Improving diagnosis of cardioembolic stroke with machine learning
By analyzing brain magnetic resonance imaging, cardiac imaging and electrocardiogram using machine learning techniques, we aim to improve the detection of patients with a cardioembolic cause of stroke for targeted prophylaxis of recurrence using anticoagulation.
Furthermore, we work on deep learning models that enable extraction of features predicting imaging readouts of atrial cardiomyopathie from routine ECG (even reduced lead holter ECG). This may help early identification of those at risk of stroke, heart failure and atrial fibrillation.
People involved:
Julian Deseö, Lisa Herzog, Hakim Baazaoui, Prof. Bjoern Menze, Ezequiel de la Rosa (DQBM)
