arXiv subjects

Yongkang Zhao

Publications and source records attributed to Yongkang Zhao.

4 recordsLinked to original sources

arXiv2026

Ultrafast optical coherence elastography for volumetric and dynamic in vivo imaging

Imaging the mechanical properties of biological tissues in vivo with high spatial and temporal resolution is essential for understanding physiological function and disease progression. Optical coherence elastography (OCE) provides label-free, micrometer-scale mapping of tissue biomechanics, but its application to dynamic and volumetric measurements has been limited by slow acquisition speeds and susceptibility to motion artifacts. Here we introduce ultrafast optical coherence elastography (ultrafast OCE), a general framework for real-time volumetric biomechanical imaging in vivo. By combining synchronized multi-phase acquisition with a demodulation strategy intrinsically robust to spectral aliasing, ultrafast OCE decouples mechanical excitation from acquisition speed, enabling reconstruction of full wave fields from only three sequential B-mode images. The method achieves frame rates up to two orders of magnitude higher than conventional approaches while preserving high sensitivity over frequencies ranging from the acoustic to ultrasonic regimes. We further develop a motion-correction strategy that compensates for bulk tissue motion under physiological conditions. We validate ultrafast OCE in dynamically stretched phantoms and pulsatile arteries and demonstrate sub-second volumetric imaging of the cornea and skin in vivo. Ultrafast OCE enables real-time interrogation of tissue biomechanics across space and time, opening new opportunities for mechanobiology, cardiovascular research, and clinical diagnostics.

physics.app-ph↗

arXiv2025

ACM-UNet: Adaptive Integration of CNNs and Mamba for Efficient Medical Image Segmentation

The U-shaped encoder-decoder architecture with skip connections has become a prevailing paradigm in medical image segmentation due to its simplicity and effectiveness. While many recent works aim to improve this framework by designing more powerful encoders and decoders, employing advanced convolutional neural networks (CNNs) for local feature extraction, Transformers or state space models (SSMs) such as Mamba for global context modeling, or hybrid combinations of both, these methods often struggle to fully utilize pretrained vision backbones (e.g., ResNet, ViT, VMamba) due to structural mismatches. To bridge this gap, we introduce ACM-UNet, a general-purpose segmentation framework that retains a simple UNet-like design while effectively incorporating pretrained CNNs and Mamba models through a lightweight adapter mechanism. This adapter resolves architectural incompatibilities and enables the model to harness the complementary strengths of CNNs and SSMs-namely, fine-grained local detail extraction and long-range dependency modeling. Additionally, we propose a hierarchical multi-scale wavelet transform module in the decoder to enhance feature fusion and reconstruction fidelity. Extensive experiments on the Synapse and ACDC benchmarks demonstrate that ACM-UNet achieves state-of-the-art performance while remaining computationally efficient. Notably, it reaches 85.12% Dice Score and 13.89mm HD95 on the Synapse dataset with 17.93G FLOPs, showcasing its effectiveness and scalability. Code is available at: https://github.com/zyklcode/ACM-UNet.

cs.CV↗

arXiv2023

An Adaptive Spatial-Temporal Local Feature Difference Method for Infrared Small-moving Target Detection

Detecting small moving targets accurately in infrared (IR) image sequences is a significant challenge. To address this problem, we propose a novel method called spatial-temporal local feature difference (STLFD) with adaptive background suppression (ABS). Our approach utilizes filters in the spatial and temporal domains and performs pixel-level ABS on the output to enhance the contrast between the target and the background. The proposed method comprises three steps. First, we obtain three temporal frame images based on the current frame image and extract two feature maps using the designed spatial domain and temporal domain filters. Next, we fuse the information of the spatial domain and temporal domain to produce the spatial-temporal feature maps and suppress noise using our pixel-level ABS module. Finally, we obtain the segmented binary map by applying a threshold. Our experimental results demonstrate that the proposed method outperforms existing state-of-the-art methods for infrared small-moving target detection.

cs.CV↗

arXiv2019

Altering nodes types in controlling complex networks

Controlling a complex network towards a desired state is of great importance in many applications. A network can be controlled by inputting suitable external signals into some selected nodes, which are called driver nodes. Previous works found there exist two control modes in dense networks: distributed and centralized modes. For networks with the distributed mode, most of the nodes can be act as driver nodes; and those with the centralized mode, most of the nodes never be the driver nodes. Here we present an efficient algorithm to change the control type of nodes, from input nodes to redundant nodes, which is done by reversing edges of the network. We conclude four possible cases when reversing an edge and show the control mode can be changed by reversing very few in-edges of driver nodes. We evaluate the performance of our algorithm on both synthetic and real networks. The experimental results show that the control mode of a network can be easily changed by reversing a few elaborately selected edges, and the number of possible driver nodes is dramatically decreased. Our methods provide the ability to design the desired control modes of the network for different control scenarios, which may be used in many application regions.

eess.SY↗