Thursday, March 30, 2023

Undergraduate Research Festivals (FURF and SURF) are held in celebration and recognition of undergraduate researchers' contributions to research, scholarly, and creative work at the University of Iowa.

See VPR's Office recap for more on SURF!

This is an incredible opportunity for Iowa students to practice and hone their communication skills,” said Bob Kirby, director of ICRU. “From designing a poster, to giving a short pitch about what they study and why it matters, to fielding questions from experts in their field as well as non-experts, everything about this experience is designed to help students boost critical communication skills that will help them succeed in whatever comes next for them.

Three students affiliated with UIowaCS presented their Undergraduate Research Wednesday, March 29 at SURF 2023:

Mia Dukle
Major: Biomedical Sciences, Bioinformatics

Graduation: Spring 2026

Mentor: Drs. Hanna Stevens and Sara Maurer (Psychiatry)

Effects of Different Stress Models on Immune Gene Expression in the Mouse Placenta

The placenta is a temporary organ that serves as the sole connection between fetus and mother. Placental health is crucial for typical embryonic development. Previous studies have shown the importance of placental immune response and found that stress leads to more immune gene expression in male placentas. These studies and many others used to assess prenatal stress used different stress models, which may or may not be comparable to each other. Thus, we utilized three different stress models in comparison to a naïve group. I hypothesize many immune genes are expressed more during prenatal stress, particularly in male placentas compared to female ones. To test this, we administered different stress models to time-mated CD1 dams and collected placental tissues on embryonic day 14. Different immune genes such as CXCR7, S100B, CCL3, CXCL10, and CCL2 will be assessed in the placenta using standard qPCR methods. These stressors that dams experience can induce changes in immune gene expression starting in the placenta and then affecting the fetus which may increase the susceptibility of the fetus to neurodevelopmental disorders as gene expressions differ from homeostatic levels. We anticipate the data will show greater immune gene expression in male placentas after experiencing prenatal stress.

Maneesh John
Major: Computer Science and Engineering

Graduation: Spring 2023

Mentor: Dr. Mathews Jacob (Electrical and Computer Engineering)

Monotone model-based deep learning using non-monotone operators

In recent years, unrolled deep learning methods that combine imaging physics with learned regularization priors have been shown to be very effective for undersampled MRI reconstruction. These algorithms typically alternate between a data consistency block and a denoising block, unrolling these blocks for a fixed number of iterations. More recently, monotone operator learning (MOL) was proposed as a memory-efficient alternative to unrolled methods, with additional guarantees of uniqueness, convergence, and robustness. In practice, this allows MOL to be applied to higher-dimensional problems that current unrolled algorithms cannot handle, such as 3D MRI reconstruction. To achieve these properties, MOL enforces a monotonicity constraint on its convolutional neural network (CNN) denoising block. However, this monotonicity constraint significantly restricts the denoising performance of the CNN. The focus of this project is to improve monotone operator learning by relaxing the monotonicity constraint. Instead of constraining the CNN, we propose to constrain the combination of the CNN and the gradient of the data consistency term. The resulting deep equilibrium formulation provides the memory efficiency and theoretical guarantees of monotone operator learning, while also offering performance competitive with unrolled methods.

Hannah Back
Major: Data Science

Graduation: Spring 2024

Mentor: Dr. Susan Meerdink (Geographical and Sustainability Sciences)

Remote Detection of Harmful Algal Blooms

Harmful algal blooms (HABs) pose a hazard to human and animal health, cause economic damage, and negatively impact the ecosystem. Currently, Iowa Department of Natural Resources take water samples weekly and only at state-owned beaches. Because HABs can arise suddenly, people can be exposed unknowingly before the next sample has been processed. There is a clear need to develop cost-effective early detection methods. We are exploring methods for remote detection of HABs using trail cameras and other remote sensing approaches at our test site, Big Spirit Lake. Our data comes from conventional digital camera imagery taken five times daily. Each image is processed by first finding a region of interest (ROI) for a given lake. To find the “greenness” of the region, we calculate the relative intensity of green to the other color channels. We then produce a time-series to see how the greenness of the water changes during the year. We also use water quality and weather data to refine our understanding of how HABs arise and spread. Our next steps include processing satellite imagery and refining our trail camera processing methodology based on information found in these datasets.

All students involved in mentored research or creative work are invited to present their work at these festivals utilizing posters or other visual displays. A variety of judges from the University of Iowa community (peers, faculty, and graduate students) and the surrounding community are on hand to discuss student work and provide valuable feedback. Every aspect of these events, from poster design to presentation, to fielding questions, is designed to help students hone research communication skills. Watch this page, ICRU Facebook page, or join their mailing list for updates about these events! 

Are you making a poster for SURF or FURF? Learn more about creating Poster Presentations.