Biomedical images such as fMRI are used for training deep learning models to track tumor growth and identify lesions. However, the availability of these images is limited and we have to restore to data augmentation and image generation procedures. We are adopting score based and consistency diffusion models popular in computer vision for medical image generation. In this presentation, we will see these two methods applied to the generation of sparse jet images, focusing on high-fidelity synthesis for high energy physics applications. Using the JetNet dataset, the diffusion models are trained to learn the visual representation of jet kinematics. The results demonstrate that consistency models achieve significantly lower Fréchet inception distance measures compared to score-based models, indicating improved image quality and generation stability. Unlike methods based on latent distributions, this approach operates directly in image space. Furthermore, the efficacy of jet image generation is demonstrated using histograms of jet mass and other metrics to highlight the strengths of image-based jet generative modeling for training deep learning models. We also show how score based diffusion models can be used for calorimeter shower image generation that have particular symmetric structures similar to brain scan images.

VIdya Manian has a Ph.D. in Computing, Information Science, and Engineering from the University of Puerto Rico, Mayagüez (UPRM), in 2004. She was a visiting scholar at West Virginia University, Morgantown, in 2005. Since 2006, she has been a faculty member in the Department of Electrical and Computer Engineering and Bioengineering at UPRM. Dr. Manian is the Director of the Artificial Intelligence Imaging Group (AIIG) at UPRM and an investigator with the MIT Institute for Artificial Intelligence and Fundamental Interactions (IAIFI). Her research interests include developing AI/ML methods for high-energy physics jet classification and anomaly detection, hyperspectral image analysis, and the analysis of neural signals from fMRI and EEG for neurodegenerative disease prediction and brain-computer interface applications. Dr. Manian was the recipient of the Distinguished Professor Award at UPRM in 2014, 2015, and 2020. With her students she has authored 35 peer-reviewed journal articles and over 50 conference papers.