Yixuan Jia

Yixuan Jia

PhD student in EECS at UMich

Posts by Collection

portfolio

publications

Y-90 Bremsstrahlung SPECT Scatter Estimation and Voxel Dosimetry Using a Deep Learning Pipeline

Published in Journal of Nuclear Medicine (SNMMI 2023), 2023

Conference abstract presented at SNMMI 2023 (Oral) on deep learning-based Y-90 SPECT scatter estimation and voxel dosimetry.

Recommended citation: Y. Jia, Z. Li, J. Fessler, Y. Dewaraja. (2023). "Y-90 Bremsstrahlung SPECT Scatter Estimation and Voxel Dosimetry Using a Deep Learning Pipeline." Journal of Nuclear Medicine, 64(supplement 1), P1211.

90Y SPECT Scatter Estimation and Voxel Dosimetry in Radioembolization Using a Unified Deep Learning Framework

Published in EJNMMI Physics, 2023

A unified three-stage deep learning framework for Y90 SPECT: CNN-based scatter estimation, reconstruction with scatter correction, and dose-rate map generation.

Recommended citation: Y. Jia, Z. Li, A. Akhavanallaf, J.A. Fessler, Y.K. Dewaraja. (2023). "90Y SPECT Scatter Estimation and Voxel Dosimetry in Radioembolization Using a Unified Deep Learning Framework." EJNMMI Physics, 10(1), 82.
Download Paper

Segment Anything Model for SPECT (SAMS): Novel Implementation in SPECT Imaging for Tumor Segmentation

Published in Journal of Nuclear Medicine (SNMMI 2024), 2024

Conference abstract presented at SNMMI 2024 (Oral) on applying the Segment Anything Model to SPECT imaging for tumor segmentation.

Recommended citation: Z. Lu, Z. Li, Y. Jia, G. Chen, M. Roseland, G. Mok, Y. Dewaraja. (2024). "Segment Anything Model for SPECT (SAMS): Novel Implementation in SPECT Imaging for Tumor Segmentation." Journal of Nuclear Medicine, 65(supplement 2), 241583.

Shorter SPECT Scans Using Self-Supervised Coordinate Learning to Synthesize Skipped Projection Views

Published in EJNMMI Physics, 2025

We adapt the neural radiance field (NeRF) concept to SPECT imaging, enabling significant reduction in acquisition time via self-supervised coordinate learning.

Recommended citation: Z. Li*, Y. Jia*, X. Xu, J. Hu, J.A. Fessler, Y.K. Dewaraja. (2025). "Shorter SPECT Scans Using Self-Supervised Coordinate Learning to Synthesize Skipped Projection Views." EJNMMI Physics, 12(1), 47.
Download Paper

FlowDAS: A Stochastic Interpolant-Based Framework for Data Assimilation

Published in Neural Information Processing Systems (NeurIPS), 2025

We introduce FlowDAS, a flow-based framework using stochastic interpolants to unify the learning of state transition dynamics and generative priors for data assimilation.

Recommended citation: S. Chen*, Y. Jia*, Q. Qu, H. Sun, J.A. Fessler. (2025). "FlowDAS: A Stochastic Interpolant-Based Framework for Data Assimilation." NeurIPS, 2025.
Download Paper

MCLR: Improving Conditional Modeling in Visual Generative Models via Inter-Class Likelihood-Ratio Maximization and Establishing the Equivalence between Classifier-Free Guidance and Alignment Objectives

Published in arXiv preprint, 2026

We propose MCLR, a principled alignment objective that maximizes inter-class likelihood-ratios during training, achieving CFG-like improvements without inference-time guidance.

Recommended citation: X. Li, Y. Jia, X. Li, J.A. Fessler, R. Wang, Q. Qu. (2026). "MCLR: Improving Conditional Modeling in Visual Generative Models via Inter-Class Likelihood-Ratio Maximization and Establishing the Equivalence between Classifier-Free Guidance and Alignment Objectives." arXiv preprint arXiv:2603.22364.
Download Paper

talks

teaching

Biomedical AI

Graduate course, University of Michigan, 2023

Graduate Student Instructor, Sep 2023 – Dec 2023.

AI Magic Summer School

Summer program, University of Michigan, 2024

Graduate Student Instructor, Jul 2024 – Aug 2024.