Lufeng Che
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Lufeng Che has a total of 101 co-authors. Here are the top co-authors by number of shared publications:
Baoyin Hou
Zhejiang University
7 shared publicationsZhi Ding
College of Information Science and Electronic Engineering, Zhejiang University
6 shared publications- Chaoran Liu
Ministry of Education Engineering Research Center of Smart Microsensors and Microsystems, College of Electronics and Information, Hangzhou Dianzi University, Hangzhou, China
5 shared publications Weidong Wang
College of Information Science and Electronic Engineering, Zhejiang University, Hangzhou, China
5 shared publicationsYe Zhu
Zhejiang University
5 shared publications- Linxi Dong
Hangzhou Dianzi University
3 shared publications
Latest Publications
- 1 August 2026
- Published by IOP Publishing in Biomedical Physics & Engineering Express
- Vol. 12 (4) , 045057
HPA‐Net: a lightweight hierarchical progressive fusion network for cardiac disease detection based on ECG-PCG signals
Objective. Cardiovascular disease remains a leading global health threat, creating a strong clinical need for convenient and rapid diagnostic methods. However, current single-modality and conventional fusion methods often fail to adequately utilize the intrinsic bioelectromechanical coupling characteristics of the heart. To address these limitations, we propose the hierarchical progressive attention fusion network (HPA-Net), a lightweight, dual-modal framework designed for efficient and reliable diagnostic performance on resource-constrained devices. Approach. HPA-Net is a compactdeep learning architecture comprising only 0.21 million parameters and 0.362G FLOPs. It utilizes a multi-scale convolutional attention encoder to extract spatiotemporal representations from electrocardiogram and phonocardiogram signals. A multi-level strategy was employed to integrate the features across multiple processing stages. Furthermore, the fast adaptive multitask optimization technique was implemented to strengthen multi-branch learning and mitigate the issue of modality dominance during the training process. Main results. Rigorous validation on the PhysioNet/CinC 2016 subset-A dataset demonstrated that HPA-Net achieved an accuracy of 97.59 ± 0.72%, sensitivity of 98.42 ± 0.46%, and specificity of 95.53 ± 1.87%, achieving highly competitive accuracy with drastically reduced computational overhead. Despite its compact size, the model exhibited strong generalization, reaching 97% accuracy on the independent unimodal datasets. Technical evaluation showed an inference latency of 41.20 ms and a real-time factor of 0.008. Significance. The proposed architecture efficiently acquires complex spatiotemporal representations while maintaining a minimal computational demand. Its high accuracy combined with low inference latency confirms the feasibility of HPA-Net for seamless deployment on edge hardware. providing a promising solution for real-time cardiovascular monitoring and early disease detection in practical clinical or home care scenarios.
- 1 August 2026
- Published by IOP Publishing in Biomedical Physics & Engineering Express
- Vol. 12 (4) , 045057
HPA‐Net: a lightweight hierarchical progressive fusion network for cardiac disease detection based on ECG-PCG signals
Objective. Cardiovascular disease remains a leading global health threat, creating a strong clinical need for convenient and rapid diagnostic methods. However, current single-modality and conventional fusion methods often fail to adequately utilize the intrinsic bioelectromechanical coupling characteristics of the heart. To address these limitations, we propose the hierarchical progressive attention fusion network (HPA-Net), a lightweight, dual-modal framework designed for efficient and reliable diagnostic performance on resource-constrained devices. Approach. HPA-Net is a compactdeep learning architecture comprising only 0.21 million parameters and 0.362G FLOPs. It utilizes a multi-scale convolutional attention encoder to extract spatiotemporal representations from electrocardiogram and phonocardiogram signals. A multi-level strategy was employed to integrate the features across multiple processing stages. Furthermore, the fast adaptive multitask optimization technique was implemented to strengthen multi-branch learning and mitigate the issue of modality dominance during the training process. Main results. Rigorous validation on the PhysioNet/CinC 2016 subset-A dataset demonstrated that HPA-Net achieved an accuracy of 97.59 ± 0.72%, sensitivity of 98.42 ± 0.46%, and specificity of 95.53 ± 1.87%, achieving highly competitive accuracy with drastically reduced computational overhead. Despite its compact size, the model exhibited strong generalization, reaching 97% accuracy on the independent unimodal datasets. Technical evaluation showed an inference latency of 41.20 ms and a real-time factor of 0.008. Significance. The proposed architecture efficiently acquires complex spatiotemporal representations while maintaining a minimal computational demand. Its high accuracy combined with low inference latency confirms the feasibility of HPA-Net for seamless deployment on edge hardware. providing a promising solution for real-time cardiovascular monitoring and early disease detection in practical clinical or home care scenarios.
- 12 March 2026
- Published by Wiley in Advanced Functional Materials
- Vol. 36 (41)
A Stretchable AI‐Enhanced Patch with Low‐Drift Iontronic Pressure and Vibration Sensor for Soft Film Thickness Recognition and Knee Motion Decoding
Flexible iontronic mechanical sensors that mimic ion flow mechanisms in human mechanoreceptors can precisely sense physical stimuli, such as strain or pressure. An easily neglected issue is that these sensors often suffer from irreparable signal‐drift under continuous static pressure owing to mechanical hardening in the sensitive interface. Additionally, a mechanical mismatch between the flexible sensor and the rigid circuit limits the reliable application on the dynamic surface of human skin. Here, inspired by the tip conductor to non‐contact guide lightning, a non‐hardening solid‐liquid contact interface is engineered to form the electric double layer under pressure or vibration, which enables low‐drift and pressure‐vibration iontronic sensing. The iontronic sensor exhibits exceptional pressure sensitivity 23 550.6 kPa − 1 , and detects subtle vibration as low as 0.03 g. Leveraging low drift and high sensitivity, the sensor measures the thickness of soft films, distinguishing minimum thicknesses down to 10 µm, and CNN model enables recognition of 7 distinct film types with 93.6% accuracy. Furthermore, a stretchable circuit patch integrating an iontronic sensor and electronic chips is applied to the knee for motion monitoring. Using a transfer learning model, the patch system achieves real‑time decoding of 30 knee motions with 94.3% accuracy by mapping bending degree and shaking counts to x/y coordinates in a character matrix.
- 12 March 2026
- Published by Wiley in Advanced Functional Materials
- Vol. 36 (41)
A Stretchable AI‐Enhanced Patch with Low‐Drift Iontronic Pressure and Vibration Sensor for Soft Film Thickness Recognition and Knee Motion Decoding
Flexible iontronic mechanical sensors that mimic ion flow mechanisms in human mechanoreceptors can precisely sense physical stimuli, such as strain or pressure. An easily neglected issue is that these sensors often suffer from irreparable signal‐drift under continuous static pressure owing to mechanical hardening in the sensitive interface. Additionally, a mechanical mismatch between the flexible sensor and the rigid circuit limits the reliable application on the dynamic surface of human skin. Here, inspired by the tip conductor to non‐contact guide lightning, a non‐hardening solid‐liquid contact interface is engineered to form the electric double layer under pressure or vibration, which enables low‐drift and pressure‐vibration iontronic sensing. The iontronic sensor exhibits exceptional pressure sensitivity 23 550.6 kPa − 1 , and detects subtle vibration as low as 0.03 g. Leveraging low drift and high sensitivity, the sensor measures the thickness of soft films, distinguishing minimum thicknesses down to 10 µm, and CNN model enables recognition of 7 distinct film types with 93.6% accuracy. Furthermore, a stretchable circuit patch integrating an iontronic sensor and electronic chips is applied to the knee for motion monitoring. Using a transfer learning model, the patch system achieves real‑time decoding of 30 knee motions with 94.3% accuracy by mapping bending degree and shaking counts to x/y coordinates in a character matrix.
- 20 January 2026
- Published by Springer Nature
- p. 147-156
Highly Sensitive and Fast-Response Iontronic Flexible Pressure Sensor for AI-Enabled Morse Code Classification
- 20 January 2026
- Published by Springer Nature
- p. 147-156
Highly Sensitive and Fast-Response Iontronic Flexible Pressure Sensor for AI-Enabled Morse Code Classification
- 20 January 2026
- Published by Springer Nature
- p. 113-122
Programmable Quasi-Zero Stiffness via 3D Printed Kresling-Origami Structures
- 20 January 2026
- Published by Springer Nature
- p. 113-122
Programmable Quasi-Zero Stiffness via 3D Printed Kresling-Origami Structures
- 20 January 2026
- Published by Springer Nature
- p. 125-133
Piezoelectric Micromachined Ultrasonic Transducer for Carotid Arterial Diameter Monitoring
- 20 January 2026
- Published by Springer Nature
- p. 125-133
Piezoelectric Micromachined Ultrasonic Transducer for Carotid Arterial Diameter Monitoring
- 20 January 2026
- Published by Springer Nature
- p. 227-235
Design and Fabrication of Simulation-Optimized Polysilicon Microhemispherical Shell Resonators with Self-aligned Curved Electrodes
- 20 January 2026
- Published by Springer Nature
- p. 227-235
Design and Fabrication of Simulation-Optimized Polysilicon Microhemispherical Shell Resonators with Self-aligned Curved Electrodes
- 12 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Microelectromechanical Systems
- Vol. 35 (1) , 172-180
Synchronous Enhancement of Sensitivity and Frequency of Piezoelectric Micromachined Ultrasonic Transducers Operating in the Second-Order Flexural Mode
Piezoelectric Micromachined Ultrasonic Transducers (PMUTs) are widely used in sensing applications, where high-order modes of PMUT can be used to enhance sensing accuracy. However, in existing works, when PMUTs work at a high-order resonance mode, their weak amplitude of flexural vibration and sound pressure cancellation from opposite membrane motions result in low sensitivity. Here, we demonstrate a novel PMUT with double-turn slots at the edge of the membrane, transforming the fully-clamped membrane into a suspended structure, thus modifying the mode shape. When the PMUT operates in the second-order mode, the structure alleviates sound pressure cancellation, and enables large vibration amplitude, thus enhancing both sensitivity and frequency. AlN-based PMUT prototypes are fabricated on a 4-inch wafer with CMOS compatible processes for validation. Compared to fully-clamped PMUTs operating in the fundamental resonance mode, the central displacements of the two devices with different slot angles are increased by 153% and 84%, their frequencies are improved by 59% and 189%, respectively, and the round-trip sensitivity is enhanced by 5.47 dB. As demonstration examples, pulse-echo-based ranging and imaging are explored using a $3\times 5$ PMUT array. The ranging accuracy is 0.65 mm with a maximum distance of 50 mm, and the imaging deviation is 0.6 mm at a distance of 2 cm. These results demonstrate great application potential of this innovative design.
- 12 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Microelectromechanical Systems
- Vol. 35 (1) , 172-180
Synchronous Enhancement of Sensitivity and Frequency of Piezoelectric Micromachined Ultrasonic Transducers Operating in the Second-Order Flexural Mode
Piezoelectric Micromachined Ultrasonic Transducers (PMUTs) are widely used in sensing applications, where high-order modes of PMUT can be used to enhance sensing accuracy. However, in existing works, when PMUTs work at a high-order resonance mode, their weak amplitude of flexural vibration and sound pressure cancellation from opposite membrane motions result in low sensitivity. Here, we demonstrate a novel PMUT with double-turn slots at the edge of the membrane, transforming the fully-clamped membrane into a suspended structure, thus modifying the mode shape. When the PMUT operates in the second-order mode, the structure alleviates sound pressure cancellation, and enables large vibration amplitude, thus enhancing both sensitivity and frequency. AlN-based PMUT prototypes are fabricated on a 4-inch wafer with CMOS compatible processes for validation. Compared to fully-clamped PMUTs operating in the fundamental resonance mode, the central displacements of the two devices with different slot angles are increased by 153% and 84%, their frequencies are improved by 59% and 189%, respectively, and the round-trip sensitivity is enhanced by 5.47 dB. As demonstration examples, pulse-echo-based ranging and imaging are explored using a $3\times 5$ PMUT array. The ranging accuracy is 0.65 mm with a maximum distance of 50 mm, and the imaging deviation is 0.6 mm at a distance of 2 cm. These results demonstrate great application potential of this innovative design.
- 28 November 2025
- Published by Wiley in Advanced Functional Materials
- Vol. 36 (24)
A Fully Integrated Patch for Real‐Time AI‐Enhanced Haptic Closed‐Loop Interaction of Complete 128 ASCII Codes
The development of human–computer interaction technology is directly linked to economic development, serving as a crucial threshold that enables information technology to integrate into society. However, current emerging HCI faces challenges with not supporting complete 128 ASCII codes and insufficient interaction feedback. Here, a wireless, real‐time and AI‐enhanced haptic HCI system based on a conformal skin patch is proposed, supporting closed‐loop interaction of complete 128 ASCII codes. The patch leverages the outstanding electrical performance of haptic sensor and actuator, the mechanical properties of flexible serpentine circuits, and the efficient processing of advanced AI algorithms. Through bionic fingerprint electrode and in situ grown fibrous microstructure, the haptic sensor achieves high sensitivity of up to 997.2 kPa −1 . Small vibration actuator modulated by PWM can provide 7 levels of haptic feedback, achieving 91% distinction accuracy of individual array elements on skin. To avoid the cumbersome process of data collection, a sensor array encoding rule is defined capable of covering complete 128 ASCII codes, and mathematical modeling simulates pressing behaviors and efficiently synthesizes large‐scale dataset. Advanced CNN deep learning, trained on the massive dataset, successfully classifies complete 128 ASCII codes, including complex characters like “G,” “o,” “!” and “<”.
- 28 November 2025
- Published by Wiley in Advanced Functional Materials
- Vol. 36 (24)
A Fully Integrated Patch for Real‐Time AI‐Enhanced Haptic Closed‐Loop Interaction of Complete 128 ASCII Codes
The development of human–computer interaction technology is directly linked to economic development, serving as a crucial threshold that enables information technology to integrate into society. However, current emerging HCI faces challenges with not supporting complete 128 ASCII codes and insufficient interaction feedback. Here, a wireless, real‐time and AI‐enhanced haptic HCI system based on a conformal skin patch is proposed, supporting closed‐loop interaction of complete 128 ASCII codes. The patch leverages the outstanding electrical performance of haptic sensor and actuator, the mechanical properties of flexible serpentine circuits, and the efficient processing of advanced AI algorithms. Through bionic fingerprint electrode and in situ grown fibrous microstructure, the haptic sensor achieves high sensitivity of up to 997.2 kPa −1 . Small vibration actuator modulated by PWM can provide 7 levels of haptic feedback, achieving 91% distinction accuracy of individual array elements on skin. To avoid the cumbersome process of data collection, a sensor array encoding rule is defined capable of covering complete 128 ASCII codes, and mathematical modeling simulates pressing behaviors and efficiently synthesizes large‐scale dataset. Advanced CNN deep learning, trained on the massive dataset, successfully classifies complete 128 ASCII codes, including complex characters like “G,” “o,” “!” and “<”.
- 18 September 2025
- Published by IOP Publishing in Journal of Micromechanics and Microengineering
- Vol. 35 (9) , 095012
Measurement of Earth tides with a MEMS gravimeter via releasing the axial force of cross-configured geometric anti-spring
Gravimeters play an important role in geophysical research, including mineral prospecting, hazard forecasting, and crustal deformation monitoring. However, existing commercial gravimeters are typically bulky and expensive. MEMS gravimeters incorporating geometric anti-spring (GAS) mechanism effectively address these limitations and exhibit high sensitivity and low resonant frequency. In this work, we employ finite element analysis to investigate the relationship between the stiffness characteristics of the GAS and its axial force. Through this analysis, we propose a novel design that adopts a crossed configuration of two sets of GASs with different dimensions, enabling the proof mass to displace both along and perpendicular to the sensitive axis. This configuration effectively reduces the axial forces in the shorter-arc-length GAS and mitigates the spring system’s negative stiffness effect. By optimizing the dimensions of GASs, the quasi-zero stiffness characteristic can be achieved. A low-cost, low-noise optical shadow displacement sensor is designed to detect the displacement of the proof mass. A 5 d measurement of Earth tides reveal that the experimental data has a correlation coefficient of 0.91 with the theoretical date from the TOSFT software. Final tests demonstrate that the micro-electro-mechanical system (MEMS) gravimeter achieves a noise power spectral density of 5 μGal √Hz–1 at 3 Hz.
- 18 September 2025
- Published by IOP Publishing in Journal of Micromechanics and Microengineering
- Vol. 35 (9) , 095012
Measurement of Earth tides with a MEMS gravimeter via releasing the axial force of cross-configured geometric anti-spring
Gravimeters play an important role in geophysical research, including mineral prospecting, hazard forecasting, and crustal deformation monitoring. However, existing commercial gravimeters are typically bulky and expensive. MEMS gravimeters incorporating geometric anti-spring (GAS) mechanism effectively address these limitations and exhibit high sensitivity and low resonant frequency. In this work, we employ finite element analysis to investigate the relationship between the stiffness characteristics of the GAS and its axial force. Through this analysis, we propose a novel design that adopts a crossed configuration of two sets of GASs with different dimensions, enabling the proof mass to displace both along and perpendicular to the sensitive axis. This configuration effectively reduces the axial forces in the shorter-arc-length GAS and mitigates the spring system’s negative stiffness effect. By optimizing the dimensions of GASs, the quasi-zero stiffness characteristic can be achieved. A low-cost, low-noise optical shadow displacement sensor is designed to detect the displacement of the proof mass. A 5 d measurement of Earth tides reveal that the experimental data has a correlation coefficient of 0.91 with the theoretical date from the TOSFT software. Final tests demonstrate that the micro-electro-mechanical system (MEMS) gravimeter achieves a noise power spectral density of 5 μGal √Hz–1 at 3 Hz.
- Top Cited PapersOpen Access
- 6 December 2024
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 10 (49) , eadr7155
Metasurface spectrometers beyond resolution-sensitivity constraints
Conventional spectrometer designs necessitate a compromise between their resolution and sensitivity, especially as device and detector dimensions are scaled down. Here, we report on a miniaturizable spectrometer platform where light throughput onto the detector is instead enhanced as the resolution is increased. This planar, CMOS-compatible platform is based around metasurface encoders designed to exhibit photonic bound states in the continuum, where operational range can be altered or extended simply through adjusting geometric parameters. This system can enhance photon collection efficiency by up to two orders of magnitude versus conventional designs; we demonstrate this sensitivity advantage through ultralow-intensity fluorescent and astrophotonic spectroscopy. This work represents a step forward for the practical utility of spectrometers, affording a route to integrated, chip-based devices that maintain high resolution and SNR without requiring prohibitively long integration times.
- 6 December 2024
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 10 (49) , eadr7155
Top Cited PapersOpen AccessMetasurface spectrometers beyond resolution-sensitivity constraints
Conventional spectrometer designs necessitate a compromise between their resolution and sensitivity, especially as device and detector dimensions are scaled down. Here, we report on a miniaturizable spectrometer platform where light throughput onto the detector is instead enhanced as the resolution is increased. This planar, CMOS-compatible platform is based around metasurface encoders designed to exhibit photonic bound states in the continuum, where operational range can be altered or extended simply through adjusting geometric parameters. This system can enhance photon collection efficiency by up to two orders of magnitude versus conventional designs; we demonstrate this sensitivity advantage through ultralow-intensity fluorescent and astrophotonic spectroscopy. This work represents a step forward for the practical utility of spectrometers, affording a route to integrated, chip-based devices that maintain high resolution and SNR without requiring prohibitively long integration times.