Lei Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Lei Zhang has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
- Huiliang Shang
School of Information Science and Technology, Fudan University, Shanghai, China
3 shared publications - Haitao Wu
ENT institute and Department of Otorhinolaryngology, Eye & ENT Hospital, Fudan University, Shanghai, China
2 shared publications - Lingjie Wu
ENT Institute and Department of Otorhinolaryngology, Eye & ENT Hospital Fudan University Shanghai China
2 shared publications - Yandan Lin
Academy for Engineering and Technology Fudan University Shanghai China
2 shared publications - Liangzhuang Wei
Academy for Engineering and Technology, Fudan University, Handan 220, Shanghai 200433, China
1 shared publication - Rong Cao
Fujian Institute of Research on the Structure of Matter Chinese Academy of Sciences Fujian, Fuzhou P.R. China
1 shared publication
Latest Publications
- Open Access
- 17 March 2023
- Published by MDPI AG in Diagnostics
- Vol. 13 (6) , 1151
A Novel Framework of Manifold Learning Cascade-Clustering for the Informative Frame Selection
Narrow band imaging is an established non-invasive tool used for the early detection of laryngeal cancer in surveillance examinations. Most images produced from the examination are useless, such as blurred, specular reflection, and underexposed. Removing the uninformative frames is vital to improve detection accuracy and speed up computer-aided diagnosis. It often takes a lot of time for the physician to manually inspect the informative frames. This issue is commonly addressed by a classifier with task-specific categories of the uninformative frames. However, the definition of the uninformative categories is ambiguous, and tedious labeling still cannot be avoided. Here, we show that a novel unsupervised scheme is comparable to the current benchmarks on the dataset of NBI-InfFrames. We extract feature embedding using a vanilla neural network (VGG16) and introduce a new dimensionality reduction method called UMAP that distinguishes the feature embedding in the lower-dimensional space. Along with the proposed automatic cluster labeling algorithm and cost function in Bayesian optimization, the proposed method coupled with UMAP achieves state-of-the-art performance. It outperforms the baseline by 12% absolute. The overall median recall of the proposed method is currently the highest, 96%. Our results demonstrate the effectiveness of the proposed scheme and the robustness of detecting the informative frames. It also suggests the patterns embedded in the data help develop flexible algorithms that do not require manual labeling.
- 17 March 2023
- Published by MDPI AG in Diagnostics
- Vol. 13 (6) , 1151
Open AccessA Novel Framework of Manifold Learning Cascade-Clustering for the Informative Frame Selection
Narrow band imaging is an established non-invasive tool used for the early detection of laryngeal cancer in surveillance examinations. Most images produced from the examination are useless, such as blurred, specular reflection, and underexposed. Removing the uninformative frames is vital to improve detection accuracy and speed up computer-aided diagnosis. It often takes a lot of time for the physician to manually inspect the informative frames. This issue is commonly addressed by a classifier with task-specific categories of the uninformative frames. However, the definition of the uninformative categories is ambiguous, and tedious labeling still cannot be avoided. Here, we show that a novel unsupervised scheme is comparable to the current benchmarks on the dataset of NBI-InfFrames. We extract feature embedding using a vanilla neural network (VGG16) and introduce a new dimensionality reduction method called UMAP that distinguishes the feature embedding in the lower-dimensional space. Along with the proposed automatic cluster labeling algorithm and cost function in Bayesian optimization, the proposed method coupled with UMAP achieves state-of-the-art performance. It outperforms the baseline by 12% absolute. The overall median recall of the proposed method is currently the highest, 96%. Our results demonstrate the effectiveness of the proposed scheme and the robustness of detecting the informative frames. It also suggests the patterns embedded in the data help develop flexible algorithms that do not require manual labeling.
- Open Access
A Lightweight Subgraph-Based Deep Learning Approach for Fall Recognition
Falls pose a great danger to social development, especially to the elderly population. When a fall occurs, the body’s center of gravity moves from a high position to a low position, and the magnitude of change varies among body parts. Most existing fall recognition methods based on deep learning have not yet considered the differences between the movement and the change in amplitude of each body part. Besides, some problems exist such as complicated design, slow detection speed, and lack of timeliness. To alleviate these problems, a lightweight subgraph-based deep learning method utilizing skeleton information for fall recognition is proposed in this paper. The skeleton information of the human body is extracted by OpenPose, and an end-to-end lightweight subgraph-based network is designed. Sub-graph division and sub-graph attention modules are introduced to add a larger perceptual field while maintaining its lightweight characteristics. A multi-scale temporal convolution module is also designed to extract and fuse multi-scale temporal features, which enriches the feature representation. The proposed method is evaluated on a partial fall dataset collected in NTU and on two public datasets, and outperforms existing methods. It indicates that the proposed method is accurate and lightweight, which means it is suitable for real-time detection and rapid response to falls.
Open AccessA Lightweight Subgraph-Based Deep Learning Approach for Fall Recognition
Falls pose a great danger to social development, especially to the elderly population. When a fall occurs, the body’s center of gravity moves from a high position to a low position, and the magnitude of change varies among body parts. Most existing fall recognition methods based on deep learning have not yet considered the differences between the movement and the change in amplitude of each body part. Besides, some problems exist such as complicated design, slow detection speed, and lack of timeliness. To alleviate these problems, a lightweight subgraph-based deep learning method utilizing skeleton information for fall recognition is proposed in this paper. The skeleton information of the human body is extracted by OpenPose, and an end-to-end lightweight subgraph-based network is designed. Sub-graph division and sub-graph attention modules are introduced to add a larger perceptual field while maintaining its lightweight characteristics. A multi-scale temporal convolution module is also designed to extract and fuse multi-scale temporal features, which enriches the feature representation. The proposed method is evaluated on a partial fall dataset collected in NTU and on two public datasets, and outperforms existing methods. It indicates that the proposed method is accurate and lightweight, which means it is suitable for real-time detection and rapid response to falls.
- Open Access
- 13 January 2022
- Published by MDPI AG in Micromachines
- Vol. 13 (1) , 126
A Novel Distribution for Representation of 6D Pose Uncertainty
The 6D Pose estimation is a crux in many applications, such as visual perception, autonomous navigation, and spacecraft motion. For robotic grasping, the cluttered and self-occlusion scenarios bring new challenges to the this field. Currently, society uses CNNs to solve this problem. The CNN models will suffer high uncertainty caused by the environmental factors and the object itself. These models usually maintain a Gaussian distribution, which is not suitable for the underlying manifold structure of the pose. Many works decouple rotation from the translation and quantify rotational uncertainty. Only a few works pay attention to the uncertainty of the 6D pose. This work proposes a distribution that can capture the uncertainty of the 6D pose parameterized by the dual quaternions, meanwhile, the proposed distribution takes the periodic nature of the underlying structure into account. The presented results include the normalization constant computation and parameter estimation techniques of the distribution. This work shows the benefits of the proposed distribution, which provides a more realistic explanation for the uncertainty in the 6D pose and eliminates the drawback inherited from the planar rigid motion.
- 13 January 2022
- Published by MDPI AG in Micromachines
- Vol. 13 (1) , 126
Open AccessA Novel Distribution for Representation of 6D Pose Uncertainty
The 6D Pose estimation is a crux in many applications, such as visual perception, autonomous navigation, and spacecraft motion. For robotic grasping, the cluttered and self-occlusion scenarios bring new challenges to the this field. Currently, society uses CNNs to solve this problem. The CNN models will suffer high uncertainty caused by the environmental factors and the object itself. These models usually maintain a Gaussian distribution, which is not suitable for the underlying manifold structure of the pose. Many works decouple rotation from the translation and quantify rotational uncertainty. Only a few works pay attention to the uncertainty of the 6D pose. This work proposes a distribution that can capture the uncertainty of the 6D pose parameterized by the dual quaternions, meanwhile, the proposed distribution takes the periodic nature of the underlying structure into account. The presented results include the normalization constant computation and parameter estimation techniques of the distribution. This work shows the benefits of the proposed distribution, which provides a more realistic explanation for the uncertainty in the 6D pose and eliminates the drawback inherited from the planar rigid motion.
- 1 November 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 123-128
Deep Convolutional Neural Networks for Accurate Diagnosis of Nasopharyngeal Carcinoma in pCLE Images
Objectives/Hypothesis: Probe-based confocal laser endomicroscopy(pCLE) can be used for the real-time cell-level optical biopsy of nasopharyngeal lesions in vivo. The development and validation of deep-learning-based automatic diagnosis of pCLE images are essential to the rapid screening and detection of nasopharyngeal carcinoma. Methods: From May 1, 2017, to November 30, 2017, the pCLE images were consecutively collected as samples in data sets from 31 patients with highly suspected NPC. The pCLE images were divided into training, cross-validation, and test sets to train, validate and test the proposed model. The proposed model's accuracy, sensitivity, and specificity are using pathological diagnosis as the gold standard. Analysis of the data is from May 15, 2019, to May 31, 2020. Results: The data set contains 3623 pCLE images in total. Eighty percent of the images were selected randomly as the training set, then another 10% as the validation set, and the last 10% as the test set. We proposed a deep learning model based on ResNet50. On the test data set, the model achieved an overall detection accuracy of 94.7%. The proposed model has higher accuracy, over 20.8%, compared to the best traditional supervised machine learning. Conclusions: Our algorithm automatically detected the nasopharyngeal carcinoma through screening images, which conduces to the otolaryngologists in their diagnosis. This deep-learning-based framework will be of great importance in the rapid screening and detecting of NPC.
- 1 November 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 123-128
Deep Convolutional Neural Networks for Accurate Diagnosis of Nasopharyngeal Carcinoma in pCLE Images
Objectives/Hypothesis: Probe-based confocal laser endomicroscopy(pCLE) can be used for the real-time cell-level optical biopsy of nasopharyngeal lesions in vivo. The development and validation of deep-learning-based automatic diagnosis of pCLE images are essential to the rapid screening and detection of nasopharyngeal carcinoma. Methods: From May 1, 2017, to November 30, 2017, the pCLE images were consecutively collected as samples in data sets from 31 patients with highly suspected NPC. The pCLE images were divided into training, cross-validation, and test sets to train, validate and test the proposed model. The proposed model's accuracy, sensitivity, and specificity are using pathological diagnosis as the gold standard. Analysis of the data is from May 15, 2019, to May 31, 2020. Results: The data set contains 3623 pCLE images in total. Eighty percent of the images were selected randomly as the training set, then another 10% as the validation set, and the last 10% as the test set. We proposed a deep learning model based on ResNet50. On the test data set, the model achieved an overall detection accuracy of 94.7%. The proposed model has higher accuracy, over 20.8%, compared to the best traditional supervised machine learning. Conclusions: Our algorithm automatically detected the nasopharyngeal carcinoma through screening images, which conduces to the otolaryngologists in their diagnosis. This deep-learning-based framework will be of great importance in the rapid screening and detecting of NPC.