Consultation regarding the use of deep learning for specific research questions
We offer advice for Radboudumc employees to show the potential benefit deep learning techniques can have for your research. Many types of data gathered in the Radboudumc, such as images, time series, etc. lend themselves to be automatically interpreted by deep learning algorithms. This can save researchers time, eliminates any subjectivity and can provide researchers valuable insights that could otherwise be missed. Are you interested in the potential of deep learning for your research? Contact us.
Assistance with data preparation and annotation
A crucial ingredient for deep learning techniques to work, is a well prepared data set. Most deep learning algorithms learn by example. This means that your data set should be annotated with the labels that you want your deep learning algorithm to learn. The RTC Deep Learning can annotate data for you, or assist you in making good annotations yourself. Another important aspect is to standardize your data, such that all samples are in the same format, value range etc. The RTC Deep Learning has broad expertise in the best ways to achieve such standardization.
Development and training of deep learning algorithms for detection, classification and segmentation problems
With a well prepared data set, the RTC Deep Learning can develop and train deep learning algorithms for tasks such as detection, classification and segmentation of pathologies or anatomical structures. For examples of the diverse range of tasks we can automate using deep learning algorithms, have a look at our 'Projects' section.
Development of web-based applications for data processing and viewing
The algorithms we build can be easily accessed through our web-based platform www.grand-chellenge.org. Here you can easily upload a sample of your data, which will be automatically processed, resulting in an output that can be viewed through our CIRRUS Core web viewer.
GPU computing solutions
Training deep learning algorithms requires a lot of GPU compute power. Our RTC Deep Learning has state-of-the-art hardware with the best GPU's currently available. Recently a new NVIDIA DGX™ A100 was added to boost our computing power. This is the largest and most advanced available AI hardware system on the market. We are the only research facility in the Netherlands to have direct access to such a machine. We offer the usage of our GPU's through SOL to other researchers within the Radboudumc. Contact us for more information about using our hardware for your research.
To share our knowledge and expertise on artificial intelligence and spefically deep learning for healthcare we organize workshops and courses regularly. We contribute to the Intelligent Systems in Medical Imaging course for students. Co-organize the AI for Health course which offers employees of the Radboudumc a deep dive in the healthcare applications of AI. Finally, we offer workshops and training opportunities for researchers in the Radboudumc that would like to make use of our software or hardware solutions.
Support MSc students and visiting researchers
Msc students and visiting researchers that are using AI in their research can receive support from the RTC Deep Learning. This can offered by usage of our hardware, access to our code base and through expert advice on how to tackle a specific AI problem.