Segmentation and Quality Analysis of OCT Images for Kidney Evaluation

This project offers opportunities for credit-based undergraduate research (e.g., independent study). Due to the multidisciplinary nature of the project, students from any department within the College of Engineering (COE), Computer Science, or other related areas are encouraged to apply.

This project focuses on the segmentation of kidney capsule structures from OCT (Optical Coherence Tomography) images, with the goal of quantitatively analyzing tissue morphology and intensity. These measurements will be used for the assessment of donor kidney quality and clinical decision-making in organ transplantation.

We currently have an extensive OCT dataset (~120 donor kidneys) and are looking for motivated undergraduate students to assist with manual labeling of the capsule. The labeled data will be used to train machine learning models for automated segmentation and quantitative analysis with clinical data. Students will receive training on the annotation tools and deep learning models and will work closely with graduate researchers involved in the project.

Students will gain experience in:

• Working directly with real clinical data
• Medical image annotation using professional imaging software such as ImageJ
• Basic quantitative analysis of morphology and radiomic features
• Fundamentals of deep learning–based segmentation workflows.

Responsibilities may include:

• Manual segmentation of OCT images using labeling software
• Organizing and maintaining labeled datasets
• Assisting with basic quantitative analysis (e.g., thickness calculation)
• (Optional) Participation in the development and validation of machine learning models.

Required Qualifications:

•  Prior experience with Python.

Preferred backgrounds include (Preferred but not required):

• Imaging software such as ImageJ
• Interest in medical imaging or biomedical device applications
• Basic knowledge of anatomy or willingness to learn.

If you are interested in gaining hands-on experience in medical image analysis and contributing to a translational research effort, please feel free to reach out!