Xin Shu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Xin Shu has a total of 32 co-authors. Here are the top co-authors by number of shared publications:
- Lei Zhang
College of Computer Science, Chengdu University, Chengdu 610106, China
15 shared publications Zizhou Wang
Institute of High Performance Computing, Agency for Science, Technology and Research (A*STAR), Singapore
8 shared publications- Zhang Yi
Sichuan University
6 shared publications - Lituan Wang
Machine Intelligence Laboratory, College of Computer Science, Sichuan University, Chengdu, China
3 shared publications Qing Lv
Department of Galactophore Surgery, West China Hospital, Sichuan University, Chengdu, Sichuan, PR China
3 shared publications- Wei Huang
Machine Intelligence Laboratory, College of Computer Science, Sichuan University, Chengdu, China
3 shared publications
Latest Publications
- 28 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 28 (12) , 7175-7183
Adaptive Annotation Correlation Based Multi-Annotation Learning for Calibrated Medical Image Segmentation
Medical image segmentation is a fundamental task in many clinical applications, yet current automated segmentation methods rely heavily on manual annotations, which are inherently subjective and prone to annotation bias. Recently, modeling annotator preference has garnered great interest, and several methods have been proposed in the past two years. However, the existing methods completely ignore the potential correlation between annotations, such as complementary and discriminative information. In this work, the A daptive annotation C orrela T ion based mult I -ann O tation Lear N ing ( ACTION ) method is proposed for calibrated medical image segmentation. ACTION employs consensus feature learning and dynamic adaptive weighting to leverage complementary information across annotations and emphasize discriminative information within each annotation based on their correlations, respectively. Meanwhile, memory accumulation-replay is proposed to accumulate the prior knowledge and integrate it into the model to enable the model to accommodate the multi-annotation setting. Two medical image benchmarks with different modalities are utilized to evaluate the performance of ACTION, and extensive experimental results demonstrate that it achieves superior performance compared to several state-of-the-art methods.
- 28 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 28 (12) , 7175-7183
Adaptive Annotation Correlation Based Multi-Annotation Learning for Calibrated Medical Image Segmentation
Medical image segmentation is a fundamental task in many clinical applications, yet current automated segmentation methods rely heavily on manual annotations, which are inherently subjective and prone to annotation bias. Recently, modeling annotator preference has garnered great interest, and several methods have been proposed in the past two years. However, the existing methods completely ignore the potential correlation between annotations, such as complementary and discriminative information. In this work, the A daptive annotation C orrela T ion based mult I -ann O tation Lear N ing ( ACTION ) method is proposed for calibrated medical image segmentation. ACTION employs consensus feature learning and dynamic adaptive weighting to leverage complementary information across annotations and emphasize discriminative information within each annotation based on their correlations, respectively. Meanwhile, memory accumulation-replay is proposed to accumulate the prior knowledge and integrate it into the model to enable the model to accommodate the multi-annotation setting. Two medical image benchmarks with different modalities are utilized to evaluate the performance of ACTION, and extensive experimental results demonstrate that it achieves superior performance compared to several state-of-the-art methods.
- 1 January 2024
- Published by Inderscience Publishers in International Journal of Reasoning-based Intelligent Systems
- Vol. 1 (1)
Multi-scale detail enhancement method of 2D animation image based on Retinex algorithm
- 1 January 2024
- Published by Inderscience Publishers in International Journal of Reasoning-based Intelligent Systems
- Vol. 1 (1)
Multi-scale detail enhancement method of 2D animation image based on Retinex algorithm
- 29 September 2023
- Published by Elsevier in Computer Methods and Programs in Biomedicine
- Vol. 242, 107826
Consistent representation via contrastive learning for skin lesion diagnosis
- 29 September 2023
- Published by Elsevier in Computer Methods and Programs in Biomedicine
- Vol. 242, 107826
Consistent representation via contrastive learning for skin lesion diagnosis
- 28 April 2023
- Published by Elsevier in Knowledge-Based Systems
- Vol. 273
Fine-grained recognition: Multi-granularity labels and category similarity matrix
- 28 April 2023
- Published by Elsevier in Knowledge-Based Systems
- Vol. 273
Fine-grained recognition: Multi-granularity labels and category similarity matrix
- 7 January 2023
- Published by Elsevier in Medical Image Analysis
- Vol. 85, 102746
Intra-class consistency and inter-class discrimination feature learning for automatic skin lesion classification
- 7 January 2023
- Published by Elsevier in Medical Image Analysis
- Vol. 85, 102746
Intra-class consistency and inter-class discrimination feature learning for automatic skin lesion classification
- 1 December 2022
- Published by Springer Nature in Applied Intelligence
- Vol. 53 (12) , 16015-16028
Multi-instance learning based on spatial continuous category representation for case-level meningioma grading in MRI images
- 1 December 2022
- Published by Springer Nature in Applied Intelligence
- Vol. 53 (12) , 16015-16028
Multi-instance learning based on spatial continuous category representation for case-level meningioma grading in MRI images
- 14 October 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cybernetics
- Vol. 53 (8) , 5323-5335
A Feature Space-Restricted Attention Attack on Medical Deep Learning Systems
Deep neural network has shown a powerful performance in the medical image analysis of a variety of diseases. However, a number of studies over the past few years have demonstrated that these deep learning systems can be vulnerable to well-designed adversarial attacks, with minor disruptions added to the input. Since both the public and academia have focused on deep learning in the health information economy, these adversarial attacks would prove more important and raise security concerns. In this article, adversarial attacks on deep learning systems in medicine are analyzed from two different points of view: 1) white box and 2) black box. A fast adversarial sample generation method, Feature Space-Restricted Attention Attack is proposed to explore more confusing adversarial samples. It is based on a generative adversarial network with bound classification space to generate perturbations to achieve attacks. Meanwhile, it can employ an attention mechanism to focus this perturbation on the lesion region. This enables the perturbation closely associated with the classification information making the attack more efficient and invisible. The performance and specificity of the proposed attack method are demonstrated by conducting extensive experiments on three different types of medical images. Finally, it is expected that this work can assist practitioners become being of current weaknesses in the deployment of deep learning systems in clinical settings. And, it further investigates domain-specific features of medical deep learning systems to enhance model generalization and resistance to attacks.
- 14 October 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cybernetics
- Vol. 53 (8) , 5323-5335
A Feature Space-Restricted Attention Attack on Medical Deep Learning Systems
Deep neural network has shown a powerful performance in the medical image analysis of a variety of diseases. However, a number of studies over the past few years have demonstrated that these deep learning systems can be vulnerable to well-designed adversarial attacks, with minor disruptions added to the input. Since both the public and academia have focused on deep learning in the health information economy, these adversarial attacks would prove more important and raise security concerns. In this article, adversarial attacks on deep learning systems in medicine are analyzed from two different points of view: 1) white box and 2) black box. A fast adversarial sample generation method, Feature Space-Restricted Attention Attack is proposed to explore more confusing adversarial samples. It is based on a generative adversarial network with bound classification space to generate perturbations to achieve attacks. Meanwhile, it can employ an attention mechanism to focus this perturbation on the lesion region. This enables the perturbation closely associated with the classification information making the attack more efficient and invisible. The performance and specificity of the proposed attack method are demonstrated by conducting extensive experiments on three different types of medical images. Finally, it is expected that this work can assist practitioners become being of current weaknesses in the deployment of deep learning systems in clinical settings. And, it further investigates domain-specific features of medical deep learning systems to enhance model generalization and resistance to attacks.
- 13 October 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive and Developmental Systems
- Vol. 15 (3) , 1419-1429
Deep Slice-Crossed Network With Local Weighted Loss for Brain Metastases Segmentation
Cognitive deficits occur in up to 90% of patients with brain metastases, can be caused by the tumor itself and whole brain radiation therapy. Latest treatment options to reduce cognitive decline require high-precision metastases segmentation. This asks existing metastases segmentation models to be further improved as their limited ability to micro lesion perception. To reduce the segmentation error on brain tissues and improve the accuracy on micro metastases, a deep slice-crossed network and a local weighted loss are proposed, respectively. The slice-crossed network contains a gated feature fusion module where the "gate" is learned individually from the slice-crossed label to roughly localize metastases. Then in tumor regions, this "gate" allows intra-slice information to pass for metastases enhancement, while in other regions, it allows inter-slice information to pass for vascular screening. The local weighted loss is proposed to address the inconsistency in size of intra-class objects. Unlike existing weighting approaches, it calculates the weight of each pixel from its neighbourhood pixel distribution to enhance the learning of pixels within small tumours, which can be applied to common losses. To evaluate them, two datasets containing 1000 and 368 patients, respectively, were collated. Experimental results demonstrate that the proposed method perform satisfactorily compared with state-of-the-art brain metastasis segmentation models.
- 13 October 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive and Developmental Systems
- Vol. 15 (3) , 1419-1429
Deep Slice-Crossed Network With Local Weighted Loss for Brain Metastases Segmentation
Cognitive deficits occur in up to 90% of patients with brain metastases, can be caused by the tumor itself and whole brain radiation therapy. Latest treatment options to reduce cognitive decline require high-precision metastases segmentation. This asks existing metastases segmentation models to be further improved as their limited ability to micro lesion perception. To reduce the segmentation error on brain tissues and improve the accuracy on micro metastases, a deep slice-crossed network and a local weighted loss are proposed, respectively. The slice-crossed network contains a gated feature fusion module where the "gate" is learned individually from the slice-crossed label to roughly localize metastases. Then in tumor regions, this "gate" allows intra-slice information to pass for metastases enhancement, while in other regions, it allows inter-slice information to pass for vascular screening. The local weighted loss is proposed to address the inconsistency in size of intra-class objects. Unlike existing weighting approaches, it calculates the weight of each pixel from its neighbourhood pixel distribution to enhance the learning of pixels within small tumours, which can be applied to common losses. To evaluate them, two datasets containing 1000 and 368 patients, respectively, were collated. Experimental results demonstrate that the proposed method perform satisfactorily compared with state-of-the-art brain metastasis segmentation models.
- 12 September 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Artificial Intelligence
- Vol. 4 (6) , 1504-1513
Boundary-Aware Network With Topological Consistency Constraint for Optic Chiasm Segmentation
Optic chiasm is a structure which is easily compressed by the tumors leading to different degrees of visual field defect and visual disturbance. When the optic chiasm is compressed, segmentation of the optic chiasm from MRI images is helpful for prognosis prediction and radiotherapy planning. However, due to the ambiguity of the optic chiasm boundary and the neglect of the anatomical structure consistency, the performance of the existing methods is limited. In this work, a Boundary-Aware Network (BANet) with topological consistency constraint is proposed for automated segmentation of optic chiasm. BANet constrained by topology loss leverages the complementary information between boundary feature and segmentation feature to effectively improve segmentation performance and topological consistency. To evaluate the effectiveness of the proposed method, a real-world specialized optic chiasm segmentation dataset is constructed. The experimental results demonstrate that the proposed method achieves higher segmentation accuracy compared with the state-of-the-art method.
- 12 September 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Artificial Intelligence
- Vol. 4 (6) , 1504-1513
Boundary-Aware Network With Topological Consistency Constraint for Optic Chiasm Segmentation
Optic chiasm is a structure which is easily compressed by the tumors leading to different degrees of visual field defect and visual disturbance. When the optic chiasm is compressed, segmentation of the optic chiasm from MRI images is helpful for prognosis prediction and radiotherapy planning. However, due to the ambiguity of the optic chiasm boundary and the neglect of the anatomical structure consistency, the performance of the existing methods is limited. In this work, a Boundary-Aware Network (BANet) with topological consistency constraint is proposed for automated segmentation of optic chiasm. BANet constrained by topology loss leverages the complementary information between boundary feature and segmentation feature to effectively improve segmentation performance and topological consistency. To evaluate the effectiveness of the proposed method, a real-world specialized optic chiasm segmentation dataset is constructed. The experimental results demonstrate that the proposed method achieves higher segmentation accuracy compared with the state-of-the-art method.
- 11 February 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 6 (5) , 1201-1210
Feature Pyramid Network With Level-Aware Attention for Meningioma Segmentation
Meningiomas are the most common primary intracranial tumors in adults, and they are dangerous and even lethal when they grow and oppress vital organs. In clinical, microsurgical resection is the most widely used treatment for most meningiomas. And tumor segmentation is an essential primary step before applying any therapy. However, due to the various meningioma locations and complicated intracranial structures, it is still challenging to segment the tumor accurately in both boundaries and contour automatically. In this work, a novel method is proposed for the automatic segmentation of meningioma. In general, the proposed method follows a coarse-to-fine strategy. Specifically, the feature pyramid structure is employed to extract multi-level features. Further, a Level-aware Attention is proposed to refine multi-level features by naturally utilizing the complementary features of different levels, thus significantly improving the segmentation performance. Moreover, to validate the proposed method, a realistic dataset of meningioma with fine labels is constructed, and experimental results on the dataset demonstrate the effectiveness of the proposed method.
- 11 February 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 6 (5) , 1201-1210
Feature Pyramid Network With Level-Aware Attention for Meningioma Segmentation
Meningiomas are the most common primary intracranial tumors in adults, and they are dangerous and even lethal when they grow and oppress vital organs. In clinical, microsurgical resection is the most widely used treatment for most meningiomas. And tumor segmentation is an essential primary step before applying any therapy. However, due to the various meningioma locations and complicated intracranial structures, it is still challenging to segment the tumor accurately in both boundaries and contour automatically. In this work, a novel method is proposed for the automatic segmentation of meningioma. In general, the proposed method follows a coarse-to-fine strategy. Specifically, the feature pyramid structure is employed to extract multi-level features. Further, a Level-aware Attention is proposed to refine multi-level features by naturally utilizing the complementary features of different levels, thus significantly improving the segmentation performance. Moreover, to validate the proposed method, a realistic dataset of meningioma with fine labels is constructed, and experimental results on the dataset demonstrate the effectiveness of the proposed method.