Hands-on with AI in Radiology

Artificial intelligence, machine learning, and deep learning: they dominate the news. Radiology is considered a good target for automation by intelligent software. Over 100 companies were presenting AI products at RSNA 2017, a number much higher than ever before. On the other hand, the number of AI products available today to radiologists is still small, and most radiologists do not use AI products routinely at all. The goal of this day is to let radiologists experience AI software hands-on, behind a workstation where you can try software out for yourself. Introduction and help by industry representatives and radiologists experienced with the software is provided. In several blocks you will have the opportunity to test yourself against the machine. The focus is on tools that are already in use in some clinics and could be used by the practicing radiologist today. The day starts with a plenary lecture introducing how AI works, how it can be integrated in the radiology workflow, and what products are available. We end with a plenary outlook and interactive discussion on the future of AI in radiology.

When: Saturday April 21, 2018

Where: Radboudumc, Nijmegen, The Netherlands

Accreditation by the NvvR has been requested. Lectures and discussion will be in Dutch, some of the presentations of software in English


  • EuSoMII members:150 EUR
  • Non-members: 180 EUR (with free EUSoMII membership)
  • Residents: 150 EUR (with free EUSoMII membership)

Register: follow this link.


  • 09.00 Coffee, registration
  • 09.30 Lecture “Artificial Intelligence in de radiologie” Bram van Ginneken, Erik Ranschaert
  • 10.15 Hands-on 1: Hand radiographs. Teacher: Dr. Willemijn Klein (Radboudumc). Visiana (Hørsholm, DK). Location: Room 35.
  • 11.00 Hands-on 2: Mammography. Teachers: Dr. Ritse Mann, Dr. Jan van Zelst (Radboudumc) ScreenPoint (Nijmegen, NL). Location: Brakkesteinkamer.
  • 11.45 Hands-on 3: Chest CT. Teacher: Dr. Colin Jacobs. MeVis (Bremen, DE), Aidence (Amsterdam, NL). Location: Room 32 & Lounge area.
  • 12.30 Lunch
  • 13.30 Hands-on 4: Neuro CT & MR. Teacher: Dr. Ewoud Smit. icometrix (Leuven, BE), Brainomix (Oxford, UK). Locations: Room 23 and Room 25.
  • 14.15 Hands-on 5: Chest x-ray. Teachers: Dr. Steven Schalekamp, Dr. Maarten de Rooij. Riverain (Miamisburg, OH). Location: reading room.
  • 15.00 Hands-on 6: Radiomics: Neuro, liver, bone. Quantib (Rotterdam, NL), Quibim (Valencia, ES). Location: Annakamer.
  • 15.45 Lecture and debate: “De toekomst van de radiologie” Erik Ranschaert, Mathias Prokop, Tim Leiner
  • 16.30 Drinks and snacks

There will be 12 participants per hand-on block. Participants rotate over the blocks, so the order varies individually. You will receive a personal schedule when you pick up your badge at the regisatration desk. The blocks will be held in various rooms at the Department of Radiology and Nuclear Medicine. The plenary lectures and lunch will be held in the Tuinzaal of Radboudumc.


During the Hands-on AI day, you can try out products from nine companies from seven countries. Their contributions are described below.


The Danish company Visiana introduced the BoneXpert software for automated bone age determination in 2009, and this CE-marked medical device has now been sold to over 110 clinics and hospitals across Europe. There are more than 15 publications with validation studies of BoneXpert. The precision – or repeatability – is 0.18 years for the software (on repeated X-rays), versus 0.58 years for human rating (rating of the same X-rays by 12 raters). The consistency of the automated method is one of its main assets. Other advantages are that it saves precious radiologist time and that it delivers the result into your PACS system immediately after the X-ray has been recorded.

BoneXpert also measures the cortical thicknesses in the metacarpals and expressed this as a Bone Health Index, including a Z-score, and this works for adults as well.


ScreenPoint Medical from Nijmegen develops Deep Learning and image analysis technology for automated reading of mammograms and digital breast tomosynthesis. We exploit the latest methods in the rapidly evolving field of machine learning, combining these with very large well curated digital image databases and a thorough understanding of the physics of mammogram image formation and the practicalities of the clinical deployment of mammographic image analysis. Our CE-marked software, Transpara™, is an image analytics suite which has been shown to improve mammography reading. It provides applications to improve mammography reading accuracy and increases the radiologist's confidence in their assessment. Furthermore, Transpara™ can automatically pre-screen mammograms to identify, with high confidence, the cases that most likely have no cancer.


MeVis Medical Solutions AG, located in Bremen, Germany, focuses on providing software solutions for image-based medical diagnosis and therapy. In close collaboration with research partners in Germany and the Netherlands, MeVis has developed the comprehensive lung cancer screening software Veolity.

With its optimized workflow, solid pulmonary nodule CAD and its advanced nodule segmentation, volumetry and follow-up registration algorithms, Veolity gives radiologists worldwide the right tools to efficiently review chest CTs in high throughput environments such as screening programs.

Veolity is CE labeled and FDA-cleared, and is already being used in large screening programs and studies in the U.S., Europe, Asia, and Australia.


Amsterdam based startup Aidence develops cutting edge AI diagnostics software based on the latest Deep Learning technology. Its first CE marked solution, Veye Chest, assists radiologists in detecting, tracking and reporting of pulmonary nodules on CT chest. Veye delivers superior accuracy with 90% sensitivity at an average of 1 false-positive per scan. Veye Chest is seamlessly integrated into your reporting workflow, which Aidence believes is key to successful adoption of AI. Veye Chest can be used in screening settings and during routine clinical practice.


Brainomix, launched as a spin-out from the University of Oxford, UK, develops imaging software to support clinical decision making, for neurological and cerebrovascular diseases. The company's e-ASPECTS software, a CE-marked, class IIa medical device, automatically implements the ASPECTS scoring system on acute ischemic stroke CT scans. The software's Artificial Intelligence (AI) amplifies the physician's ability to interpret the CT scans enabling rapid and consistent quantification of early ischemic brain damage. Physicians can review e-ASPECTS results, anywhere, less than one minute after the CT scan is performed. e-ASPECTS is in use in hospitals across Europe and worldwide, supporting the fast, consistent diagnosis of patients that would benefit from thrombolysis or endovascular treatment.


icometrix, founded as a spin-off of the universities and university hospitals of Leuven and Antwerp, provides radiologists with standardized measurements on brain scans in order to improve personalized care of patients with neurological disorders. Its CE-marked and FDA-cleared cloud software, icobrain, quantifies brain atrophy and lesion evolution in MS patients, lobar and hippocampal atrophy for dementia, and brain atrophy and axonal injuries in traumatic brain injury patients.


Riverain Technologies from Miamisburg, Ohio, provides software tools to aid the clinician in early detection of lung disease. Riverain’s ClearRead applications for chest X-ray remove ribs and clavicles from a standard chest X-ray; identify regions of interest such as pulmonary nodules; accentuate lines and tubes on portable chest X-ray images; and provide a subtraction image to enhance interval change between current and prior chest X-ray exams. AllClearRead applications are FDA and CE certified and are in active use in hospitals worldwide.


Quantib, a spin-off of Erasmus Medical Center Rotterdam, is building a complete neurosuite for the radiologist. Currently available CE marked products cover brain atrophy measurement including lobe segmentation and the detection and longitudinal tracking of white matter hyperintensities, combined with normative reference curves for brain volume, enabling the physician to compare the evolution of their patient's results to that of a healthy population. Additionally, a radiomics based product is in development in collaboration with the Biomedical Imaging Group Rotterdam. This software will aid in the detection, classification and quantification of brain tumors, supporting faster and more differentiated diagnosis.


QUIBIM is an innovative spin-off company of La Fe Research Institute in Valencia, created as an initiative from radiologists and biomedical engineers. We combine Artificial Intelligence algorithms, based on Deep Learning techniques, with Imaging Biomarkers analysis to objectively measure disease or treatment-derived alterations in organs and tissues, offering additional quantitative information to the eye of radiologists.

QUIBIM integrates its structured reports directly into PACS or through our platform QUIBIM Precision®, a cloud-based platform for image processing.

All the QUIBIM biomarkers have been validated in clinical settings and provide useful parametric information for diagnosis and monitoring of high socioeconomic impact diseases including osteoporosis, cancer, dementia, COPD and diffuse liver diseases.

We provide a comprehensive range of imaging biomarkers to professionals who want to distinguish themselves by choosing to offer an added value to diagnostic radiology, research, and clinical trials.