Safety and Efficacy of Drug Coated Balloon Therapy for de Novo Lesions in Patients With Coronary Heart Disease Under the Guidance of QFR (UNIQUE-DCB-I Study )

Since Gruntzig successfully performed percutaneous coronary balloon angioplasty in 1977, percutaneous coronary intervention has developed rapidly. From bare metal stents to drug-eluting stents (DES), the symptoms and prognosis of patients with coronary heart disease (CHD) have been greatly improved. Although DES has reduced the probability of in-stent restenosis (ISR) and thrombosis compared with BMS since its clinical application, it can not completely solve this problem. Even if the new generation of DES requires revascularization, the incidence of ISR is still as high as 5%-10%. DES treatment is associated with delayed endothelial healing, late acquired poor stent adherence and new atherosclerosis, which lead to late ISR and thrombosis. In addition, DES is still not ideal for the treatment of small vessel disease, diffuse long lesion and bifurcation lesion. Therefore, drug coated balloon (DCB) has attracted people's attention. Balloon-loaded antiproliferative drugs can fully release the drugs to the vascular wall during balloon dilation, which can inhibit the restenosis process from the beginning of injury, and show good efficacy and safety in some specific lesions. Many clinical studies have shown that DCB has good efficacy and safety in some specific lesions (ISR, small vessel disease, bifurcation disease, in situ lesion). Especially in the treatment of ISR, researchers believe that its efficacy is not inferior to DES, and it has the advantage of non-metal residues.
Quantitative flow ratio (QFR) is the second generation FFR detection method based on angiographic images. The diagnostic accuracy of QFR 0.80 for myocardial ischemic stenosis was 92.7%. Compared with QCA, the positive predictive value and negative predictive value of QFR were also significantly better than those of QCA. The latest FAVOR II results also confirm that QFR is more sensitive and specific in diagnosing myocardial ischemia caused by coronary artery stenosis than QCA, and confirm the feasibility of using QFR online in catheter lab to evaluate the functional significance of coronary artery critical lesions. However, there is no report on the treatment of de novo lesions in patients with coronary heart disease by DCB under the guidance of QFR. The aim of this study was to evaluate the safety and efficacy of drug balloon therapy for de novo lesions in patients with CHD under the guidance of QFR compared with DES implantation.

开始日期2026-01-01

申办/合作机构

南京医科大学第一附属医院

NCT04119986

/ Not yet recruitingN/AIIT

Safety and Efficacy of Drug Coated Balloon Therapy for Coronary In-stent Restenosis in Patients With Coronary Heart Disease Under the Guidance of QFR (UNIQUE-DCB-II Study)

In 1970, the first percutaneous balloon coronary angioplasty opened a new chapter of interventional therapy. However, the incidence of intracoronary restenosis was about 30%. Subsequently, bare metal stents and drug-eluting stents (DES) reduced the incidence of in-stent restenosis (ISR) to 5%-10% and it was still a bottleneck treated by percutaneous coronary intervention (PCI). Currently, ISR is mainly treated by balloon angioplasty, stent implantation and coronary artery bypass grafting.
In 2014, the guidelines of the European Society of Cardiology recommended that drug balloon therapy (DCB) and new generation DES should be the preferred strategies for ISR treatment. Compared with DES, DCB treatment can avoid the inflammation of intima caused by multi-layer stent strut, and reduce the risk of intimal hyperplasia and thrombosis in stent. However, DCB lacks sustained radial support. Even if the residual stenosis is less than 30% after sufficient pre-dilation, the elastic retraction of the intima still exists. In addition, the antiproliferative effect of paclitaxel is significantly worse than that of sirolimus and its derivatives, and there is a lack of long-term sustained release of anti-proliferative drugs. Compared with DCB, DES can obtain long-term stable radial support and long-term anti-proliferation effect, but stent struts exposed in the vascular lumen are at risk of stent thrombosis. The new generation of DES improves the design of stent platform, improves the polymer coating, and applies new anti-proliferative drugs. It effectively reduces the inflammation of vascular wall, speeds up the process of vascular re-endothelialization, promotes early vascular repair, and significantly reduces the incidence of stent thrombosis. Recent BIOLUXRCT, RESTORE and DARE studies provide more powerful evidence for the treatment of ISR by new generation DES.
Quantitative flow ratio (QFR) is the second generation FFR detectional method based on coronary contrast image. The latest FAVOR II results also confirm that QFR is more sensitive and specific than quantitative coronary analysis (QCA) in the diagnosis of myocardial ischemia caused by coronary artery stenosis. However, there is no report of ISR treated with DCB under the guidance of QFR. The aim of this study was to evaluate the safety and efficacy of DCB in the treatment of in-stent restenosis in patients with coronary heart disease (CHD) under the guidance of QFR compared with DES implantation.

开始日期2026-01-01

申办/合作机构

南京医科大学第一附属医院

ChiCTR2500108745

临床4期IIT

Adebrelimab combined with vunakizumab for first-line maintenance treatment of extensive-stage small cell lung cancer: A prospective, single-center, single-arm clinical study

开始日期2025-09-15

申办/合作机构

南京医科大学第一附属医院

[+2]

100 项与南京医科大学第一附属医院相关的临床结果

登录后查看更多信息

0 项与南京医科大学第一附属医院相关的专利（医药）

登录后查看更多信息

12,258

项与南京医科大学第一附属医院相关的文献（医药）

2026-01-01·BIOMATERIALS

Gold-embedded yolk-shell mesoporous organosilica nanocomposite for microwave-enhanced targeted chemotherapy and immune modulation in hepatocellular carcinoma

Article

作者: Wang, Shouju ; Liu, Sheng ; Sun, Hanyao ; Zhang, Jie ; Tian, Wei ; Li, Yang ; Luo, Xi ; Xu, Jiaqi ; Luo, Yafei ; Li, Na ; Tang, Jiajia ; Shi, Haibin ; Lu, Shangyu ; Yuan, Chen ; Xu, Ziqing

Microwave ablation (MWA), an innovative therapy for hepatocellular carcinoma (HCC), faces challenges of limited thermal effects and a tumor immunosuppressive microenvironment. Addressing these, we developed an advanced Au@PMO@DOX-Lac nanocomposite, aimed at enhancing the microwave thermal effect, while simultaneously facilitating targeted chemotherapy and immunomodulation. This nanocomposite, incorporating gold-embedded yolk-shell mesoporous organosilica nanoparticles, amplifies microwave thermal effects from two aspects: molecular hotspots created at the gold-silica interface and the confinement effect within its hollow nanostructure. These enhancements facilitate more effective tumor ablation with reduced microwave power requirements and shorter treatment durations. Additionally, surface modification with lactobionic acid and loading with doxorubicin allow the nanocomposite to perform precise, targeted synergistic chemotherapy. After ablation, the nanocomposite increased cytotoxic T cells and reduced regulatory T cells, while shifting macrophages from the immunosuppressive M2 to the anti-tumor M1 phenotype, significantly enhancing localized and systemic anti-tumor immune responses. Overall, this multifunctional nanocomposite not only overcomes the thermal limitations of MWA but also addresses the tumor's immunosuppressive environment, providing a promising approach for treating HCC and potentially other cancers with similar therapeutic challenges.

2026-01-01·PATTERN RECOGNITION

Multi-graph Graph matching for coronary artery semantic labeling in invasive coronary angiograms

Article

作者: Xu, Zhihui ; Zhao, Chen ; Esposito, Michel ; Zhou, Weihua ; Baral, Pukar

Coronary artery disease (CAD) stands as the leading cause of death worldwide, and invasive coronary angiography (ICA) remains the gold standard for assessing vascular anatomical information. However, deep learning-based methods encounter challenges in generating semantic labels for arterial segments, primarily due to the morphological similarity between arterial branches and varying anatomy of arterial system between different projection view angles and patients. To address this challenge, we model the vascular tree as a graph and propose a multi-graph matching (MGM) algorithm for coronary artery semantic labeling. The MGM algorithm assesses the similarity between arterials in multiple vascular tree graphs, considering the cycle consistency between each pair of graphs. As a result, the unannotated arterial segments are appropriately labeled by matching them with annotated segments. Through the incorporation of anatomical graph structure, radiomics features, and semantic mapping, the proposed MGM model achieves an impressive accuracy of 0.9471 for coronary artery semantic labeling using our multi-site dataset with 718 ICAs. With the semantic labeled arteries, an overall accuracy of 0.9155 was achieved for stenosis detection. The proposed MGM presents a novel tool for coronary artery analysis using multiple ICA-derived graphs, offering valuable insights into vascular health and pathology.

2025-12-31·RENAL FAILURE

Delta corticomedullary apparent diffusion coefficient on MRI as a biomarker for prognosis in IgA nephropathy

Article

作者: Jiang, Ling ; Zhang, Chengning ; Zhang, Bo ; Yuan, Yanggang ; Sun, Bin ; Lu, Fang ; Xing, Changying ; Huang, Zhimin ; Mao, Huijuan ; Qian, Li ; Duan, Suyan ; Zeng, Ming ; Pan, Ying ; Wang, Zitao ; Zhang, Yudong

OBJECTIVES:

To explore the association of the cortico-medullary difference in apparent diffusion coefficient (ΔADC) with clinicopathological parameters of disease activity at the time of biopsy, and with the prognositic risk stratification in IgA nephropathy (IgAN) patients.

METHODS:

We included 112 patients with biopsy-proven IgAN who measured ΔADC. Patients underwent a kidney biopsy and diffusion-weighted magnetic resonance imaging within one week of the biopsy. Clinicopathological characteristics were compared according to different ΔADC levels. The effect of ΔADC on eGFR and kidney fibrosis was explored using multivariate regression and ROC analysis. An individual's 5-year risk probability of progressing to ESKD or decreasing of eGFR > 50% was calculated by the guidelines-recommended international risk-prediction tool in IgAN. The effect of ΔADC on prognostic risk stratification was assessed. Net reclassification improvement (NRI) was used to evaluate the model performance.

RESULTS:

The average ΔADC was 168.89 ± 85.1 x10^-6 mm²/s. ΔADC levels decreased significantly with increasing chronic kidney disease (CKD) stages (p = 0.0038). Spearman correlation analysis revealed that ΔADC was positively correlated with eGFR, hemoglobin, serum albumin, while negatively correlated with levels of serum creatine (Scr), blood urea nitrogen (BUN), T score of Oxford classification and Lee grades (p < 0.05). Moreover, we showed that ΔADC was independently associated with eGFR (β = 0.04, 95% CI = [0.003, 0.077], p = 0.033) demonstrated by a backward stepwise multivariate linear regression analysis. Besides, ΔADC, a combination of ΔADC and eGFR showed an AUC of 0.776 (60% sensitivity and 85.3% specificity) and an AUC of 0.875 (100% sensitivity and 69.6% specificity) respectively for evaluating kidney interstitial fibrosis (IF) severity. Furthermore, ΔADC showed an AUC of 0.792 (95% CI 0.677-0.906) for differentiating higher progression risk categories from lower categories (specificity = 91.6%, sensitivity = 58.8%). The low-ΔADC group (≤ median value 167.1 × 10^-6 mm²/s) was associated with 7.509-fold higher likelihood of higher progression risk compared to the high-ΔADC group (>167.1 × 10^-6 mm²/s) in a fully-adjusted model. And reclassification analyses confirmed that the final adjusted model improved NRI.

CONCLUSIONS:

ΔADC was significantly associated with kidney function and enabled a reliable evaluation of kidney IF severity in IgAN patients. Low ΔADC can predict a high 5-year kidney progression risk in IgAN, independent of important clinical factors. Moreover, the predictive ability to identify patients at high risk of severe kidney fibrosis and adverse progression estimates with satisfactory accuracy, facilitating ΔADC a promising and noninvasive tool in complementarily evaluating disease activity and the prognostic risk stratification in patients with IgAN.

436

项与南京医科大学第一附属医院相关的新闻（医药）

2025-09-05

·先声药业

新一代抗失眠药达利雷生开出全国首张处方，开启科学睡眠新时代

2025年9月5日，全球新一代抗失眠药物达利雷生在首都医科大学宣武医院成功开出全国首张处方，标志着该药正式进入中国临床使用，为我国失眠患者带来全新的治疗选择。失眠已成为影响公众健康的重要问题。《2025年中国睡眠健康调查报告》显示，近半数中国成年人存在睡眠障碍。传统抗失眠药物虽能缩短入睡时间，但常伴随日间嗜睡、认知功能受影响等副作用，且长期使用存在依赖风险。达利雷生作为新型双食欲素受体拮抗剂，不仅能有效提升夜间睡眠质量，还可显著改善日间功能，为患者提供安全、可长期使用的治疗方案。与传统药物通过广泛抑制大脑活动的作用机制不同，达利雷生精准作用于导致失眠的“过度觉醒”机制，促进生理性自然睡眠。其约8小时的血药浓度半衰期设计，既有效减少夜间觉醒，又避免次日药物残留影响，确保患者白天保持清醒。开出达利雷生首张处方的宣武医院神经内科首席专家王玉平教授表示：“达利雷生使患者次日不会出现昏沉、嗜睡等不适，更有利于工作、学习和日常生活，尤其适合工作强度较高的上班族。偶尔失眠者可临时服用改善睡眠，慢性失眠患者则建议每晚一片长期服用，以持续提升睡眠质量。”江苏省人民医院神经内科主任医师张克忠教授介绍：它可以调节睡眠周期，让睡眠紊乱的患者进入正常的睡眠结构，提高睡眠质量。建议患者遵照医嘱按疗程服用，同时保持良好睡眠习惯和心态，多管齐下改善睡眠。南方医科大学南方医院睡眠研究室主任张斌教授表示：“失眠治疗的核心在于帮助患者重拾充满活力的生活。达利雷生在改善日间功能方面的独特优势以及良好的安全性，为中国的失眠患者提供了全新的治疗选择，推动整个领域从单纯关注‘改善睡眠’向‘全面恢复日间功能，提升生活质量’这一更高目标迈进。”中国III期临床试验显示，达利雷生不仅能加快入睡、减少夜间觉醒，还可显著改善日间功能，治疗剂量连续使用1年未出现依赖性或耐药性。达利雷生源自瑞士，由先声药业与Idorsia公司合作研发引进。目前该药已在全球美国、英国、瑞士、日本、加拿大等十多个国家和地区获批上市，是唯一获欧洲药品管理局批准可改善日间功能的双食欲素受体拮抗剂类失眠药，并获得中国相关治疗指南的高规格推荐。此外，该药未列入精神药品管制目录，患者获取更为便捷。目前达利雷生已在阿里健康电商平台首发。随着该药陆续进入全国零售药店，失眠患者将能更方便地获得这一兼顾夜间睡眠与日间功能的创新药物。

2025-09-03

·肿瘤界

Nature Cancer | 任善成教授团队开发AI大模型实现前列腺癌无创精准诊断与分级

点击蓝字关注我们 2025年9月2日，海军军医大学第二附属医院（上海长征医院）任善成教授联合北京大学第三医院张树栋教授、南京医科大学第一附属医院李杰教授、北京友谊医院王良教授、青岛大学附属医院聂佩教授、安徽大学邵立智教授等多学科医工交叉团队，在国际顶尖医学期刊Nature Cancer（IF=28.5）发表了题为An MRI-Pathology Foundation Model for Non-Invasive Diagnosis and Grading of Prostate Cancer的高质量研究成果。该研究开发并验证了一个基于多中心真实临床数据的用于前列腺癌高效、准确且无创诊断和分级的影像-病理基础模型。研究背景前列腺癌是全世界男性第二大癌症，随着中国人口老龄化和生活方式的西化，近年来我国前列腺癌发病率现已位居我国男性恶性肿瘤第六位。临床实践发现：50岁以上健康男性体检中，1/3的男性B超发现有前列腺结节，近10% 出现肿瘤指标PSA异常升高，这些“可疑信号”给“患者”带来了巨大的心理压力。目前，全球各大临床指南推荐磁共振（PI-RADS评分）进一步确诊。然而，PI-RADS评分存在两大缺陷（1）主观性缺陷——该评分依赖放射科医生经验判断。研究显示不同医生间判读差异高达30%（2）准确性缺陷。PI-RADS评分分为1-5级，对应不同癌变可能性。即使是最有经验的医生给出了“准确”的PI-RADS 评分，也无法“准确”判断有无肿瘤。因此，亟需一种高效、准确且无创的预测工具，用于辅助“临床可疑”患者的诊断和恶性程度分级。研究成果该研究开发并验证了一个基于多中心真实临床数据的用于前列腺癌高效、准确且无创诊断和分级的影像-病理基础模型，展示了人工智能（AI）结合MRI如何定量反映前列腺肿瘤的病理特征，进一步增强了AI基础模型在临床实践中用于癌症高效、准确且无创诊断和分级的潜力，有望显著减少不必要的前列腺穿刺活检，有效降低患者痛苦和并发症风险，最终为前列腺癌患者带来更舒适、安全的诊疗体验与福祉。研究团队招募了来自多个中心的回顾性和前瞻性患者队列（n=5747），并收集了放射学、病理学和临床检查数据。AI模型的评估涵盖了时间外部测试、空间外部测试、人口外部测试及前瞻性测试。为减少图像序列遗漏、过拟合和仪器差异带来的负面影响，研究使用了1,296,950对影像数据进行基础模型的构建，集成了自监督学习、任务多重学习、Transformer及基础模型迁移学习技术，显著提升了预测性能。研究团队设计的AI模型（MRI-based Predicted Transformer for Prostate Cancer，MRI-PTPCa），通过三个磁共振影像序列（T2WI、DWI、ADC）预测只有在病理评估中才能获得的肿瘤侵袭性信息，从而辅助临床医生诊断前列腺癌、临床显著前列腺癌及病理分级。回顾性研究中，7个机构、4个医疗中心和一个国际数据集参与了多中心测试。前瞻性研究则通过将AI模型作为独立系统、平行系统和预警系统进行测试。在实际测试中，MRI-PTPCa的预测结果与病理学评估结果具有显著一致性（P<0.001），并且优于临床评估和其他预测模型（前列腺癌AUC=0.983，95%CI 0.98-0.986；临床显著前列腺癌AUC=0.978，95%CI 0.975-0.98；分级准确率=89.1%，95%CI 88.2%-89.9%）。值得注意的是，基于MRI-PTPCa与多参数MRI联合使用，在非侵入性诊断和分级方面，其表现与病理评估不相上下。MRI-PTPCa有望成为一种新型的非侵入性前列腺癌诊断和分级工具。在可解释性方面，研究团队将影像、前列腺根治术大切片、人工智能可视化热图与量化特征进行了对照分析，以从放射学、病理学和血液学的角度解释MRI-PTPCa模型的优越性能。结果表明，MRI-PTPCa得分与真实的格里森分级之间存在显著正相关。基于类别激活映射（Class Activation Mapping，CAM）的注意力热图突出了对预测结果做出贡献的关键区域和范围。此外，T2WI、DWI和ADC在前列腺癌的诊断和分级中的贡献性也通过量化分析，证明了它们与PI-RADS专家共识的高度一致性。MRI-PTPCa的融合特征与前列腺肿瘤中的细胞强度、形态和纹理信息显著相关（P<0.01），进一步验证了影像学与病理学关联的可行性。MRI-PTPCa中的多参数MRI特征在前列腺癌的病理表型中具有显著性，包括非前列腺癌、前列腺癌和临床显著前列腺癌。基础模型的编码特征还与tPSA、fPSA和f/t PSA呈现显著相关性，为前列腺癌的分子层面信息提供了有力支持。综上所述，研究团队开发并验证了一个基于多中心真实临床数据的用于前列腺癌高效、准确且无创诊断和分级的影像-病理基础模型。研究展示了AI结合MRI如何定量反映前列腺肿瘤的病理特征，进一步增强了AI基础模型在临床实践中用于癌症高效、准确且无创诊断和分级的能力。作者介绍任善成泌尿外科主任主任医师、教授、博士生导师国家杰出青年科学基金获得者、教育部青年长江学者、国家重大科技专项首席科学家，享受国务院政府特殊津贴，国家重点学科带头人，上海市重点学科带头人，全军临床重点专科主任，全军泌尿外科中心主任，上海市一带一路泌尿系统肿瘤国际联合重点实验室主任临床主攻方向为泌尿系统肿瘤特别是前列腺癌的微创手术和综合治疗，主刀完成泌尿系统微创手术逾5000例、其中机器人手术4000余例，完成亚洲首例单孔机器人前列腺癌根治术和经会阴机器人前列腺癌根治术，获达芬奇机器人手术中国十万例杰出贡献奖；作为首席科学家，主持研发舰载手术机器人系统“海图”，完成首例跨洲际超远程舰载机器人手术主持国家自然科学基金重点项目、国家重点研发计划、军队后勤科研重点项目等国家和省部级课题计划30余项，获批国家发明专利17项；代表性成果发表在Nat Med、Nat Genet、Nature Cancer、Cell Res、Nat Cell Biol、Mol Cell、 J Clin Invest、Nat Commun、Eur Urol等国际权威期刊，总影响因子800余分，其中IF>20分11篇、IF>10分10篇，3篇论文被评为ESI高被引论文；主编/副主编中英文专著3部担任国家自然科学基金委医学科学部咨询委员会委员，亚洲泌尿外科机器人学会学术委员会主席、中华医学会泌尿外科分会青年委员会秘书长、中国医师协会泌尿外科医师分会委员、上海市医师协会泌尿外科医师分会副会长，系《中华泌尿外科杂志》编委、Asian Journal of Urology执行主编、Cancer Biology & Medicine和Military Medical Research编委以第一完成人获教育部科技进步一等奖、上海市自然科学奖一等奖，获树兰医学青年奖、上海市卫生系统青年人才最高奖“银蛇奖”一等奖；入选“国之名医·青年新锐”、张江国家自主创新示范区杰出创新创业人才等10余项人才计划声明：本文由“肿瘤界”整理与汇编，欢迎分享转载，如需使用本文内容，请务必注明出处。来源：上海长征医院官微编辑：lagertha 审核：南星

2025-09-03

·生物世界

Nature Cancer：任善成团队等开发AI大模型，实现前列腺癌无创精准诊断与分级

编辑丨王多鱼排版丨水成文前列腺癌是全世界男性第二大癌症，随着中国人口老龄化和生活方式的西化，近年来我国前列腺癌发病率以每年 13% 的增速狂飙，现已位居我国男性恶性肿瘤第六位，且发病率增速还在持续攀升。2024 年我国前列腺癌新发病例数达到 14.4 万人，预计 2030 年将达到 19.9 万人，2035 年将达到 25 万人。前列腺癌的诊断主要依赖于血液前列腺特异性抗原（PSA）、B 超和医生的直肠指诊。然而临床实践发现：50 岁以上健康男性体检中，1/3 的男性 B 超发现有前列腺结节，近 10% 出现 PSA 异常升高，这些大量的“可疑患者”给临床诊断带来极大的困难，也给“患者”带来了巨大的心理压力。目前，全球各大临床指南推荐磁共振检查（PI-RADS评分）进一步确诊，并据此进行前列腺穿刺活检来最终确定患者的诊断结果。然而，磁共振 PI-RADS 评分存在两大缺陷：1）主观性缺陷——该评分本质上依赖放射科医生经验判断。以往研究显示不同医生间判读差异高达 30%。这种主观偏差使精准诊断如同“轮盘赌局”——患者可能因假阴性延误治疗，或因假阳性承受过度穿刺。2）准确性缺陷。PI-RADS 评分分为 1-5 级，对应不同癌变可能性。即使是最高水平最有经验的医生给出了“准确”的 PI-RADS 评分，也无法“准确”判断有无肿瘤，可能导致过度穿刺或者漏诊。因此，亟需一种高效、准确且无创的预测工具，用于辅助“临床可疑”患者的诊断和恶性程度分级。人工智能（AI）等新兴技术的出现，为影像学数据与病理学结果之间的关联提供了新的工具，从而为实现无创精准诊断与分级开辟了新的可能。 2025 年 9 月 2 日，海军军医大学第二附属医院（上海长征医院）任善成教授联合北京大学第三医院张树栋教授、南京医科大学第一附属医院李杰教授、北京友谊医院王良教授、青岛大学附属医院聂佩教授、安徽大学邵立智教授等多学科医工交叉团队，在 Nature 子刊 Nature Cancer 上发表了题为：An MRI-Pathology Foundation Model for Non-Invasive Diagnosis and Grading of Prostate Cancer 的论文。该研究开发并验证了一个基于多中心真实临床数据的用于前列腺癌高效、准确且无创诊断和分级的影像-病理基础模型——MRI-PTPCa，展示了人工智能（AI）结合磁共振成像（MRI）如何定量反映前列腺肿瘤的病理特征，进一步增强了 AI 基础模型在临床实践中用于癌症高效、准确且无创诊断和分级的潜力，有望显著减少不必要的前列腺穿刺活检，有效降低患者痛苦和并发症风险，最终为前列腺癌患者带来更舒适、安全的诊疗体验与福祉。研究团队招募了来自多个中心的回顾性和前瞻性患者队列（n=5747），并收集了放射学、病理学和临床检查数据。AI 模型的评估涵盖了时间外部测试、空间外部测试、人口外部测试及前瞻性测试。为减少图像序列遗漏、过拟合和仪器差异带来的负面影响，研究使用了 1296950 对影像数据进行基础模型的构建，集成了自监督学习、任务多重学习、Transformer 及基础模型迁移学习技术，显著提升了预测性能。研究团队设计的 AI 模型——MRI-based Predicted Transformer for Prostate Cancer（MRI-PTPCa），通过三个磁共振影像序列（T2WI、DWI、ADC）预测只有在病理评估中才能获得的肿瘤侵袭性信息，从而辅助临床医生诊断前列腺癌、临床显著前列腺癌及病理分级。回顾性研究中，7 个机构、4 个医疗中心和 1 个国际数据集参与了多中心测试。前瞻性研究则通过将 AI 模型作为独立系统、平行系统和预警系统进行测试。在实际测试中，MRI-PTPCa 的预测结果与病理学评估结果具有显著一致性，并且优于临床评估和其他预测模型（前列腺癌 AUC=0.983，临床显著前列腺癌 AUC=0.978，分级准确率=89.1%）。值得注意的是，基于 MRI-PTPCa 与多参数 MRI 联合使用，在非侵入性诊断和分级方面，其表现与病理评估不相上下。MRI-PTPCa 有望成为一种新型的非侵入性前列腺癌诊断和分级工具。在可解释性方面，研究团队将影像、前列腺根治术大切片、人工智能可视化热图与量化特征进行了对照分析，以从放射学、病理学和血液学的角度解释 MRI-PTPCa 模型的优越性能。结果表明，MRI-PTPCa 得分与真实的格里森分级之间存在显著正相关。基于类别激活映射（Class Activation Mapping，CAM）的注意力热图突出了对预测结果做出贡献的关键区域和范围。此外，T2WI、DWI 和 ADC 在前列腺癌的诊断和分级中的贡献性也通过量化分析，证明了它们与 PI-RADS 专家共识的高度一致性。MRI-PTPCa 的融合特征与前列腺肿瘤中的细胞强度、形态和纹理信息显著相关，进一步验证了影像学与病理学关联的可行性。MRI-PTPCa 中的多参数 MRI 特征在前列腺癌的病理表型中具有显著性，包括非前列腺癌、前列腺癌和临床显著前列腺癌。基础模型的编码特征还与 tPSA、fPSA 和 f/t PSA 呈现显著相关性，为前列腺癌的分子层面信息提供了有力支持。综上所述，研究团队开发并验证了一个基于多中心真实临床数据的用于前列腺癌高效、准确且无创诊断和分级的影像-病理基础模型。研究展示了 AI 结合 MRI 如何定量反映前列腺肿瘤的病理特征，进一步增强了 AI 基础模型在临床实践中用于癌症高效、准确且无创诊断和分级的能力。论文链接： https://www-nature-com.libproxy1.nus.edu.sg/articles/s43018-025-01041-x 设置星标，不错过精彩推文开放转载欢迎转发到朋友圈和微信群微信加群为促进前沿研究的传播和交流，我们组建了多个专业交流群，长按下方二维码，即可添加小编微信进群，由于申请人数较多，添加微信时请备注：学校/专业/姓名，如果是PI/教授，还请注明。点在看，传递你的品味

100 项与南京医科大学第一附属医院相关的药物交易

登录后查看更多信息

100 项与南京医科大学第一附属医院相关的转化医学

登录后查看更多信息

组织架构

使用我们的机构树数据加速您的研究。

或

查看完整样例数据

管线布局

2025年09月09日管线快照

管线布局中药物为当前组织机构及其子机构作为药物机构进行统计，早期临床1期并入临床1期，临床1/2期并入临床2期，临床2/3期并入临床3期

临床前

早期临床1期

临床2期

其他

登录后查看更多信息

当前项目

药物(靶点)	适应症	全球最高研发状态

Mesothelin CAR-T cell therapy(Shenzhen BinDeBio Ltd.) ( MSLN )	胃癌更多	临床2期
褐藻多糖硫酸酯 ( CLEC-2 x CTNNB )	幽门螺杆菌感染更多	临床1/2期
KT032 ( MSLN )	间皮瘤更多	早期临床1期
177Lu-PSMA-0057 ( PSMA )	转移性前列腺腺癌更多	早期临床1期
黄连温胆汤	日夜节律睡眠障碍更多	临床阶段不明