turn enhance knowledge discovery, experiment reproducibility and research impact.
Biography
Dr David Poger (https://orcid.org/0000-0001-8794-5688) was awarded his PhD from Joseph Fourier University (Grenoble, France) in 2005. He then moved to Australia where he worked as a Research Fellow at The University of Queensland. Since 2020, David has been the Research Data Manager at Microscopy Australia. He assists microscopy facilities in data management by developing best practices and guiding facilities in their journey to FAIR data. David participates actively in several projects and working groups in Australia and overseas to develop and promote best practices in research data management and facilitate the interaction between microscopy facilities, data managers and IT/eResearch specialists.