Adipose Tissue Segmentation in Children using Deep Neural Networks
Radiological Sciences; Bioengineering
Description of Research Project:
Obese children have larger amounts of subcutaneous and visceral adipose tissue (SAT, VAT) and are at high risk for cardiometabolic disease. SAT and VAT can be quantified using free-breathing magnetic resonance imaging (MRI) in children. However, the reference standard for SAT and VAT analysis requires manual annotation, which depends on expert knowledge and is time consuming. In our lab, we are developing deep neural networks to automatically segment SAT and VAT on MRI and quantify their fat content and volume in children. Our neural networks will only take seconds to accurately segment the desired tissues.
Description of Student Responsibilities:
(1) Create reference manual annotations for adipose tissue on MR images. (2) Perform literature survey, especially in the areas of machine and deep learning tools for adipose tissue segmentation. (3) Train/test neural networks for adipose tissue segmentation and report the performance metrics.