Jiajing Zhang

I am a Ph.D. candidate at the Department of Electrical and Computer Engineering (ECE), the University of Hong Kong (HKU), supervised by Prof. Wei-Ning Lee.

My research focuses on Computer Vision, Multimodal Image Processing, Vision-Language Models, and AI Agents for Science. My recent work mainly contributes to the ultrasound imaging field, including image (and IQ signal) reconstruction, denoising, segmentation, disease diagnosis, and biomarker measurement.

Research

Ultrasound reconstruction and image enhancement

I develop learning-based methods for reconstructing high-quality ultrasound data from sparse or degraded measurements. My work on synthetic-aperture ultrasound investigates spatiotemporal IQ signal recovery from sparse transmissions, improving acquisition efficiency. I also propose self-supervised test-time training for single-shot ultrasound denoising, with an emphasis on robustness under distribution shifts.

Medical image understanding and foundation models

I investigate anatomy-aware and modality-aware representations for generalizable medical image analysis by incorporating anatomical structure or imaging physics into model design. This includes developing foundation models for multimodal segmentation, as well as adapting VLMs to effectively understand cardiac ultrasound images for precise clinical interpretation.

Translational biomedical AI

I develop clinically oriented models to measure quantitative information from medical images or physiological signals. Applications include ultrasound-based musculoskeletal assessment, CT-based robotic surgery planning, MRI-based cancer segmentation, facial weakness assessment for rare myopathy, and electrocardio panorama synthesis. This research highlights interpretability and practical integration into imaging or intervention systems.