Philips brings AI to radiology
December 11, 2018
Philips has introduced an open platform to enable the development and deployment of artificial intelligence (AI) to support radiologists in their clinical and translational research.
AI has the potential to improve patient care and increase the efficiency of care delivery. However, there are problems when it comes to introducing AI into healthcare clinical practice. Health systems are consistently faced with questions on how to collect and prepare high quality data, which methods of training and validating the tools are most appropriate, and how to deploy AI without disruption.
IntelliSpace Discovery 3.0 combines AI and other technologies with knowledge of the clinical and operational context in which they are used – a people-centred approach called adaptive intelligence – to develop integrated methods that adapt to the needs of healthcare providers.
IntelliSpace Discovery is already a proven research platform, used by more than 50 hospitals and academic institutions worldwide, for the development of radiology applications for rendering, segmentation and quantification. For example, it is being used for research and clinical validation and subsequently the tools and applications can be deployed into the radiology workflow on the IntelliSpace portal.
With the introduction of IntelliSpace Discovery 3.0, the platform now provides research applications and tools for radiologists to aggregate, normalise and anonymise data, which can be visualised and annotated to train and validate deep learning algorithms. They can then easily deploy these algorithms as plug-in apps into the research workflow to analyse new datasets and help facilitate clinical research in radiology, oncology, neurology and cardiology.
"Together with our customers we're enabling research in adaptive intelligence with the goal to create solutions that augment healthcare professionals and improve patient care and efficiencies of care delivery, both inside and outside of the hospital," said Jeroen Tas, chief innovation and strategy officer at Philips. "AI is the connective tissue to seamlessly integrate data and technology to enable precision diagnosis."
The building blocks of IntelliSpace Discovery include:
- Front-end applications – Integration with existing clinical infrastructure provides seamless access to vendor agnostic and multi-modality data sets, and research visualisation tools allow data to be prepared and processed for AI training with existing AI tools.
- Study management – Scalable infrastructure contains a vendor neutral research archive for structured and unstructured data, used to upload and process data.
- Machine learning – Research data and deployment platform obtains batch processing in a scalable high-performance computing environment to support iterative development and validation.
- Clinical research – AI assets and capabilities include multi-parametric enablement, segmentation and annotation, tumour quantification and stratification, deep learning networks and classification.
- Research services – Accessibility to Philips R&D for a range of services from consultancy to development.
IntelliSpace Discovery is for research use only and cannot be used for patient diagnosis or treatment selection.