Research
Ultrafast Radiation Detector
Our research focuses on ultrafast radiation detectors that enable next-generation medical imaging modalities. To achieve superior time resolution, we integrate advanced photonics technologies to enhance scintillator light yield and optimize light transfer between scintillators and photosensors, together with cutting-edge electronics and Edge AI–based signal processing for precise timing extraction. These approaches allow detectors to capture radiation events with unprecedented temporal precision—approaching less than 10 ps—paving the way for transformative imaging applications.
By harnessing these ultrafast detectors, we aim to realize direct Positron Emission Imaging (dPEI), which enables high-speed molecular imaging without the need for complex reconstruction, and time-of-flight computed tomography (TOF-CT), where timing information can be used to reject scattered events and dramatically improve image quality.
Ultimately, this research seeks to establish a new paradigm in radiation imaging systems that combine both speed and accuracy, advancing diagnostic performance and patient care.
Next-Generation Medical Imaging System
Our research explores next-generation medical imaging systems that achieve unprecedented temporal and spectral resolution.
In this effort, we are developing dedicated brain PET systems that integrate time-of-flight (TOF), depth-of-interaction (DOI), and AI-driven image reconstruction technologies to dramatically enhance image quality and quantitative accuracy. Another promising direction is direct positron emission imaging (dPEI), which detects annihilation photons without conventional scintillation conversion, enabling picosecond-level timing and real-time visualization of ultra-fast biological processes.
We are also pursuing photon-counting computed tomography (PCCT) combined with TOF technology. By implementing stationary CT architectures with ultra-fast detectors, this approach aims to accelerate image acquisition and achieve precise, motion-free imaging of dynamic organs such as the heart.
AI-Based Medical Imaging and Image Intelligence
AI-based medical imaging aims to enhance image quality, diagnostic reliability, and quantitative accuracy across diverse imaging modalities. By applying a broad range of AI techniques, including deep learning, physics-informed modeling, and data-driven optimization, image noise and artifacts can be effectively reduced while preserving clinically meaningful structures.
Our research investigates comprehensive AI frameworks for medical image denoising and enhancement, ranging from conventional deep neural networks to generative and diffusion-based models. In parallel, we explore multimodal AI approaches that integrate medical images with acquisition parameters, clinical context, and textual knowledge using large language models (LLMs), enabling adaptive and context-aware image processing.
By combining image-level AI, multimodal intelligence, and medical imaging domain knowledge, this research aims to establish a general-purpose AI platform that supports low-dose imaging, improves diagnostic confidence, and bridges the gap between raw imaging data and clinical decision-making.
Semiconductor Photodetectors and Fast Electronics
Semiconductor-based photodetectors such as silicon photomultipliers (SiPMs) are redefining the standard for fast and reliable photon detection. Beyond using existing devices, our work extends to the design and fabrication of novel photodetector architectures, focusing on higher photon detection efficiency, improved timing precision, and better scalability for large-area imaging systems.
Complementing sensor development, we create custom application-specific integrated circuits (ASICs) that provide low-noise amplification, precise time-to-digital conversion, and high-throughput data processing. By co-developing sensors and electronics, we ensure that the interface between photodetector and readout is fully optimized, enabling performance levels unattainable by either component alone.
This dual focus—advancing both semiconductor sensors and their dedicated ASICs—opens the path to compact, high-performance modules for PET, CT, and hybrid imaging modalities, supporting breakthroughs in medical imaging and radiation detection technologies.
