IMSA-Artificial intelligence (AI) and generative models in Biomedical Data en Argonne

Internship Description

Background:
Artificial intelligence (AI) and generative models are opening new frontiers in biomedical research by creating realistic synthetic datasets that preserve important biological signals while protecting patient privacy. Recent advances, such as diffusion models, have shown the ability to generate tumor whole-slide image (WSI) tiles directly from RNA-sequencing data. These synthetic images not only capture variation in cell-type distributions linked to gene expression but can also be used to train machine-learning models that outperform those built from scratch. Evaluating different diffusion models is important to understand the trade-offs between computational cost and biological fidelity in generating synthetic biomedical data.

Description of Student Internship:
In this internship, the student will explore how different diffusion models perform when generating synthetic tumor image tiles from RNA-sequencing data. The project will involve comparing computational performance (e.g., runtime, memory usage, scalability) and evaluating accuracy using model metrics (e.g., image quality scores, alignment with biological features such as cell-type distributions). Students will gain practical experience in running and evaluating generative AI models, using Python frameworks (e.g., PyTorch, Hugging Face Diffusers), and designing experiments to balance efficiency and accuracy. The project will culminate in a short written report and presentation comparing at least two diffusion models, with an emphasis on both technical performance and biomedical relevance.
 

Education and Experience Requirements

Required Skills:
      •     Curiosity about artificial intelligence and biomedical applications
      •     Interest in coding with Python (prior exposure is helpful but not required)
      •     Willingness to learn about machine learning concepts and generative models
      •     Enthusiasm for analyzing results and thinking critically about their scientific meaning

Internship Family

Internship Category