Xiaofeng Xu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Xiaofeng Xu has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
- Fei Ning
School of Life Sciences and Agricultural Engineering, Nanyang Normal University, Wolong, Nanyang, China
1 shared publication He Li
Henan Engineering Research Center of Intelligent Processing for Big Data of Digital Image, School of Computer Science and Technology, Nanyang Normal University, Nanyang, China
1 shared publication- He Li
School of Artificial Intelligence and Software Engineering, Nanyang Normal University, Nanyang, China
1 shared publication Jiangdan Shan
Henan Engineering Research Center of Intelligent Processing for Big Data of Digital Image, School of Computer Science and Technology, Nanyang Normal University, Nanyang, China
1 shared publication- Qinglei Qi
School of Artificial Intelligence and Software Engineering, Nanyang Normal University, Nanyang, China
1 shared publication - Quan Liu
Affiliations School of Artificial Intelligence and Software Engineering, Nanyang Normal University, Nanyang, China, Henan Provincial Engineering Research Center for Image Big Data Intelligent Processing, Nanyang, China
1 shared publication
Latest Publications
- Open Access
- 3 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 123236-123249
Survival Prediction for Non-Small Cell Lung Cancer Based on Multimodal Fusion and Deep Learning
Non-small cell Lung Cancer (NSCLC) is one of the most common types of lung cancer, accounting for approximately 80% to 85% of lung cancer cases, and survival prediction in this cancer is an essential medical task. Traditional survival prediction methods, which only rely on demographic and clinical variables, fail to fully characterize patients’ pathological characteristics and clinical factors, thus limiting the prediction effect. With the development of medical imaging technology and genomic methods, survival prediction studies have ushered in new analytical perspectives in recent years. However, most of the existing advanced techniques only rely on one class and few classes of medical data, which do not comprehensively characterize patients’ pathological features and clinical conditions, and similarly limit the predictive effect. To this end, in this paper a new method is proposed to solve this problem, that is, using flexible interpretable graph structure to fuse and model the multimodal data (including clinical data and CT images, etc.) of patients, so as to solve the problem of fragmentation and one-sidedness among multi-class data of survival prediction. A new multi-modal fusion graph convolutional network (FGCN) is proposed according to the characteristics of multi-modal graph data. The main body of the network structure is composed of a SAGE graph convolution layer with the self-attention mechanism, which accelerates the convergence of the model compared to the traditional one and ensures that the risk predicted by the model is as close to the actual situation as possible. More importantly, this study is the first to introduce the TopKPooling strategy in this model to address the problems of feature redundancy and excessive noise features arising from multimodal data fusion and to reduce the model complexity. A large number of ablation and comparison experiments and analyses on public non-small cell lung cancer datasets show that the method proposed in this paper achieves better prediction results, with a C-index value of 0.76, which exceeds the currently known advanced techniques, and fully proves the effectiveness and superiority of the method in this paper.
- 3 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 123236-123249
Open AccessSurvival Prediction for Non-Small Cell Lung Cancer Based on Multimodal Fusion and Deep Learning
Non-small cell Lung Cancer (NSCLC) is one of the most common types of lung cancer, accounting for approximately 80% to 85% of lung cancer cases, and survival prediction in this cancer is an essential medical task. Traditional survival prediction methods, which only rely on demographic and clinical variables, fail to fully characterize patients’ pathological characteristics and clinical factors, thus limiting the prediction effect. With the development of medical imaging technology and genomic methods, survival prediction studies have ushered in new analytical perspectives in recent years. However, most of the existing advanced techniques only rely on one class and few classes of medical data, which do not comprehensively characterize patients’ pathological features and clinical conditions, and similarly limit the predictive effect. To this end, in this paper a new method is proposed to solve this problem, that is, using flexible interpretable graph structure to fuse and model the multimodal data (including clinical data and CT images, etc.) of patients, so as to solve the problem of fragmentation and one-sidedness among multi-class data of survival prediction. A new multi-modal fusion graph convolutional network (FGCN) is proposed according to the characteristics of multi-modal graph data. The main body of the network structure is composed of a SAGE graph convolution layer with the self-attention mechanism, which accelerates the convergence of the model compared to the traditional one and ensures that the risk predicted by the model is as close to the actual situation as possible. More importantly, this study is the first to introduce the TopKPooling strategy in this model to address the problems of feature redundancy and excessive noise features arising from multimodal data fusion and to reduce the model complexity. A large number of ablation and comparison experiments and analyses on public non-small cell lung cancer datasets show that the method proposed in this paper achieves better prediction results, with a C-index value of 0.76, which exceeds the currently known advanced techniques, and fully proves the effectiveness and superiority of the method in this paper.
- Open Access
- 22 September 2023
- Published by Association for Computing Machinery (ACM)
- p. 346-350
Automatic Segmentation Method for the Nuclei Contour of High Resolution Rat Brain
The brain consists of massive nuclei with different functions. In neuroscience research, the precise recognition and delineation of nucleus boundaries is the crux of brain atlas illustration. Here, we propose a method based on Markov Random Field. This method introduces fractional differentiation into texture feature extraction and a new potential energy function is defined based on kernel region information. To handle the problem of large data volume, we propose a strategy of dual MRF processing with down sampling for pre classification. Finally, the fuzzy entropy criterion is used to optimize the segmentation results. Experiments on model data and real tissue slice data show that this method can not only segment the regions with obvious differences in the density of the nuclei with insignificant differences in the density of nuclei, but also divide the nuclei without insignificant density differences by the cell shapes and textures. In addition, this method can not only segment a specific target kernel in one calculation, but also simultaneously partition multiple target kernels. The segmentation algorithm shows great potential for illustrating high-resolution 3D brain atlases.
- 22 September 2023
- Published by Association for Computing Machinery (ACM)
- p. 346-350
Open AccessAutomatic Segmentation Method for the Nuclei Contour of High Resolution Rat Brain
The brain consists of massive nuclei with different functions. In neuroscience research, the precise recognition and delineation of nucleus boundaries is the crux of brain atlas illustration. Here, we propose a method based on Markov Random Field. This method introduces fractional differentiation into texture feature extraction and a new potential energy function is defined based on kernel region information. To handle the problem of large data volume, we propose a strategy of dual MRF processing with down sampling for pre classification. Finally, the fuzzy entropy criterion is used to optimize the segmentation results. Experiments on model data and real tissue slice data show that this method can not only segment the regions with obvious differences in the density of the nuclei with insignificant differences in the density of nuclei, but also divide the nuclei without insignificant density differences by the cell shapes and textures. In addition, this method can not only segment a specific target kernel in one calculation, but also simultaneously partition multiple target kernels. The segmentation algorithm shows great potential for illustrating high-resolution 3D brain atlases.
- Open Access
- 23 September 2022
- Published by Association for Computing Machinery (ACM)
- p. 178-182
An Automatic Method for Fast Synaptic Signal Detection and Count
Synapses are the special connection structure between brain neurons, and they are the core element of brain plasticity and information transmission. In order to better understand the process of brain information transmission, it is necessary to quantitatively calculate the synapse, but the automatic identification of a large range of synaptic images is still a difficult problem. According to the characteristics of synaptic signal in microscopy optical mouse brain images, this paper proposes the automatic fast synapse recognition method based on density peak clustering, which solves the identification problem of dense small synaptic signal, and realizes the rapid synapse positioning and counting in a large brain area. The results show that the proposed methods in accuracy and efficiency bring great convenience to the researchers. CCS CONCEPTS•Applied computing
- 23 September 2022
- Published by Association for Computing Machinery (ACM)
- p. 178-182
Open AccessAn Automatic Method for Fast Synaptic Signal Detection and Count
Synapses are the special connection structure between brain neurons, and they are the core element of brain plasticity and information transmission. In order to better understand the process of brain information transmission, it is necessary to quantitatively calculate the synapse, but the automatic identification of a large range of synaptic images is still a difficult problem. According to the characteristics of synaptic signal in microscopy optical mouse brain images, this paper proposes the automatic fast synapse recognition method based on density peak clustering, which solves the identification problem of dense small synaptic signal, and realizes the rapid synapse positioning and counting in a large brain area. The results show that the proposed methods in accuracy and efficiency bring great convenience to the researchers. CCS CONCEPTS•Applied computing