Grant: Project Grant

Each year 400,000 North Americans suffer sudden cardiac death (SCD), usually from a life threatening heart rhythm. One in five of us will die from SCD, often without any warning and where help is not available. An implantable cardioverter defibrillator (ICD) is a special type of pacemaker that can prevent SCD. However, the current guidelines of who should get an ICD use only one measurement, the "ejection fraction" (EF), to decide if someone should or should not get one. Using this approach only 1 out of 10 patients getting an ICD actually get benefit, and more than 80% of patients who end up suffering SCD are never given one. Therefore, we must find better ways to figure out who should and should not be given an ICD. Our lab has been working for over 10 years to find features on picture of the heart that pick out people at highest risk of SCD. To date we have published 8 features that strongly predict this. We have now made a computer software that can automatically build a 4-dimensional heart model (beating 3D heart) from images taken with an MRI scanner. We believe this can find people at high risk of SCD. To do this we have partnered with the Alberta Machine Intelligence Institute, a group of scientists that use approaches (similar to Google) to solve complex problems. We plan to teach a computer program, using a total of 1,500 MRI studies in patients sent for an ICD, what a "high risk" patient looks like. We will test the ability of this new software to predict if and when a new patient will suffer SCD. If successful, we will make this software available to the community to improve how we deliver ICDs to those patients who need them most. This work could therefore have a major impact by saving a larger number of lives in people with heart disease.