About me
I am a fifth-year Ph.D. student in Computer Science at Johns Hopkins University, fortunately advised by Professor Kenichi Oishi and Professor Craig Jones. Previously, I had the opportunity to collaborate with Professor René Vidal.
My research lies at the intersection of deep learning and medical imaging, with a focus on brain MRI analysis. Specifically, I work on transformer-based models for brain age estimation and diffusion-based approaches for unsupervised neonatal lesion detection and inpainting. More broadly, my interests span multimodal representation learning, clinically informed model design, and improving the generalizability of medical imaging models.
My long-term research vision is to advance methods for multimodal information representation through the Vision-Language models and foundation models, and to leverage deep learning to bridge AI and clinical practice, with the goal of deepening our knowledge of brain cognition, neurodevelopment, and neurodegeneration.
Previously, I received my BS in Computer Science, Mathematics, and Economics, with a minor in Physics, from the University of Wisconsin - Madison, where I was advised by Professor Vikas Singh.
Recent News
Our preprint “OpenMAP-HIE: Pathology Informed Score-based Diffusion Model for Unsupervised Neonatal Hypoxic-Ischemic Encephalopathy Lesion Segmentation” is available (preprint, code).
Our manuscript “OpenMAP-BrainAge: Generalizable and Interpretable Brain Age Predictor from MRI” was accepted to Radiology Advances.
Our work “Clinically Interpretable Transformer Model Reveals Neurodevelopmental Delay in Preterm and NICU Infants Using Diffusion MRI” was accepted to the 2026 ISMRM & ISMRT Annual Meeting & Exhibition (Cape Town, South Africa, May 9–14, 2026).
At the 150th Annual Meeting of the American Neurological Association (ANA 2025, Baltimore, Sep 2025), our work “Multiview Transformer for Brain Age Prediction” received a Poster Award, was selected for an oral presentation (one of five talks chosen from hundreds of abstracts), and was named an Abstract of Distinction; I also received a Travel Award and presented as an Emerging Scholar (speaker profile). The Johns Hopkins Computer Science department featured these recognitions.
I gave a guest lecture titled “Introduction to Transformer in Computer Vision” for the Fall 2023 Computer Vision course at Johns Hopkins University (Nov 2023).
Publications
OpenMAP-HIE: Pathology Informed Score-based Diffusion Model for Unsupervised Neonatal Hypoxic-Ischemic Encephalopathy Lesion Segmentation
Preprint, 2026
A pathology-informed score-based diffusion model that localizes neonatal HIE lesions on ADC maps without voxel-wise lesion supervision.
Clinically Interpretable Transformer Model Reveals Neurodevelopmental Delay in Preterm and NICU Infants Using Diffusion MRI
Accepted, ISMRM, 2026
A clinically interpretable transformer applied to diffusion MRI to predict biological brain age of neonates and discover associations between neurodevelopmental delay to preterm and NICU conditions.
OpenMAP-BrainAge: Generalizable and Interpretable Brain Age Predictor
Radiology Advances, 2025
A multiview transformer framework with infomration fusion across different anatomical views and numerical brain volume features for brain age prediction in linear computational cost, which emphasizes generalization across sites and interpretability of imaging features.
TransMAD: Transformer-based Masked Adversarial Defenses to Patch Attacks
Preprint, 2023
Introduces a transformer-based masked adversarial defense with simple plug-in that reduces vulnerability to patch attacks while maintaining high performance on benign data.
A Guided Latent Dirichlet Allocation Approach to Investigate Real-time Latent Topics of Twitter Data during Hurricane Laura
Journal of Information Science, 2021
Uses guided LDA to track evolving public discussion themes in Twitter data collected during Hurricane Laura.
Towards Diffeomorphism Invariant Convolution Neural Networks
Differential Geometry in Computer Vision and Machine Learning workshop, CVPR, 2021
Develops convolutional blocks through jacobian space whose outputs are stable under diffeomorphic deformations of the input domain.
Exploring Adversarial Robustness of Deep Metric Learning
arXiv preprint, 2021
Empirically studies how adversarial examples behave in deep metric learning and defenses to improve the robustness of deep metric learning.
Application of Image Segmentation in Surface Water Extraction of Freshwater Lakes using Radar Data
ISPRS International Journal of Geo-Information, 2020
Applies semantic segmentation to SAR imagery to map surface water in freshwater lakes for hydrologic monitoring.
Teachings
- Teaching Assistant, EN 601.783 Vision as Bayesian Inference - Spring 2024 (with Prof. Alan Yuille)
- Teaching Assistant, EN 601.661 Computer Vision - Fall 2023 (with Prof. Kapil D. Katyal)