Research

In-memory Computing System  

Our research on in-memory computing (IMC) systems focuses on overcoming the memory bottleneck in conventional von Neumann architectures by tightly integrating computation within memory arrays. We are targeting energy-efficient digital and mixed-signal IMC architectures that enable high-throughput and low-power processing for data-intensive applications such as artificial intelligence and sensor signal processing. 

Biomedical Sensor Interface System

Our work covers interfaces for capacitive, resistive, ultrasound, optical (light), and acoustic (voice) sensors, addressing the diverse characteristics and signal requirements of each modality. Through analog and mixed-signal circuit design, I focus on achieving high sensitivity, low noise, wide dynamic range, and low power consumption, which are critical for wearable and implantable biomedical systems.

Intelligent Neural Recording and Stimulation

Our research on biomedical sensor interface systems focuses on the design of biosignal readout architectures that can flexibly support a wide dynamic range while achieving high noise performance and power efficiency. We aim to develop robust analog and mixed-signal front-end circuits that enable accurate acquisition of weak and noisy physiological signals across diverse sensing modalities, including electrical, optical, and acoustic biosignals. Moreover, by combining in-memory computing hardware with deep learning-based neural signal classification, our research targets the realization of intelligent, closed-loop neuromodulation systems that can generate adaptive and optimized neural stimulation patterns enhancing the therapeutic efficacy for various neurological disorders.

Ultrasound Image Classification Deep learning System

Our research on ultrasound image classification using deep learning systems focuses on developing robust and efficient AI-based diagnostic frameworks for medical ultrasound imaging. We aim to design end-to-end systems that integrate ultrasound signal acquisition, image reconstruction, and deep learning-based classification, enabling accurate and reliable interpretation of ultrasound data. In particular, our work targets high-accuracy classification of ultrasound images under limited data and noisy imaging conditions, which are common challenges in real clinical environments.

Image Restoration Compressive Sensing System

Our research on image restoration and compressive sensing systems focuses on developing efficient and robust signal acquisition and reconstruction frameworks for high-quality image recovery under limited sampling and degraded imaging conditions. We aim to design end-to-end systems that jointly optimize sensing strategies, reconstruction algorithms, and hardware-efficient system architectures, enabling accurate image restoration from compressed, noisy, or incomplete measurements.