Ying Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Ying Chen has a total of 72 co-authors. Here are the top co-authors by number of shared publications:
- Yun Ge
School of Electronic Science and Engineering, Nanjing University, Nanjing, China
8 shared publications - Jie Yuan
School of Electronic Science and Engineering, Nanjing University, Nanjing, China
4 shared publications Weifeng Li
School of Electronic Science and Engineering, Nanjing University, Nanjing, China
4 shared publicationsXiaolin Huang
School of Electronic Science and Engineering, Nanjing University, Nanjing, China
4 shared publications- Xiaolin Huang
School of Electronic Science and Engineering, Nanjing University, Nanjing, China
4 shared publications - Weiyu Sun
School of Electronic Science and Engineering, Nanjing University, Nanjing, China
3 shared publications
Latest Publications
- 30 August 2026
- Published by Springer Nature in Network Modeling Analysis in Health Informatics and Bioinformatics
- Vol. 15 (1) , 191
MulGM-Net: A decision-based gated memory network with contrastive learning for multi-task cancer diagnosis
- 30 August 2026
- Published by Springer Nature in Network Modeling Analysis in Health Informatics and Bioinformatics
- Vol. 15 (1) , 191
MulGM-Net: A decision-based gated memory network with contrastive learning for multi-task cancer diagnosis
- Open Access
- 27 February 2026
- Published by MDPI AG in Diagnostics
- Vol. 16 (5) , 710
SBMN: Similarity-Based Memory Network for the Diagnosis of Vertical Root Fracture in Dental Imaging
Background/Objectives: Medical image analysis of vertical root fractures (VRFs) is challenged by limited annotated data, class imbalance, and subtle inter-class differences. To address these issues, we propose an SBMN: a Similarity-Based Memory Network that integrates Category Memory with the Basic SBMN Module and a similarity-based classifier. Methods: An SBMN stores representative features for each class and leverages similarity-based gating to enhance feature discrimination. Experiments were conducted on a CBCT dataset of fractured and non-fractured teeth to evaluate performance. Results: The SBMN achieved up to 97.1% and 99.7% classification accuracy on automatically and manually segmented images, respectively. Memory manipulation experiments confirm the critical role of Category Memory in controlling classification outcomes. Conclusions: These results indicate that SBMNs offer an effective and interpretable approach for small-sample medical image classification and diagnosis.
- 27 February 2026
- Published by MDPI AG in Diagnostics
- Vol. 16 (5) , 710
Open AccessSBMN: Similarity-Based Memory Network for the Diagnosis of Vertical Root Fracture in Dental Imaging
Background/Objectives: Medical image analysis of vertical root fractures (VRFs) is challenged by limited annotated data, class imbalance, and subtle inter-class differences. To address these issues, we propose an SBMN: a Similarity-Based Memory Network that integrates Category Memory with the Basic SBMN Module and a similarity-based classifier. Methods: An SBMN stores representative features for each class and leverages similarity-based gating to enhance feature discrimination. Experiments were conducted on a CBCT dataset of fractured and non-fractured teeth to evaluate performance. Results: The SBMN achieved up to 97.1% and 99.7% classification accuracy on automatically and manually segmented images, respectively. Memory manipulation experiments confirm the critical role of Category Memory in controlling classification outcomes. Conclusions: These results indicate that SBMNs offer an effective and interpretable approach for small-sample medical image classification and diagnosis.
- Open Access
- 21 February 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
IGMNN: a diagnosis method for vertical root fractures based on an information gated memory neural network
In medical image analysis, acquiring large-scale labeled datasets remains challenging, and images often exhibit high overall similarity that requires expert-level interpretation, differing substantially from natural image processing. To address these issues, we introduce the Information-Gated Memory (IGM) unit, a memory mechanism that enables deep networks to store and compare category-specific information. Unlike traditional CNNs or RNNs, the IGM unit performs memory-guided contrastive matching, allowing the network to focus on diagnostically relevant features and enhance classification performance. Using a CBCT dataset of 392 individuals, divided according to the presence or absence of artifacts, the proposed IGMNN achieved classification accuracies of \(97.3\%\) and \(93.9\%\), respectively.- 21 February 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessIGMNN: a diagnosis method for vertical root fractures based on an information gated memory neural network
In medical image analysis, acquiring large-scale labeled datasets remains challenging, and images often exhibit high overall similarity that requires expert-level interpretation, differing substantially from natural image processing. To address these issues, we introduce the Information-Gated Memory (IGM) unit, a memory mechanism that enables deep networks to store and compare category-specific information. Unlike traditional CNNs or RNNs, the IGM unit performs memory-guided contrastive matching, allowing the network to focus on diagnostically relevant features and enhance classification performance. Using a CBCT dataset of 392 individuals, divided according to the presence or absence of artifacts, the proposed IGMNN achieved classification accuracies of \(97.3\%\) and \(93.9\%\), respectively. - 17 July 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Letters
- Vol. 31 (10709908) , 2595-2599
Tackling the Non-IID Issue in Heterogeneous Federated Learning by Gradient Harmonization
Federated learning (FL) is a privacy-preserving paradigm for collaboratively training a global model from decentralized clients. However, the performance of FL is hindered by non-independent and identically distributed (non-IID) data and device heterogeneity. In this letter, we revisit this key challenge through the lens of gradient conflicts on the server side. Specifically, we first investigate the gradient conflict phenomenon among multiple clients and reveal that stronger heterogeneity leads to more severe gradient conflicts. To tackle this issue, we propose FedGH, a simple yet effective method that mitigates local drifts through Gradient Harmonization. This technique projects one gradient vector onto the orthogonal plane of the other within conflicting client pairs. Extensive experiments demonstrate that FedGH consistently enhances multiple state-of-the-art FL baselines across diverse benchmarks and non-IID scenarios. Moreover, FedGH yields more significant improvements in scenarios with stronger heterogeneity. As a plug-and-play module, FedGH can seamlessly integrate into any FL framework without requiring hyperparameter tuning.
- 17 July 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Letters
- Vol. 31 (10709908) , 2595-2599
Tackling the Non-IID Issue in Heterogeneous Federated Learning by Gradient Harmonization
Federated learning (FL) is a privacy-preserving paradigm for collaboratively training a global model from decentralized clients. However, the performance of FL is hindered by non-independent and identically distributed (non-IID) data and device heterogeneity. In this letter, we revisit this key challenge through the lens of gradient conflicts on the server side. Specifically, we first investigate the gradient conflict phenomenon among multiple clients and reveal that stronger heterogeneity leads to more severe gradient conflicts. To tackle this issue, we propose FedGH, a simple yet effective method that mitigates local drifts through Gradient Harmonization. This technique projects one gradient vector onto the orthogonal plane of the other within conflicting client pairs. Extensive experiments demonstrate that FedGH consistently enhances multiple state-of-the-art FL baselines across diverse benchmarks and non-IID scenarios. Moreover, FedGH yields more significant improvements in scenarios with stronger heterogeneity. As a plug-and-play module, FedGH can seamlessly integrate into any FL framework without requiring hyperparameter tuning.
- 30 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. 26 (15209210) , 10290-10305
Self-Similarity Prior Distillation for Unsupervised Remote Physiological Measurement
Remote photoplethysmography (rPPG) is a non-invasive technique that aims to capture subtle variations in facial pixels caused by changes in blood volume resulting from cardiac activities. Most existing unsupervised methods for rPPG tasks focus on the contrastive learning between samples while neglecting the inherent self-similarity prior in physiological signals. In this paper, we propose a Self-Similarity Prior Distillation (SSPD) framework for unsupervised rPPG estimation, which capitalizes on the intrinsic temporal self-similarity of cardiac activities. Specifically, we first introduce a physical-prior embedded augmentation technique to mitigate the effect of various types of noise. Then, we tailor a self-similarity-aware network to disentangle more reliable self-similar physiological features. Finally, we develop a hierarchical self-distillation paradigm for self-similarity-aware learning and rPPG signal decoupling. Comprehensive experiments demonstrate that the unsupervised SSPD framework achieves comparable or even superior performance compared to the state-of-the-art supervised methods. Meanwhile, SSPD has the lowest inference time and computation cost among end-to-end models.
- 30 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. 26 (15209210) , 10290-10305
Self-Similarity Prior Distillation for Unsupervised Remote Physiological Measurement
Remote photoplethysmography (rPPG) is a non-invasive technique that aims to capture subtle variations in facial pixels caused by changes in blood volume resulting from cardiac activities. Most existing unsupervised methods for rPPG tasks focus on the contrastive learning between samples while neglecting the inherent self-similarity prior in physiological signals. In this paper, we propose a Self-Similarity Prior Distillation (SSPD) framework for unsupervised rPPG estimation, which capitalizes on the intrinsic temporal self-similarity of cardiac activities. Specifically, we first introduce a physical-prior embedded augmentation technique to mitigate the effect of various types of noise. Then, we tailor a self-similarity-aware network to disentangle more reliable self-similar physiological features. Finally, we develop a hierarchical self-distillation paradigm for self-similarity-aware learning and rPPG signal decoupling. Comprehensive experiments demonstrate that the unsupervised SSPD framework achieves comparable or even superior performance compared to the state-of-the-art supervised methods. Meanwhile, SSPD has the lowest inference time and computation cost among end-to-end models.
- 1 November 2023
- Published by SPIE-Intl Soc Optical Eng in Journal of Medical Imaging
- Vol. 10 (06) , 066002-14
Characterization of anisotropy of elastic modulus with three-dimensional freehand scan shear wave elasticity imaging
PurposeThe purpose of this study is to develop a freehand scan three-dimensional (3D) shear wave elasticity imaging (SWEI) method for characterizing the anisotropy of elastic properties in biological tissues. The motivation behind this work lies in addressing the limitations of traditional two-dimensional (2D) SWEI, which only measures shear wave speeds in a single direction, as well as fulfilling the clinical demand for improved medical imaging.ApproachOur imaging system utilizes a high-definition optical camera to continuously track the ultrasonic transducer, collecting spatial position-angle data of the transducer and corresponding two-dimensional SWEI data. By reconstructing three-dimensional SWEI images using these data, we achieved freehand SWEI.ResultsWe validated the accuracy of 2D SWEI on a standard elastic phantom, and then performed 3D SWEI on the pork tenderloin and the triceps brachii of two volunteers. We obtained shear wave speed of 1.82 to 3.12 m / s in the pork tenderloin, shear wave speed of 1.16 to 2.36 m / s in the triceps brachii of non-exercising volunteers, and shear wave speed of 0.55 to 1.63 m / s in the triceps brachii of exercising volunteers, and the maximum shear wave speed directions were generally aligned with the orientation of muscle fibers.ConclusionsWe proposed a method that can overcome the limitations of 2D-SWEI regarding imaging angle while also extending the imaging angle of 3D-SWEI, which could have significant implications for improving the accuracy and safety of medical diagnoses.
- 1 November 2023
- Published by SPIE-Intl Soc Optical Eng in Journal of Medical Imaging
- Vol. 10 (06) , 066002-14
Characterization of anisotropy of elastic modulus with three-dimensional freehand scan shear wave elasticity imaging
PurposeThe purpose of this study is to develop a freehand scan three-dimensional (3D) shear wave elasticity imaging (SWEI) method for characterizing the anisotropy of elastic properties in biological tissues. The motivation behind this work lies in addressing the limitations of traditional two-dimensional (2D) SWEI, which only measures shear wave speeds in a single direction, as well as fulfilling the clinical demand for improved medical imaging.ApproachOur imaging system utilizes a high-definition optical camera to continuously track the ultrasonic transducer, collecting spatial position-angle data of the transducer and corresponding two-dimensional SWEI data. By reconstructing three-dimensional SWEI images using these data, we achieved freehand SWEI.ResultsWe validated the accuracy of 2D SWEI on a standard elastic phantom, and then performed 3D SWEI on the pork tenderloin and the triceps brachii of two volunteers. We obtained shear wave speed of 1.82 to 3.12 m / s in the pork tenderloin, shear wave speed of 1.16 to 2.36 m / s in the triceps brachii of non-exercising volunteers, and shear wave speed of 0.55 to 1.63 m / s in the triceps brachii of exercising volunteers, and the maximum shear wave speed directions were generally aligned with the orientation of muscle fibers.ConclusionsWe proposed a method that can overcome the limitations of 2D-SWEI regarding imaging angle while also extending the imaging angle of 3D-SWEI, which could have significant implications for improving the accuracy and safety of medical diagnoses.
- Open Access
- 26 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 8214-8224
Resolve Domain Conflicts for Generalizable Remote Physiological Measurement
- 26 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 8214-8224
Open AccessResolve Domain Conflicts for Generalizable Remote Physiological Measurement
- Open Access
- 20 June 2023
- Published by MDPI AG in Applied Sciences
- Vol. 13 (12) , 7317
Temperature-Controlled Hyperthermia with Non-Invasive Temperature Monitoring through Speed of Sound Imaging
Hyperthermia therapy (HT) is used to treat diseases through heating of high temperature usually in conjunction with some other medical therapeutics such as chemotherapy and radiotherapy. In this study, we propose a promising temperature-controlled hyperthermia method that uses high-intensity focused ultrasound (HIFU) for clinical tumor treatment combined with diagnostic ultrasound image guidance and non-invasive temperature monitoring through speed of sound (SOS) imaging. HIFU heating is realized by a ring ultrasound transducer array with 256 elements. In this study, tumors in the human thigh were set as heating targets. The inner structure information of thigh tissue is obtained by B-mode ultrasound imaging. Since the relationship between temperature and SOS in different human tissue is available, the temperature detection is converted to the SOS detection obtained by the full-wave inversion (FWI) method. Simulation results show that our model can achieve expected hyperthermia of constant temperature on tumor target with 0.2 °C maximum temperature fluctuation for 5 h. Through simulation, our proposed thermal therapy model achieves accurate temperature control of ±0.2 °C in human thigh tumors, which verifies the feasibility of the proposed temperature-controlled hyperthermia model. Furthermore, the temperature measurement can share the same ring ultrasound transducer array for HIFU heating and B-mode ultrasound imaging, which provides a guiding significance for clinical application.
- 20 June 2023
- Published by MDPI AG in Applied Sciences
- Vol. 13 (12) , 7317
Open AccessTemperature-Controlled Hyperthermia with Non-Invasive Temperature Monitoring through Speed of Sound Imaging
Hyperthermia therapy (HT) is used to treat diseases through heating of high temperature usually in conjunction with some other medical therapeutics such as chemotherapy and radiotherapy. In this study, we propose a promising temperature-controlled hyperthermia method that uses high-intensity focused ultrasound (HIFU) for clinical tumor treatment combined with diagnostic ultrasound image guidance and non-invasive temperature monitoring through speed of sound (SOS) imaging. HIFU heating is realized by a ring ultrasound transducer array with 256 elements. In this study, tumors in the human thigh were set as heating targets. The inner structure information of thigh tissue is obtained by B-mode ultrasound imaging. Since the relationship between temperature and SOS in different human tissue is available, the temperature detection is converted to the SOS detection obtained by the full-wave inversion (FWI) method. Simulation results show that our model can achieve expected hyperthermia of constant temperature on tumor target with 0.2 °C maximum temperature fluctuation for 5 h. Through simulation, our proposed thermal therapy model achieves accurate temperature control of ±0.2 °C in human thigh tumors, which verifies the feasibility of the proposed temperature-controlled hyperthermia model. Furthermore, the temperature measurement can share the same ring ultrasound transducer array for HIFU heating and B-mode ultrasound imaging, which provides a guiding significance for clinical application.
- Open Access
- 5 September 2022
- Published by Springer Nature in BMC Oral Health
- Vol. 22 (1) , 1-9
Diagnosis of in vivo vertical root fracture using deep learning on cone-beam CT images
Objectives: Evaluating the diagnostic efficiency of deep learning models to diagnose vertical root fracture in vivo on cone-beam CT (CBCT) images. Materials and methods: The CBCT images of 276 teeth (138 VRF teeth and 138 non-VRF teeth) were enrolled and analyzed retrospectively. The diagnostic results of these teeth were confirmed by two chief radiologists. There were two experimental groups: auto-selection group and manual selection group. A total of 552 regions of interest of teeth were cropped in manual selection group and 1118 regions of interest of teeth were cropped in auto-selection group. Three deep learning networks (ResNet50, VGG19 and DenseNet169) were used for diagnosis (3:1 for training and testing). The diagnostic efficiencies (accuracy, sensitivity, specificity, and area under the curve (AUC)) of three networks were calculated in two experiment groups. Meanwhile, 552 teeth images in manual selection group were diagnosed by a radiologist. The diagnostic efficiencies of the three deep learning network models in two experiment groups and the radiologist were calculated. Results: In manual selection group, ResNet50 presented highest accuracy and sensitivity for diagnosing VRF teeth. The accuracy, sensitivity, specificity and AUC was 97.8%, 97.0%, 98.5%, and 0.99, the radiologist presented accuracy, sensitivity, and specificity as 95.3%, 96.4 and 94.2%. In auto-selection group, ResNet50 presented highest accuracy and sensitivity for diagnosing VRF teeth, the accuracy, sensitivity, specificity and AUC was 91.4%, 92.1%, 90.7% and 0.96. Conclusion: In manual selection group, ResNet50 presented higher diagnostic efficiency in diagnosis of in vivo VRF teeth than VGG19, DensenNet169 and radiologist with 2 years of experience. In auto-selection group, Resnet50 also presented higher diagnostic efficiency in diagnosis of in vivo VRF teeth than VGG19 and DensenNet169. This makes it a promising auxiliary diagnostic technique to screen for VRF teeth.
- 5 September 2022
- Published by Springer Nature in BMC Oral Health
- Vol. 22 (1) , 1-9
Open AccessDiagnosis of in vivo vertical root fracture using deep learning on cone-beam CT images
Objectives: Evaluating the diagnostic efficiency of deep learning models to diagnose vertical root fracture in vivo on cone-beam CT (CBCT) images. Materials and methods: The CBCT images of 276 teeth (138 VRF teeth and 138 non-VRF teeth) were enrolled and analyzed retrospectively. The diagnostic results of these teeth were confirmed by two chief radiologists. There were two experimental groups: auto-selection group and manual selection group. A total of 552 regions of interest of teeth were cropped in manual selection group and 1118 regions of interest of teeth were cropped in auto-selection group. Three deep learning networks (ResNet50, VGG19 and DenseNet169) were used for diagnosis (3:1 for training and testing). The diagnostic efficiencies (accuracy, sensitivity, specificity, and area under the curve (AUC)) of three networks were calculated in two experiment groups. Meanwhile, 552 teeth images in manual selection group were diagnosed by a radiologist. The diagnostic efficiencies of the three deep learning network models in two experiment groups and the radiologist were calculated. Results: In manual selection group, ResNet50 presented highest accuracy and sensitivity for diagnosing VRF teeth. The accuracy, sensitivity, specificity and AUC was 97.8%, 97.0%, 98.5%, and 0.99, the radiologist presented accuracy, sensitivity, and specificity as 95.3%, 96.4 and 94.2%. In auto-selection group, ResNet50 presented highest accuracy and sensitivity for diagnosing VRF teeth, the accuracy, sensitivity, specificity and AUC was 91.4%, 92.1%, 90.7% and 0.96. Conclusion: In manual selection group, ResNet50 presented higher diagnostic efficiency in diagnosis of in vivo VRF teeth than VGG19, DensenNet169 and radiologist with 2 years of experience. In auto-selection group, Resnet50 also presented higher diagnostic efficiency in diagnosis of in vivo VRF teeth than VGG19 and DensenNet169. This makes it a promising auxiliary diagnostic technique to screen for VRF teeth.
- Open Access
- 1 August 2022
- Published by Frontiers Media SA in Frontiers in Oncology
- Vol. 12, 919088
Using deep learning to distinguish malignant from benign parotid tumors on plain computed tomography images
Objectives: Evaluating the diagnostic efficiency of deep-learning models to distinguish malignant from benign parotid tumors on plain computed tomography (CT) images.Materials and methods: The CT images of 283 patients with parotid tumors were enrolled and analyzed retrospectively. Of them, 150 were benign and 133 were malignant according to pathology results. A total of 917 regions of interest of parotid tumors were cropped (456 benign and 461 malignant). Three deep-learning networks (ResNet50, VGG16_bn, and DenseNet169) were used for diagnosis (approximately 3:1 for training and testing). The diagnostic efficiencies (accuracy, sensitivity, specificity, and area under the curve [AUC]) of three networks were calculated and compared based on the 917 images. To simulate the process of human diagnosis, a voting model was developed at the end of the networks and the 283 tumors were classified as benign or malignant. Meanwhile, 917 tumor images were classified by two radiologists (A and B) and original CT images were classified by radiologist B. The diagnostic efficiencies of the three deep-learning network models (after voting) and the two radiologists were calculated.Results: For the 917 CT images, ResNet50 presented high accuracy and sensitivity for diagnosing malignant parotid tumors; the accuracy, sensitivity, specificity, and AUC were 90.8%, 91.3%, 90.4%, and 0.96, respectively. For the 283 tumors, the accuracy, sensitivity, and specificity of ResNet50 (after voting) were 92.3%, 93.5% and 91.2%, respectively.Conclusion: ResNet50 presented high sensitivity in distinguishing malignant from benign parotid tumors on plain CT images; this made it a promising auxiliary diagnostic method to screen malignant parotid tumors.
- 1 August 2022
- Published by Frontiers Media SA in Frontiers in Oncology
- Vol. 12, 919088
Open AccessUsing deep learning to distinguish malignant from benign parotid tumors on plain computed tomography images
Objectives: Evaluating the diagnostic efficiency of deep-learning models to distinguish malignant from benign parotid tumors on plain computed tomography (CT) images.Materials and methods: The CT images of 283 patients with parotid tumors were enrolled and analyzed retrospectively. Of them, 150 were benign and 133 were malignant according to pathology results. A total of 917 regions of interest of parotid tumors were cropped (456 benign and 461 malignant). Three deep-learning networks (ResNet50, VGG16_bn, and DenseNet169) were used for diagnosis (approximately 3:1 for training and testing). The diagnostic efficiencies (accuracy, sensitivity, specificity, and area under the curve [AUC]) of three networks were calculated and compared based on the 917 images. To simulate the process of human diagnosis, a voting model was developed at the end of the networks and the 283 tumors were classified as benign or malignant. Meanwhile, 917 tumor images were classified by two radiologists (A and B) and original CT images were classified by radiologist B. The diagnostic efficiencies of the three deep-learning network models (after voting) and the two radiologists were calculated.Results: For the 917 CT images, ResNet50 presented high accuracy and sensitivity for diagnosing malignant parotid tumors; the accuracy, sensitivity, specificity, and AUC were 90.8%, 91.3%, 90.4%, and 0.96, respectively. For the 283 tumors, the accuracy, sensitivity, and specificity of ResNet50 (after voting) were 92.3%, 93.5% and 91.2%, respectively.Conclusion: ResNet50 presented high sensitivity in distinguishing malignant from benign parotid tumors on plain CT images; this made it a promising auxiliary diagnostic method to screen malignant parotid tumors.