Diabetes Prediction

Open Project: Diabetes Prediction

Type: Master Thesis

Contact Person: Sebastian Pölsterl, Christian Wachinger

 
Type 2 diabetes (T2D) is the most common form of diabetes. In this form of diabetes, insulin is less efficiently used than normal, leading to abnormally high blood sugar levels. In healthy patients, insulin decreases blood glucose concentration by stimulating intake of glucose by muscle, fat, and liver cells, where it is used for energy. Patients with T2D have insulin resistance and more and more insulin needs to be produced to keep blood sugar levels in the normal range. Eventually, the pancreas is unable to secrete enough insulin to keep blood glucose levels normal. Many complications are associated with the disease, including an increased risk for cardiovascular disease. Diabetes is a leading cause of kidney failure, lower limb amputations, and retinopathy, which can result in blindness.

Excess abdominal fat accumulation is a major risk factor for insulin resistance. Therefore, detecting structural changes, such as lipid deposition in target abdominal organs, could be an effective approach for diabetes screening. In this project, we will use deep learning techniques to identify structural changes in whole-body magnetic resonance images (MRI) that are associated with T2D.

Requirements:
   – Very good programming skills in Python.
   – Experience in machine learning.
   – Experience in medical image analysis and deep learning (TensorFlow, PyTorch, etc.) is desirable.