MRI Biomarkers of the Pelvis for Opportunistic Screening

At the NIH Clinical Center, the IBCAD lab develops automated tools for MRI to measure the biomarkers of various diseases, such as Prostate/Endometrial cancer and Anasarca. One area of interest is the pelvis where there are multiple structures, such as organs (bladder, prostate/uterus, rectum), skeleton (hip, femur, sacrum), and muscles (glutes, iliopsoas). Delineating these regions is useful for the following clinical applications:

1. Opportunistic screening – monitoring degenerative change of pelvic structures in patients over time
2. Tracking patient recovery – using volumetric measurements of muscle and fat for tracking changes in patients with cancer
3. Dose planning in radiotherapy – irradiating only organs at risk while protecting normal tissue

It is tedious and labor-intensive to manually delineate pelvic structures, and we want to use automated tools to identify these structures in T1- and T2-weighted MRI.

The candidate for this project will perform the following duties:

1. Investigate existing deep learning models for segmenting multiple pelvic structures on public and internal MRI datasets
2. Analyze and summarize results from experiments, visualize findings, and perform basic statistical analysis
3. Prepare a manuscript for a scientific journal

The candidate’s efforts will contribute to the development of a first-of-its-kind AI tool that can improve patient outcomes.