Yuanzhe Zhang

Yuanzhe Zhang

Auto-generated profile, based on publicly available information

Yuanzhe Zhang

University of Strathclyde, Department of Biomedical Engineering, Glasgow, UK

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
30 October 2024 in Computer Methods and Programs in Biomedicine
Verified
Journal Article
Open access
Zhenya ZangZhenya Zang
Zhenya Zang
QWQuan Wang
Quan Wang
MPMingliang Mingliang Pan
Mingliang Mingliang Pan
YZYuanzhe Zhang
Yuanzhe Zhang
XCXi Chen
Xi Chen
DLDavid Li
David Li
This study proposes a compact deep learning (DL) architecture and a highly parallelized computing hardware platform to reconstruct the blood flow index (BFi) in diffuse correlation spectroscopy (DCS). We leveraged a rigorous analytical model to generate autocorrelation functions (ACFs) to train the DL network. We assessed the accuracy of the proposed DL using simulated and milk phantom data. Compared to convolutional neural networks (CNN), our lightweight DL architecture achieves 66.7% and 18.5% improvement in MSE for BFi and the coherence factor /3, using synthetic data evaluation. The accuracy of rBFi over different algorithms was also investigated. We further simplified the DL computing primitives using subtraction for feature extraction, considering further hardware implementation. We extensively explored computing parallelism and fixed-point quantization within the DL architecture. With the DL model's compact size, we employed unrolling and pipe- lining optimizations for computation-intensive for-loops in the DL model while storing all learned parameters in on-chip BRAMs. We also achieved pixel-wise parallelism, enabling simultaneous, real-time processing of 10 and 15 autocorrelation functions on Zynq-7000 and Zynq-UltraScale+ field programmable gate array (FPGA), respectively. Unlike existing FPGA accelerators that produce BFi and the /3 from autocorrelation functions on standalone hardware, our approach is an encapsulated, end-to-end on-chip conversion process from intensity photon data to the temporal intensity ACF and subsequently reconstructing BFi and /3. This hardware platform achieves an on-chip solution to replace post-processing and miniaturize modern DCS systems that use single- photon cameras. We also comprehensively compared the computational efficiency of our FPGA accelerator to CPU and GPU solutions.

4 November 2025 in Biomedical Optics Express
Verified
Journal Article
Open access
YZYuanzhe Zhang
Yuanzhe Zhang
MPMingliang Mingliang Pan
Mingliang Mingliang Pan
CLChenxu Li
Chenxu Li
ZJZiao Jiao
Ziao Jiao
YHYuanyuan Hua
Yuanyuan Hua
DLDavid Li
David Li
Diffuse correlation spectroscopy (DCS) is a widely used noninvasive optical technique for measuring tissue blood flow. Accurate blood flow estimation with DCS requires a high signal-to-noise ratio (SNR), but achieving high SNR is often limited by safety constraints on the optical irradiance (maximum permissible exposure) that can be delivered to tissue. To overcome this limitation, we investigated the possibility of replacing the conventional multi-mode fiber (MMF) with a liquid light guide (LLG) for illumination. The LLG provides a more uniform illumination profile and higher photon throughput to the tissue under the same irradiance limit, resulting in a significantly increased detected photon count rate and enhanced SNR. In experiments under identical power-density conditions, the LLG-based system achieved approximately a three-fold increase in SNR compared to the traditional MMF configuration. This improvement arises from the uniform beam profile and efficient light delivery of the LLG, which permits safe use of higher total power. These results indicate that LLG illumination effectively enhances DCS sensitivity without exceeding safety limits, potentially enabling more sensitive and accurate blood flow monitoring in biomedical applications.

24 October 2025 in Journal of Innovative Optical Health Sciences
Verified
Journal Article
Open access
MPMingliang Mingliang Pan
Mingliang Mingliang Pan
QWQuan Wang
Quan Wang
YZYuanzhe Zhang
Yuanzhe Zhang
CLChenxu Li
Chenxu Li
DLDavid Li
David Li
Diffuse correlation spectroscopy (DCS) is an optical technology for extracting blood flow index (BFi) by measuring intensity fluctuations of the back-scattered light emitted from tissues. The remarkable characteristics of DCS, such as its noninvasiveness, deep penetration depth, and cost-effectiveness, have led to its widespread application for human health evaluation. However, traditional DCS data processing utilizes the analytical solution of the correlation diffusion equation to fit the measured autocorrelation function [ ], which is computationally demanding and susceptible to noise, especially when using the multi-layer analytical models to detect cerebral BFi. These drawbacks limit its further application for human BFi monitoring. To accelerate BFi extraction, deep learning (DL) techniques have been introduced. DL-assisted DCS has demonstrated fast and improved noise robustness for BFi extraction compared to traditional curve-fitting methods. In this review, the development of DL models for DCS data processing, current challenges and outlooks will be discussed, aiming to provide a reference for further promotion of the development of DL-based DCS technology for human health analysis.

21 April 2026 in Optics Letters
Verified
Journal Article
Open access
YZYuanzhe Zhang
Yuanzhe Zhang
MPMingliang Mingliang Pan
Mingliang Mingliang Pan
CLChenxu Li
Chenxu Li
DQDing Qin
Ding Qin
YCYi-Chieh Chen
Yi-Chieh Chen
DLDavid Li
David Li
Speckle contrast optical spectroscopy (SCOS) is often limited by short exposure bias and a reduced dynamic range at a long exposure time. We present a mechanics-free SCOS approach that simultaneously adjusts the exposure time and illumination spot size using an electrically tunable liquid lens (−3 to +3 dpt), redistributing photons across a broader exposure range. Using independently the measured field autocorrelation functions g 1 (τ), we validate the SCOS model and identify a reliable operating range of the normalized mean intensity between 0.2 and 0.5. Measurements acquired under different optical conditions are normalized by their respective speckle contrast factor β and combined into a single composite dataset, enabling joint analysis across multiple exposure times. The extracted correlation time ( τ c ) agrees with the reference value to within <1%. These results show that adaptive spot control with unified data combination enables a robust multi-exposure-time SCOS analysis.

9 September 2026 in Optics Express
Verified
Journal Article
Open access
JGJ. A. Gray
J. A. Gray
YZY. Zhang
Y. Zhang
JMJonathan Mckendry
Jonathan Mckendry
DLDavid Li
David Li
RHRobert Henderson
Robert Henderson
JHJ. Herrnsdorf
J. Herrnsdorf
Time of flight (ToF) ranging is used to gather 3D depth information in an environment and has become vital in applications from autonomous vehicles to facial recognition. While pulsed lasers are commonly used to illuminate the scene, these rely on additional optical components and are typically separate from a technology’s wireless communication system. LEDs — especially in micro-LED format — have emerged as an alternative pulsed illumination source, with their ubiquity and fast modulation unlocking Gbps data transmission with visible light. Here, we present a 128 × 128 micro-LED display that can act as a normal video display while also sensing objects in its environment through ranging with a SPAD detector. We demonstrate data-through-display communications at 0.68 Gbps while simultaneously performing depth imaging with 4.4 cm RMSE accuracy.

23 July 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
XLXingda Li
Xingda Li
Zhenya ZangZhenya Zang
Zhenya Zang
XCXi Chen
Xi Chen
MPMingliang Mingliang Pan
Mingliang Mingliang Pan
YZYuanzhe Zhang
Yuanzhe Zhang
DLDavid Day-Uei Li
David Day-Uei Li
Diffuse correlation spectroscopy (DCS) is a powerful non-invasive technique for assessing tissue blood-flow dynamics. However, computing the autocorrelation function (ACF) is computationally intensive and hardware demanding, creating a bottleneck for low-cost, scalable clinical DCS systems. This paper presents a novel approximate normalizer based on Mitchell’s logarithmic approximation, together with an efficient hardware implementation, to reduce the hardware resources required for ACF computation while controlling approximation-induced deviation. A novel error compensation scheme is proposed in which correction coefficients are precomputed using an analytical error model derived from the Mitchell algorithm. We evaluated the arithmetic error of a 32/16 divider with the proposed error compensation scheme using peak absolute relative error (PRE) and average absolute relative error (ARE) relative to exact division. With a resource-accuracy trade-off configuration, achieves up to 9.2x and 14x reductions in PRE and ARE, respectively, compared with existing approaches. To evaluate performance in practical applications, both in vitro (milk phantom) and in vivo (forearm cuff occlusion) experiments were conducted. The results show that the deviations remain small relative to observed experimental variability, indicating close agreement with exact normalization for both autocorrelation computation and downstream blood-flow estimation under the tested conditions. Compared with Xilinx high-radix fixed-point divider IP cores, the proposed design eliminates digital signal processing block (DSP) usage entirely and reduces flip-flop (FF) usage by 41.0%, with competitive latency. This enables a low-cost, resource-efficient solution for ACF computation in scalable DCS systems.