文献速递肾癌与肾肿瘤

基于术中大体外观预测肾脏肿瘤病理:一项人工智能初步研究

Predicting Renal Tumor Pathology From Intraoperative Gross Appearance: An AI-based Pilot Study.
PubMed 文献 · Urology · 2026-06-22

本研究回顾性提取2008至2024年部分肾切除术中的术中图像,训练卷积神经网络(CNN)与视觉Transformer(ViT)对6种肾脏肿瘤进行病理分类。CNN在二分类中表现中等,多分类性能不稳定,ViT表现较差;CNN模型为术中AI辅助决策提供了概念验证。

要点

  • CNN在二分类中预测ccRCC、chRCC、混合肿瘤和AML的AUC分别为0.74、0.70、0.73和0.70。
  • 多分类CNN对pRCC和oncocytoma的AUC分别为0.70和0.71,其他类别表现不一。
  • GSViT模型多数类别表现不佳,存在向ccRCC预测偏倚,且解冻预训练骨干导致快速过拟合。

核心结论

  • 基于CNN的模型能够以中等能力从术中大体外观对肾脏肿瘤病理进行分类,为AI辅助手术决策提供了概念验证。
  • 临床应⽤前需要更⼤的数据集和外部验证。

研究设计与样本

  • 回顾性提取2008至2024年接受部分肾切除术患者的术中图像,共纳入118例患者(136次手术)的443张图像。
  • 训练ResNet50-based CNN和General Surgery Vision Transformer(GSViT)对6种肿瘤类型进行分类,采用迁移学习,在留出测试数据上评估。

主要发现

  • 二分类中CNN对ccRCC(AUC 0.74)、chRCC(AUC 0.70)、混合肿瘤(AUC 0.73)和AML(AUC 0.70)取得最高AUC。
  • 多分类CNN性能更可变,pRCC(AUC 0.70)和oncocytoma(AUC 0.71)的AUC较显著。
  • GSViT模型在大多数类别中表现较差,表现出对ccRCC的预测偏倚。
  • 尝试解冻预训练骨干导致快速过拟合,突显了数据集局限性。

关键数值与推荐

  • 分类6种肿瘤类型:透明细胞肾细胞癌(ccRCC)、乳头状肾细胞癌(pRCC)、嫌色细胞肾细胞癌(chRCC)、混合嗜酸细胞肿瘤、嗜酸细胞瘤和血管平滑肌脂肪瘤(AML)。
  • 评估指标包括准确率、AUC-ROC和混淆矩阵。

临床怎么用

  • CNN模型可在术中以中等能力提示肾脏肿瘤病理类型,辅助手术决策。
  • 目前尚不能替代病理诊断,需更大数据集和外部验证。

局限性

  • 数据集规模有限,解冻预训练骨干导致快速过拟合。
  • GSViT模型表现不佳且存在预测偏倚。
  • 回顾性设计,需外部验证。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

OBJECTIVE: To evaluate the feasibility of using convolutional neural networks (CNNs) and vision transformers (ViTs) to predict renal tumor pathology intraoperatively based on gross appearance. MATERIAL AND METHODS: Intraoperative images were retrospectively extracted from surgical recordings of patients undergoing partial nephrectomy between 2008 and 2024. Static frames obtained prior to arterial clamping were curated and linked with final pathology. A ResNet50-based CNN and the General Surgery Vision Transformer (GSViT) were trained to classify 6 tumor types: clear cell RCC (ccRCC), papillary RCC (pRCC), chromophobe RCC (chRCC), hybrid oncocytic tumors, oncocytoma, and angiomyolipoma (AML). Models were trained with transfer learning, evaluated on held-out test data, and assessed using accuracy, AUC-ROC, and confusion matrices. RESULTS: A total of 443 images from 118 patients (136 surgeries) were analyzed, including ccRCC (n = 149), pRCC (n = 97), chRCC (n = 42), hybrid tumors (n = 81), oncocytoma (n = 43), and AML (n = 31). In binary classification, the CNN achieved the highest AUCs for ccRCC (0.74), chRCC (0.70), hybrid tumors (0.73), and AML (0.70). Multi-class CNN performance was more variable, with notable AUCs for pRCC (0.70) and oncocytoma (0.71). The GSViT model underperformed across most categories, demonstrating prediction bias toward ccRCC. Attempts to unfreeze pretrained backbones led to rapid overfitting, underscoring dataset limitations. CONCLUSION: CNN-based models demonstrate moderate ability to classify renal tumor pathology intraoperatively from gross appearance, providing proof of concept for AI-assisted surgical decision-making. Larger datasets and external validation are needed before clinical application.

来源:Urology,查看原文。版权归原作者所有。

引用

  1. Patel MH, Lodha R, Millan B, et al. Predicting Renal Tumor Pathology From Intraoperative Gross Appearance: An AI-based Pilot Study. Urology. 2026. PMID: 42331293. doi:10.1016/j.urology.2026.06.026

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。