Grant: Project Grant

Stroke, mainly caused by a blood clot blocking a brain artery, is the leading cause of morbidity in the developed world. Removing the clot quickly saves at-risk brain tissue, improving the patient's chance of a good outcome. A current clinical method for removing blood clots is giving a drug that dissolves clot (clot-buster); however, this method only works 15-25% of the time for larger clots, and may cause bleeding into the brain in some patients. A more recent method of clot removal is to pull it out mechanically using a catheter inserted through an artery in the groin. This technique has a much higher success rate, but is only available at large hospitals and requires a talented surgeon. This procedure is therefore not available in community hospitals. When treating a patient with stroke, a community physician therefore needs to know if the clot in the patient's brain blood vessel dissolves quickly with the clot-busting drug or will the patient need to be sent to a larger hospital where the clot could be pulled out with a catheter? The community physician has to make decisions on treatment and transfer fast as every minute counts with stroke treatment. On the other hand, decisions have to be appropriate and correct, as otherwise some patients may be harmed while others who do not need the catheter-based procedure may be transferred unnecessarily to the bigger hospital. Moreover, the community physician sees a small number of patients with stroke each year and therefore lacks expertise in interpreting brain scans and making these decisions. To address this gap in expertise between large hospitals and community hospitals, we propose to develop an automated imaging-based triaging decision support tool that can help community physicians make clinical decisions on triage of patients with acute ischemic stroke at any time of the day or night quickly and correctly. We will automate many important imaging concepts we have developed until now for this purpose.