Artificial Intelligence in Medical Imaging for novel Cancer User Support
The research aim of AMICUS is to develop and utilise technology for distributed deep learning on medical images in the Netherlands and beyond using the Personal Health Train approach and to show in compelling examples its value to cancer care organisations and patients.
Medical imaging is the cornerstone of screening, detection, diagnosis, staging, treatment and follow-up in almost every cancer patient and without a doubt is Big Data. However the processing and interpretation of images is still mainly a human task. With AMICUS we will develop technology that supports radiologists & oncologists and other stakeholders including the patient in extracting relevant information from images.
The technology used for extracting information from images in AMICUS is deep learning, a form of artificial intelligence which has proven to be a breakthrough in the field, but which requires large volumes of imaging data to be successful. However, getting access to imaging data is a problem as it is dispersed across hospitals and is very privacy sensitive.
In AMICUS a technology will be developed that allows deep learning from these distributed imaging datasets without the need for these data to leave the hospital. For this, the previously developed Personal Health Train approach will be leveraged which allows privacy-preserving learning from distributed FAIR (Findable Accessible Interoperable and Reusable) clinical data and which AMICUS will extend with FAIR imaging data and deep learning.
The work plan of AMICUS consists of four work packages. WP1 and WP2 will focus on developing technology to make imaging and related data FAIR and to perform distributed deep learning in a way which requires only minimal computing resources at each hospital. WP3 and WP4 will focus on creating and demonstrating the value of AMICUS technology by implementing compelling solutions for care organisations to improve cancer care processes and for patients to improve cancer outcomes.
