Ao Li
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
Research Subjects
Ao Li has a total of 152 co-authors. Here are the top co-authors by number of shared publications:
- Minghui Wang
The School of Information Science and Technology, University of Science and Technology of China, Hefei, China
47 shared publications Minghui Wang
School of Information Science and Technology, University of Science and Technology of China, Hefei AH230027, China
18 shared publicationsHuanqing Feng
Department of Electronic Science and Technology, School of Information Science and Technology, University of Science and Technology of China, Hefei, Anhui 230027, P.R. China
13 shared publications- Jianing Xi
School of Biomedical Engineering, Guangzhou Medical University, Guangzhou, China
9 shared publications - Honglei Liu
School of Information Science and Technology, University of Science and Technology of China, Hefei, China
6 shared publications Yi Shen
School of Information Science and Technology
6 shared publications
Latest Publications
- Open Access
- 24 April 2026
- Published by Springer Nature in npj Precision Oncology
- Vol. 10
Deep learning for predicting pituitary neuroendocrine tumour lineage and high-risk subtypes from histology
Pituitary neuroendocrine tumours (PitNETs) exhibit significant heterogeneity, posing challenges for clinical management. We developed a deep learning model to predict PitNET lineage, high-risk subtypes, and recurrence directly from routine H&E-stained whole-slide images. Trained on 925 patients from USTC and externally validated on cohorts from Taihe Hospital (n = 226) and Huashan Hospital (n = 193), the model achieved a micro-average AUC of 0.912 for lineage classification (SF1: 0.926, PIT1: 0.932, TPIT: 0.904; Without distinct lineage: 0.706). High-risk subtype prediction yielded AUCs of 0.805 (PIT1), 0.753 (TPIT), and 0.733 (null cell). Recurrence prediction reached an AUC of 0.641. Analysis of the tumour microenvironment revealed that compared with primary tumours, recurrence tumours were characterized by an increased density of M2 macrophages and decreased infiltration of CD8 + T cells. Spatial transcriptomics further elucidated distinct molecular pathways associated with recurrence, providing mechanistic insights into prognostic predictions. Our deep learning model accurately predicts PitNET characteristics from routine H&E slides, and spatial biology validation identified distinct immune and molecular features associated with recurrence.
- 24 April 2026
- Published by Springer Nature in npj Precision Oncology
- Vol. 10
Open AccessDeep learning for predicting pituitary neuroendocrine tumour lineage and high-risk subtypes from histology
Pituitary neuroendocrine tumours (PitNETs) exhibit significant heterogeneity, posing challenges for clinical management. We developed a deep learning model to predict PitNET lineage, high-risk subtypes, and recurrence directly from routine H&E-stained whole-slide images. Trained on 925 patients from USTC and externally validated on cohorts from Taihe Hospital (n = 226) and Huashan Hospital (n = 193), the model achieved a micro-average AUC of 0.912 for lineage classification (SF1: 0.926, PIT1: 0.932, TPIT: 0.904; Without distinct lineage: 0.706). High-risk subtype prediction yielded AUCs of 0.805 (PIT1), 0.753 (TPIT), and 0.733 (null cell). Recurrence prediction reached an AUC of 0.641. Analysis of the tumour microenvironment revealed that compared with primary tumours, recurrence tumours were characterized by an increased density of M2 macrophages and decreased infiltration of CD8 + T cells. Spatial transcriptomics further elucidated distinct molecular pathways associated with recurrence, providing mechanistic insights into prognostic predictions. Our deep learning model accurately predicts PitNET characteristics from routine H&E slides, and spatial biology validation identified distinct immune and molecular features associated with recurrence.
- 18 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Imaging
- Vol. 45 (5) , 2065-2076
VGFA: Variation-Robust Graph-Level Feature Alignment for Domain Adaptive Nuclei Detection
Accurate detection of nuclei is a crucial step in advancing pathology image analysis for disease diagnosis and treatment. However, significant domain discrepancies exist among pathology images, which severely degrade the performance of detection models. Despite promising results from existing domain adaptation approaches, they may overlook the detrimental impact of intra-domain variation (IDV) at both cell and image scales. The IDV issue typically manifests as dramatic differences in nuclei composition, morphology, and spatial arrangement, which occurs not only across different pathology images but also within a single image. This inherent heterogeneity produces highly complex feature distributions, ultimately making cross-domain alignment significantly more arduous. Moreover, the presence of IDV further injects noise into pseudo-labels, reducing the signal-to-noise ratio in the feature space and complicating the alignment process. To tackle these challenges, we propose a novel variation-robust graph-level feature alignment (VGFA) framework for unsupervised domain adaptive nuclei detection. Specifically, our method first incorporates a prior-based nuclei graph pruning scheme that harnesses nuclei spatial contextual priors and dynamically eliminates unreliable nodes from the nuclei graph. Then, a local-global nuclei encoding network is designed to learn nuclei graph representations that holistically encapsulate the consistent traits among various nuclei, thereby mitigating challenges posed by cell-scale IDV. Moreover, VGFA leverages a nuclei graph discrepancy loss that is resilient to image-scale IDV, achieving effective feature alignment in cross-domain graph feature space. Extensive experiments across different adaptation scenarios demonstrate that our VGFA framework achieves state-of-the-art performance, outperforming existing feature alignment methods in domain adaptive nuclei detection.
- 18 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Imaging
- Vol. 45 (5) , 2065-2076
VGFA: Variation-Robust Graph-Level Feature Alignment for Domain Adaptive Nuclei Detection
Accurate detection of nuclei is a crucial step in advancing pathology image analysis for disease diagnosis and treatment. However, significant domain discrepancies exist among pathology images, which severely degrade the performance of detection models. Despite promising results from existing domain adaptation approaches, they may overlook the detrimental impact of intra-domain variation (IDV) at both cell and image scales. The IDV issue typically manifests as dramatic differences in nuclei composition, morphology, and spatial arrangement, which occurs not only across different pathology images but also within a single image. This inherent heterogeneity produces highly complex feature distributions, ultimately making cross-domain alignment significantly more arduous. Moreover, the presence of IDV further injects noise into pseudo-labels, reducing the signal-to-noise ratio in the feature space and complicating the alignment process. To tackle these challenges, we propose a novel variation-robust graph-level feature alignment (VGFA) framework for unsupervised domain adaptive nuclei detection. Specifically, our method first incorporates a prior-based nuclei graph pruning scheme that harnesses nuclei spatial contextual priors and dynamically eliminates unreliable nodes from the nuclei graph. Then, a local-global nuclei encoding network is designed to learn nuclei graph representations that holistically encapsulate the consistent traits among various nuclei, thereby mitigating challenges posed by cell-scale IDV. Moreover, VGFA leverages a nuclei graph discrepancy loss that is resilient to image-scale IDV, achieving effective feature alignment in cross-domain graph feature space. Extensive experiments across different adaptation scenarios demonstrate that our VGFA framework achieves state-of-the-art performance, outperforming existing feature alignment methods in domain adaptive nuclei detection.
- Open Access
- 27 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 1161-1170
Domain-Specific Interactive Prompting for Generalized Nuclei Classification
- 27 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 1161-1170
Open AccessDomain-Specific Interactive Prompting for Generalized Nuclei Classification
- 30 September 2025
- Published by Elsevier in Expert Systems with Applications
- Vol. 298
Symmetric difference consistent semi-supervised brain stroke lesion segmentation with uncertainty guidance
- 30 September 2025
- Published by Elsevier in Expert Systems with Applications
- Vol. 298
Symmetric difference consistent semi-supervised brain stroke lesion segmentation with uncertainty guidance
- Open Access
- 26 September 2025
- Published by Springer Nature in Diagnostic Pathology
- Vol. 20 (1) , 1-12
Artificial Intelligence-driven image analysis for standardised programmed death-ligand 1 expression evaluation in non-small cell lung cancer
Accurate assessment of programmed death-ligand 1 (PD-L1) immunohistochemical (IHC) expression is critical for immunotherapy in patients with non-small cell lung cancer (NSCLC). Yet, interpreting its staining is challenging, time-consuming, and causes inter-observer variability, potentially mis-stratifying patients. This necessitates the development of an artificial intelligence (AI) model to effectively quantify PD-L1 expression. Hence, we developed an AI-based deep-learning approach to automatically assess PD-L1 expression in NSCLC using IHC 22C3 assay-stained whole slide images (WSIs). A total of 706 patients with NSCLC were included in this study and 1212 WSIs were collected from three distinct study cohorts. We accurately matched the hematoxylin and eosin-stained images of the internal dataset with the IHC WSIs. Foreground regions containing tumor tissue were extracted from WSIs, and a multi-granular multiple-instance learning approach employing instance embeddings with coarse and fine granularities was implemented to extract patch-level morphological features. A multi-grained expression interpreter-based model aggregated these features to stratify PD-L1 expression status. The model showed strong interpretive ability in all three cohorts and wide applicability to different specimen types. The macro-average area under the receiver operating characteristic curve (AUC) were 0.940/0.915/0.944 for surgical specimens, 0.955/0.844/0.865 for biopsy specimens, and 0.901/0.958/0.883 for metastases. This study emphasizes the potential benefits of deep learning in automatically, rapidly, and accurately inferring PD-L1 expression from complex IHC images. It also showcases how AI frameworks can improve routine digital pathology workflows in current PD-L1 detection methods.
- 26 September 2025
- Published by Springer Nature in Diagnostic Pathology
- Vol. 20 (1) , 1-12
Open AccessArtificial Intelligence-driven image analysis for standardised programmed death-ligand 1 expression evaluation in non-small cell lung cancer
Accurate assessment of programmed death-ligand 1 (PD-L1) immunohistochemical (IHC) expression is critical for immunotherapy in patients with non-small cell lung cancer (NSCLC). Yet, interpreting its staining is challenging, time-consuming, and causes inter-observer variability, potentially mis-stratifying patients. This necessitates the development of an artificial intelligence (AI) model to effectively quantify PD-L1 expression. Hence, we developed an AI-based deep-learning approach to automatically assess PD-L1 expression in NSCLC using IHC 22C3 assay-stained whole slide images (WSIs). A total of 706 patients with NSCLC were included in this study and 1212 WSIs were collected from three distinct study cohorts. We accurately matched the hematoxylin and eosin-stained images of the internal dataset with the IHC WSIs. Foreground regions containing tumor tissue were extracted from WSIs, and a multi-granular multiple-instance learning approach employing instance embeddings with coarse and fine granularities was implemented to extract patch-level morphological features. A multi-grained expression interpreter-based model aggregated these features to stratify PD-L1 expression status. The model showed strong interpretive ability in all three cohorts and wide applicability to different specimen types. The macro-average area under the receiver operating characteristic curve (AUC) were 0.940/0.915/0.944 for surgical specimens, 0.955/0.844/0.865 for biopsy specimens, and 0.901/0.958/0.883 for metastases. This study emphasizes the potential benefits of deep learning in automatically, rapidly, and accurately inferring PD-L1 expression from complex IHC images. It also showcases how AI frameworks can improve routine digital pathology workflows in current PD-L1 detection methods.
- 12 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Imaging
- Vol. 44 (11) , 4524-4535
Tackling Tumor Heterogeneity Issue: Transformer-Based Multiple Instance Enhancement Learning for Predicting EGFR Mutation via CT Images
Accurate and non-invasive prediction of epidermal growth factor receptor (EGFR) mutation is crucial for the diagnosis and treatment of non-small cell lung cancer (NSCLC). While computed tomography (CT) imaging shows promise in identifying EGFR mutation, current prediction methods heavily rely on fully supervised learning, which overlooks the substantial heterogeneity of tumors and therefore leads to suboptimal results. To tackle tumor heterogeneity issue, this study introduces a novel weakly supervised method named TransMIEL, which leverages multiple instance learning techniques for accurate EGFR mutation prediction. Specifically, we first propose an innovative instance enhancement learning (IEL) strategy that strengthens the discriminative power of instance features for complex tumor CT images by exploring self-derived soft pseudo-labels. Next, to improve tumor representation capability, we design a spatial-aware transformer (SAT) that fully captures inter-instance relationships of different pathological subregions to mirror the diagnostic processes of radiologists. Finally, an instance adaptive gating (IAG) module is developed to effectively emphasize the contribution of informative instance features in heterogeneous tumors, facilitating dynamic instance feature aggregation and increasing model generalization performance. Experimental results demonstrate that TransMIEL significantly outperforms existing fully and weakly supervised methods on both public and in-house NSCLC datasets. Additionally, visualization results show that our approach can highlight intra-tumor and peri-tumor areas relevant to EGFR mutation status. Therefore, our method holds significant potential as an effective tool for EGFR prediction and offers a novel perspective for future research on tumor heterogeneity.
- 12 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Imaging
- Vol. 44 (11) , 4524-4535
Tackling Tumor Heterogeneity Issue: Transformer-Based Multiple Instance Enhancement Learning for Predicting EGFR Mutation via CT Images
Accurate and non-invasive prediction of epidermal growth factor receptor (EGFR) mutation is crucial for the diagnosis and treatment of non-small cell lung cancer (NSCLC). While computed tomography (CT) imaging shows promise in identifying EGFR mutation, current prediction methods heavily rely on fully supervised learning, which overlooks the substantial heterogeneity of tumors and therefore leads to suboptimal results. To tackle tumor heterogeneity issue, this study introduces a novel weakly supervised method named TransMIEL, which leverages multiple instance learning techniques for accurate EGFR mutation prediction. Specifically, we first propose an innovative instance enhancement learning (IEL) strategy that strengthens the discriminative power of instance features for complex tumor CT images by exploring self-derived soft pseudo-labels. Next, to improve tumor representation capability, we design a spatial-aware transformer (SAT) that fully captures inter-instance relationships of different pathological subregions to mirror the diagnostic processes of radiologists. Finally, an instance adaptive gating (IAG) module is developed to effectively emphasize the contribution of informative instance features in heterogeneous tumors, facilitating dynamic instance feature aggregation and increasing model generalization performance. Experimental results demonstrate that TransMIEL significantly outperforms existing fully and weakly supervised methods on both public and in-house NSCLC datasets. Additionally, visualization results show that our approach can highlight intra-tumor and peri-tumor areas relevant to EGFR mutation status. Therefore, our method holds significant potential as an effective tool for EGFR prediction and offers a novel perspective for future research on tumor heterogeneity.
- 15 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 29 (9) , 6311-6322
Agnostic-Specific Modality Learning for Cancer Survival Prediction From Multiple Data
Cancer is a pressing public health problem and one of the main causes of mortality worldwide. The development of advanced computational methods for predicting cancer survival is pivotal in aiding clinicians to formulate effective treatment strategies and improve patient quality of life. Recent advances in survival prediction methods show that integrating diverse information from various cancer-related data, such as pathological images and genomics, is crucial for improving prediction accuracy. Despite promising results of existing approaches, there are great challenges of modality gap and semantic redundancy presented in multiple cancer data, which could hinder the comprehensive integration and pose substantial obstacles to further enhancing cancer survival prediction. In this study, we propose a novel agnostic-specific modality learning (ASML) framework for accurate cancer survival prediction. To bridge the modality gap and provide a comprehensive view of distinct data modalities, we employ an agnostic-specific learning strategy to learn the commonality across modalities and the uniqueness of each modality. Moreover, a cross-modal fusion network is exerted to integrate multimodal information by modeling modality correlations and diminish semantic redundancy in a divide-and-conquer manner. Extensive experiment results on three TCGA datasets demonstrate that ASML reaches better performance than other existing cancer survival prediction methods for multiple data.
- 15 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 29 (9) , 6311-6322
Agnostic-Specific Modality Learning for Cancer Survival Prediction From Multiple Data
Cancer is a pressing public health problem and one of the main causes of mortality worldwide. The development of advanced computational methods for predicting cancer survival is pivotal in aiding clinicians to formulate effective treatment strategies and improve patient quality of life. Recent advances in survival prediction methods show that integrating diverse information from various cancer-related data, such as pathological images and genomics, is crucial for improving prediction accuracy. Despite promising results of existing approaches, there are great challenges of modality gap and semantic redundancy presented in multiple cancer data, which could hinder the comprehensive integration and pose substantial obstacles to further enhancing cancer survival prediction. In this study, we propose a novel agnostic-specific modality learning (ASML) framework for accurate cancer survival prediction. To bridge the modality gap and provide a comprehensive view of distinct data modalities, we employ an agnostic-specific learning strategy to learn the commonality across modalities and the uniqueness of each modality. Moreover, a cross-modal fusion network is exerted to integrate multimodal information by modeling modality correlations and diminish semantic redundancy in a divide-and-conquer manner. Extensive experiment results on three TCGA datasets demonstrate that ASML reaches better performance than other existing cancer survival prediction methods for multiple data.
- 18 September 2024
- Published by Elsevier in Expert Systems with Applications
- Vol. 260
Seeking multi-view commonality and peculiarity: A novel decoupling method for lung cancer subtype classification
- 18 September 2024
- Published by Elsevier in Expert Systems with Applications
- Vol. 260
Seeking multi-view commonality and peculiarity: A novel decoupling method for lung cancer subtype classification
- 7 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 29 (5) , 3247-3258
MIF: Multi-Shot Interactive Fusion Model for Cancer Survival Prediction Using Pathological Image and Genomic Data
Accurate cancer survival prediction is crucial for oncologists to determine therapeutic plan, which directly influences the treatment efficacy and survival outcome of patient. Recently, multimodal fusion-based prognostic methods have demonstrated effectiveness for survival prediction by fusing diverse cancer-related data from different medical modalities, e.g., pathological images and genomic data. However, these works still face significant challenges. First, most approaches attempt multimodal fusion by simple one-shot fusion strategy, which is insufficient to explore complex interactions underlying in highly disparate multimodal data. Second, current methods for investigating multimodal interactions face the capability-efficiency dilemma, which is the difficult balance between powerful modeling capability and applicable computational efficiency, thus impeding effective multimodal fusion. In this study, to encounter these challenges, we propose an innovative multi-shot interactive fusion method named MIF for precise survival prediction by utilizing pathological and genomic data. Particularly, a novel multi-shot fusion framework is introduced to promote multimodal fusion by decomposing it into successive fusing stages, thus delicately integrating modalities in a progressive way. Moreover, to address the capacity-efficiency dilemma, various affinity-based interactive modules are introduced to synergize the multi-shot framework. Specifically, by harnessing comprehensive affinity information as guidance for mining interactions, the proposed interactive modules can efficiently generate low-dimensional discriminative multimodal representations. Extensive experiments on different cancer datasets unravel that our method not only successfully achieves state-of-the-art performance by performing effective multimodal fusion, but also possesses high computational efficiency compared to existing survival prediction methods.
- 7 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 29 (5) , 3247-3258
MIF: Multi-Shot Interactive Fusion Model for Cancer Survival Prediction Using Pathological Image and Genomic Data
Accurate cancer survival prediction is crucial for oncologists to determine therapeutic plan, which directly influences the treatment efficacy and survival outcome of patient. Recently, multimodal fusion-based prognostic methods have demonstrated effectiveness for survival prediction by fusing diverse cancer-related data from different medical modalities, e.g., pathological images and genomic data. However, these works still face significant challenges. First, most approaches attempt multimodal fusion by simple one-shot fusion strategy, which is insufficient to explore complex interactions underlying in highly disparate multimodal data. Second, current methods for investigating multimodal interactions face the capability-efficiency dilemma, which is the difficult balance between powerful modeling capability and applicable computational efficiency, thus impeding effective multimodal fusion. In this study, to encounter these challenges, we propose an innovative multi-shot interactive fusion method named MIF for precise survival prediction by utilizing pathological and genomic data. Particularly, a novel multi-shot fusion framework is introduced to promote multimodal fusion by decomposing it into successive fusing stages, thus delicately integrating modalities in a progressive way. Moreover, to address the capacity-efficiency dilemma, various affinity-based interactive modules are introduced to synergize the multi-shot framework. Specifically, by harnessing comprehensive affinity information as guidance for mining interactions, the proposed interactive modules can efficiently generate low-dimensional discriminative multimodal representations. Extensive experiments on different cancer datasets unravel that our method not only successfully achieves state-of-the-art performance by performing effective multimodal fusion, but also possesses high computational efficiency compared to existing survival prediction methods.
- Open Access
- 27 December 2023
- Published by Association for Computing Machinery (ACM)
- p. 106-110
Multi-view Representation Learning for Histologic Subtype Classification of Lung Cancer
It is crucial to accurately discerning adenocarcinoma (ADC) from squamous cell carcinoma (SCC) histological subtypes utilizing computed tomography (CT) scans for guiding treatment decisions among individuals with non-small cell lung cancer (NSCLC). While the existing multi-view deep learning methods have shown promise in this domain, they usually struggle to mine common information efficiently due to the presence of intra-view interference, resulting in limited performance. In this research, we introduce an efficient multi-view representation learning technique to eliminate interference, thus obtaining higher-quality multi-view representations. Our approach involves the implementation of attention-based representation decoupling to project axial, coronal and sagittal views into both view-invariant and view-specific subspaces, respectively. Moreover, we employ a novel adversarial loss to ensure that the learned peculiar representations are not tailored to histological subtype classification, thereby effectively eliminating intra-view interference. Extensive experiments confirm our method's effectiveness in addressing the challenges that existing multi-view methods cannot resolve, consistently outperforming state-of-the-art techniques.
- 27 December 2023
- Published by Association for Computing Machinery (ACM)
- p. 106-110
Open AccessMulti-view Representation Learning for Histologic Subtype Classification of Lung Cancer
It is crucial to accurately discerning adenocarcinoma (ADC) from squamous cell carcinoma (SCC) histological subtypes utilizing computed tomography (CT) scans for guiding treatment decisions among individuals with non-small cell lung cancer (NSCLC). While the existing multi-view deep learning methods have shown promise in this domain, they usually struggle to mine common information efficiently due to the presence of intra-view interference, resulting in limited performance. In this research, we introduce an efficient multi-view representation learning technique to eliminate interference, thus obtaining higher-quality multi-view representations. Our approach involves the implementation of attention-based representation decoupling to project axial, coronal and sagittal views into both view-invariant and view-specific subspaces, respectively. Moreover, we employ a novel adversarial loss to ensure that the learned peculiar representations are not tailored to histological subtype classification, thereby effectively eliminating intra-view interference. Extensive experiments confirm our method's effectiveness in addressing the challenges that existing multi-view methods cannot resolve, consistently outperforming state-of-the-art techniques.