ICH and PHE segmentation

Non-traumatic spontaneous intracerebral hemorrhage (ICH) is the most deadly stroke subtype with a 30-day mortality of 40%. ICH volume is an important predictor of early neurological deterioration and functional outcome and in one-third of patients the hemorrhage expands further, particularly in the first hours. Furthermore, secondary injury due to the development of perihematomal edema (PHE) contributes to disability and mortality. Prevention of early hematoma growth has become an important treatment target, although medical therapies have not been proven effective so far. ICH and PHE volumes are therefore important imaging biomarkers for patient stratification, treatment monitoring and outcome prediction. However, current quantification methods are mainly based on visual estimations or over-simplified assumptions, such as the ABC method, and are laborious, prone to observer variability and inaccurate. In addition, no method exists for PHE quantification in non-contrast CT (NCCT), and no model exists for longitudinal analysis.

The DUTCH ICH Surgery Trial (DIST) investigates the effectiveness of minimally-invasive endoscopy-guided surgery for spontaneous ICH and the potential effect on patients’ functional outcome. In this study, neuro-imaging will be performed at multiple time points to assess, among others, ICH and PHE volumes. DIST is a part of the Collaboration for New Treatments of Acute Stroke (CONTRAST) consortium which is a collaboration of clinical and translational scientists from all academic and large clinical centers who want to act together to improve the treatment of acute stroke.

The algorithm produced in this project has been made available to members of the CONTRAST consortium in the form of a web-based application for the accurate segmentation and measurement of ICH and PHE on NCCT. The algorithm can be used on the Grand Challenge platform. All processing is performed on the platform; there is no specialized hardware required to try out the algorithm. Running the algorithm requires an account for Grand Challenge. If you don't have an account yet, you can register on the website; alternatively, you can log in using a Google account. After registering for a new account, or logging in to an existing Grand Challenge account, you can request access to the algorithm.

Try out the algorithm

People

Karin Klijn

Karin Klijn

Professor

Neurology, Radboudumc

Floris Schreuder

Floris Schreuder

Neurologist

Neurology, Radboudumc

Ajay Patel

Ajay Patel

Coordinator RTC Deep Learning

 Rashindra Manniesing

Rashindra Manniesing

 Anton Meijer

Anton Meijer

Bram van Ginneken

Bram van Ginneken

Professor

Diagnostic Image Analysis Group