Ji Xiong
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Ji Xiong has a total of 94 co-authors. Here are the top co-authors by number of shared publications:
Feng Tang
Department of Pathology Huashan Hospital, Fudan University Shanghai PR China
4 shared publicationsYin Wang
Department of Pathology Huashan Hospital, Fudan University Shanghai PR China
4 shared publicationsJianpeng Liu
Department of Radiology, Huashan hospital, Fudan University, Shanghai, China
2 shared publicationsJiaqi Tu
Department of Radiology, Huashan hospital, Fudan University, Shanghai, China
2 shared publicationsKai Lun Xu
Xinhua Hospital, School of medicine, Shanghai Jiao Tong University, shanghai, China
2 shared publications- Shuyong Liu
Department of Nuclear Medicine, The Second Affiliated Hospital of Shandong First Medical University, Taian and
2 shared publications
Latest Publications
- Open Access
- 1 July 2026
- Published by Wiley in The Journal of Pathology: Clinical Research
- Vol. 12 (4) , e70103
SOX10 distinguishes oligodendrogliomas from low‐grade glioneuronal tumors/low‐grade gliomas with oligodendrocyte‐like features
Oligodendrogliomas exhibit significant morphological overlap with low‐grade glioneuronal tumors/low‐grade gliomas with oligodendrocyte‐like features, posing challenges in histopathological diagnosis. SOX10, a transcription factor critical for oligodendrocyte maturation and central nervous system myelination, may serve as a diagnostic marker. We evaluated the diagnostic utility of SOX10 immunohistochemical expression in 300 oligodendrogliomas versus 52 low‐grade glioneuronal tumors/low‐grade gliomas with oligodendrocyte‐like features. Notably, SOX10 was negative in 251/300 (83.7%) oligodendrogliomas but diffusely positive in 47/52 (90.4%) low‐grade glioneuronal tumors/low‐grade gliomas with oligodendrocyte‐like features (except polymorphous low‐grade neuroepithelial tumors of the young, PLNTY). SOX10 expression was significantly higher in the latter group ( p < 0.001) and showed a WHO grade dependent up‐regulation in oligodendrogliomas ( p < 0.001). The diagnostic utility of SOX10 was evident in IDH1‐immunonegative cases, yielding 72% sensitivity, 90.4% specificity, and a 78.3% positive predictive value for oligodendroglioma. Combining H3K27me3 loss with SOX10 negativity markedly improved diagnostic sensitivity for oligodendroglioma to 88.5%, substantially surpassing the 35% sensitivity achieved with H3K27me3 loss alone. Additionally, an immunohistochemical panel comprising SOX10, Olig2, and IDH1 aided in discriminating non‐neoplastic oligodendrocyte hyperplasia from oligodendroglioma and evaluating the presence of peritumoral tumor residue, particularly in IDH1‐immunonegative oligodendroglioma. In conclusion, SOX10 negativity effectively excludes low‐grade glioneuronal tumors/low‐grade gliomas with oligodendrocyte‐like features (except PLNTY) and non‐neoplastic oligodendrocyte hyperplasia from oligodendrogliomas. SOX10 serves as a specific diagnostic marker, enhancing pathological diagnostic accuracy.
- 1 July 2026
- Published by Wiley in The Journal of Pathology: Clinical Research
- Vol. 12 (4) , e70103
Open AccessSOX10 distinguishes oligodendrogliomas from low‐grade glioneuronal tumors/low‐grade gliomas with oligodendrocyte‐like features
Oligodendrogliomas exhibit significant morphological overlap with low‐grade glioneuronal tumors/low‐grade gliomas with oligodendrocyte‐like features, posing challenges in histopathological diagnosis. SOX10, a transcription factor critical for oligodendrocyte maturation and central nervous system myelination, may serve as a diagnostic marker. We evaluated the diagnostic utility of SOX10 immunohistochemical expression in 300 oligodendrogliomas versus 52 low‐grade glioneuronal tumors/low‐grade gliomas with oligodendrocyte‐like features. Notably, SOX10 was negative in 251/300 (83.7%) oligodendrogliomas but diffusely positive in 47/52 (90.4%) low‐grade glioneuronal tumors/low‐grade gliomas with oligodendrocyte‐like features (except polymorphous low‐grade neuroepithelial tumors of the young, PLNTY). SOX10 expression was significantly higher in the latter group ( p < 0.001) and showed a WHO grade dependent up‐regulation in oligodendrogliomas ( p < 0.001). The diagnostic utility of SOX10 was evident in IDH1‐immunonegative cases, yielding 72% sensitivity, 90.4% specificity, and a 78.3% positive predictive value for oligodendroglioma. Combining H3K27me3 loss with SOX10 negativity markedly improved diagnostic sensitivity for oligodendroglioma to 88.5%, substantially surpassing the 35% sensitivity achieved with H3K27me3 loss alone. Additionally, an immunohistochemical panel comprising SOX10, Olig2, and IDH1 aided in discriminating non‐neoplastic oligodendrocyte hyperplasia from oligodendroglioma and evaluating the presence of peritumoral tumor residue, particularly in IDH1‐immunonegative oligodendroglioma. In conclusion, SOX10 negativity effectively excludes low‐grade glioneuronal tumors/low‐grade gliomas with oligodendrocyte‐like features (except PLNTY) and non‐neoplastic oligodendrocyte hyperplasia from oligodendrogliomas. SOX10 serves as a specific diagnostic marker, enhancing pathological diagnostic accuracy.
- 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.
- Open Access
- 23 March 2026
- Published by Springer Nature in Diagnostic Pathology
- Vol. 21
Clinicopathological analysis of 19 cases of mixed gangliocytoma-pituitary neuroendocrine tumors
Mixed gangliocytoma-pituitary neuroendocrine tumors (PitNETs) occurring in the sellar region are extremely rare neoplasms. To improve the understanding of
- 23 March 2026
- Published by Springer Nature in Diagnostic Pathology
- Vol. 21
Open AccessClinicopathological analysis of 19 cases of mixed gangliocytoma-pituitary neuroendocrine tumors
Mixed gangliocytoma-pituitary neuroendocrine tumors (PitNETs) occurring in the sellar region are extremely rare neoplasms. To improve the understanding of
- 19 February 2026
- Published by Elsevier in Academic Radiology
- Vol. 33 (6) , 2657-2670
Diagnostic Performance of Microstructural Parameters Derived From Time-dependent Diffusion MRI for Grading Meningiomas: A Prospective Study
- 19 February 2026
- Published by Elsevier in Academic Radiology
- Vol. 33 (6) , 2657-2670
Diagnostic Performance of Microstructural Parameters Derived From Time-dependent Diffusion MRI for Grading Meningiomas: A Prospective Study
- Open Access
- 11 February 2026
- Published by Springer Nature in Research Square
Foundation Model-Architected Sparse Dictionaries for Fully Quantitative Interpretability in Whole Slide Image Decoding
Artificial intelligence (AI)-based pathological image analysis has achieved remarkable progress. However, a critical limitation persists: beyond attention-based visualization methods—which offer only partial interpretability—existing AI approaches fail to provide quantifiable diagnostic evidence that meets clinically actionable standards and aligns with pathologists' domain expertise. Consequently, the interpretability gap between AI diagnostics and pathologists impedes clinical adoption. To resolve this impasse, we propose PATH-SparseD (Pathology-Aware Transformer–Hybridized Sparse Dictionary Learning), a framework that reformulates AI-driven diagnostic inference as a dictionary-query process. Central to PATH-SparseD is the novel integration of foundation models with sparse representation theory. This synergy empowers the model to deconstruct the highly complex and heterogeneous data of a WSI into a sparse combination of fundamental, clinically meaningful visual "primitives"—each acting as an atomic diagnostic unit. Specifically, for any WSI tile, PATH-SparseD does not merely extract features; it encodes the tile by identifying and activating a minimal set of these primitives from a multi-scale dictionary. This process effectively reformulates the entire WSI as a quantifiable histogram of primitive occurrences, creating a transparent and pattern-based transcript of the tissue that directly mirrors a pathologist's diagnostic reasoning. Extensive experiments on 10 datasets covering 8 tumor types demonstrated that PATH-SparseD not only provides an interpretable and quantitatively verifiable pathological analysis framework recognized by clinical pathologists, but also significantly outperforms state-of-the-art foundation-model approaches, yielding accuracy improvements of 5.6% in glioma grading, 4.4% in IDH1 genotyping, 14.5% in TNM grading, 17.0% in tumor differentiation, 11.8% in cellular origin, and 8.6% in organ origin. Ultimately, PATH-SparseD establishes a novel paradigm for WSI decoding that harnesses the performance advantages of foundation models while providing an effective solution to the interpretability challenge.
- 11 February 2026
- Published by Springer Nature in Research Square
Open AccessFoundation Model-Architected Sparse Dictionaries for Fully Quantitative Interpretability in Whole Slide Image Decoding
Artificial intelligence (AI)-based pathological image analysis has achieved remarkable progress. However, a critical limitation persists: beyond attention-based visualization methods—which offer only partial interpretability—existing AI approaches fail to provide quantifiable diagnostic evidence that meets clinically actionable standards and aligns with pathologists' domain expertise. Consequently, the interpretability gap between AI diagnostics and pathologists impedes clinical adoption. To resolve this impasse, we propose PATH-SparseD (Pathology-Aware Transformer–Hybridized Sparse Dictionary Learning), a framework that reformulates AI-driven diagnostic inference as a dictionary-query process. Central to PATH-SparseD is the novel integration of foundation models with sparse representation theory. This synergy empowers the model to deconstruct the highly complex and heterogeneous data of a WSI into a sparse combination of fundamental, clinically meaningful visual "primitives"—each acting as an atomic diagnostic unit. Specifically, for any WSI tile, PATH-SparseD does not merely extract features; it encodes the tile by identifying and activating a minimal set of these primitives from a multi-scale dictionary. This process effectively reformulates the entire WSI as a quantifiable histogram of primitive occurrences, creating a transparent and pattern-based transcript of the tissue that directly mirrors a pathologist's diagnostic reasoning. Extensive experiments on 10 datasets covering 8 tumor types demonstrated that PATH-SparseD not only provides an interpretable and quantitatively verifiable pathological analysis framework recognized by clinical pathologists, but also significantly outperforms state-of-the-art foundation-model approaches, yielding accuracy improvements of 5.6% in glioma grading, 4.4% in IDH1 genotyping, 14.5% in TNM grading, 17.0% in tumor differentiation, 11.8% in cellular origin, and 8.6% in organ origin. Ultimately, PATH-SparseD establishes a novel paradigm for WSI decoding that harnesses the performance advantages of foundation models while providing an effective solution to the interpretability challenge.
- Open Access
- 11 July 2025
- Published by Springer Nature in Journal of Big Data
- Vol. 12 (1) , 1-15
Deep learning-based tumor segmentation and radiogenomic model for predicting EGFR amplification and assessing intratumoural heterogeneity in glioblastoma
- 11 July 2025
- Published by Springer Nature in Journal of Big Data
- Vol. 12 (1) , 1-15
Open AccessDeep learning-based tumor segmentation and radiogenomic model for predicting EGFR amplification and assessing intratumoural heterogeneity in glioblastoma
- 8 November 2024
- Published by Elsevier in Clinical Radiology
- Vol. 80, 106746
MRI-based radiomics virtual biopsy for BCL6 in primary central nervous system lymphoma
- 8 November 2024
- Published by Elsevier in Clinical Radiology
- Vol. 80, 106746
MRI-based radiomics virtual biopsy for BCL6 in primary central nervous system lymphoma
- 6 July 2024
- Published by Elsevier in European Journal of Radiology
- Vol. 178, 111603
MRI-based radiomics signatures for preoperative prediction of Ki-67 index in primary central nervous system lymphoma
- 6 July 2024
- Published by Elsevier in European Journal of Radiology
- Vol. 178, 111603
MRI-based radiomics signatures for preoperative prediction of Ki-67 index in primary central nervous system lymphoma
- 8 February 2024
- Published by Springer Nature in Acta Neurochirurgica
- Vol. 166 (1) , 1-11
Development and validation of a 23-gene expression signature for molecular subtyping of medulloblastoma in a long-term Chinese cohort
Purpose Medulloblastoma is the most common childhood malignant brain tumor and is a leading cause of cancer-related death in children. Recent transcriptional studies have shown that medulloblastomas comprise at least four molecular subgroups, each with distinct demographics, genetics, and clinical outcomes. Medulloblastoma subtyping has become critical for subgroup-specific therapies. The use of gene expression assays to determine the molecular subgroup of clinical specimens is a long-awaited application of molecular biology for this pediatric cancer. Methods In the current study, we established a medulloblastoma transcriptome database of 460 samples retrieved from three published datasets (GSE21140, GSE37382, and GSE37418). With this database, we identified a 23-gene signature that is significantly associated with the medulloblastoma subgroups and achieved a classification accuracy of 95.2%. Results The 23-gene signature was further validated in a long-term cohort of 142 Chinese medulloblastoma patients. The 23-gene signature classified 21 patients as WNT (15%), 41 as SHH (29%), 16 as Group 3 (11%), and 64 as Group 4 (45%). For patients of WNT, SHH, Group 3, and Group 4, 5-year overall-survival rate reached 80%, 62%, 27%, and 47%, respectively (p < 0.0001), meanwhile 5-year progression-free survival reached 80%, 52%, 27%, and 45%, respectively (p < 0.0001). Besides, SHH/TP53-mutant tumors were associated with worse prognosis compared with SHH/TP53 wild-type tumors and other subgroups. We demonstrated that subgroup assignments by the 23-gene signature and Northcott’s NanoString assay were highly comparable with a concordance rate of 96.4%. Conclusions In conclusion, we present a novel gene signature that is capable of accurately and reliably assigning FFPE medulloblastoma samples to their molecular subgroup, which may serve as an auxiliary tool for medulloblastoma subtyping in the clinic. Future incorporation of this gene signature into prospective clinical trials is warranted to further evaluate its clinical.
- 8 February 2024
- Published by Springer Nature in Acta Neurochirurgica
- Vol. 166 (1) , 1-11
Development and validation of a 23-gene expression signature for molecular subtyping of medulloblastoma in a long-term Chinese cohort
Purpose Medulloblastoma is the most common childhood malignant brain tumor and is a leading cause of cancer-related death in children. Recent transcriptional studies have shown that medulloblastomas comprise at least four molecular subgroups, each with distinct demographics, genetics, and clinical outcomes. Medulloblastoma subtyping has become critical for subgroup-specific therapies. The use of gene expression assays to determine the molecular subgroup of clinical specimens is a long-awaited application of molecular biology for this pediatric cancer. Methods In the current study, we established a medulloblastoma transcriptome database of 460 samples retrieved from three published datasets (GSE21140, GSE37382, and GSE37418). With this database, we identified a 23-gene signature that is significantly associated with the medulloblastoma subgroups and achieved a classification accuracy of 95.2%. Results The 23-gene signature was further validated in a long-term cohort of 142 Chinese medulloblastoma patients. The 23-gene signature classified 21 patients as WNT (15%), 41 as SHH (29%), 16 as Group 3 (11%), and 64 as Group 4 (45%). For patients of WNT, SHH, Group 3, and Group 4, 5-year overall-survival rate reached 80%, 62%, 27%, and 47%, respectively (p < 0.0001), meanwhile 5-year progression-free survival reached 80%, 52%, 27%, and 45%, respectively (p < 0.0001). Besides, SHH/TP53-mutant tumors were associated with worse prognosis compared with SHH/TP53 wild-type tumors and other subgroups. We demonstrated that subgroup assignments by the 23-gene signature and Northcott’s NanoString assay were highly comparable with a concordance rate of 96.4%. Conclusions In conclusion, we present a novel gene signature that is capable of accurately and reliably assigning FFPE medulloblastoma samples to their molecular subgroup, which may serve as an auxiliary tool for medulloblastoma subtyping in the clinic. Future incorporation of this gene signature into prospective clinical trials is warranted to further evaluate its clinical.
- Open Access
- 12 January 2023
- Published by Frontiers Media SA in Frontiers in Oncology
- Vol. 12, 1061604
Corrigendum: H3K27M mutation doesn’t mean worse prognosis in old patients
Corrigendum: H3K27M mutation doesn’t mean worse prognosis in old patients
- 12 January 2023
- Published by Frontiers Media SA in Frontiers in Oncology
- Vol. 12, 1061604
Open AccessCorrigendum: H3K27M mutation doesn’t mean worse prognosis in old patients
Corrigendum: H3K27M mutation doesn’t mean worse prognosis in old patients