3D分割联合大语言模型提高mpMRI临床显著前列腺癌检出
本研究回顾性纳入5050例接受mpMRI的患者,其中3896例用于模型开发,1154例用于多中心外部验证。研究将3D V-Net体素分割与MedGemma-IT大语言模型逐层分类指标整合为两个联合模型。外部验证中联合模型AUROC最高为0.900,显著优于所有单一指标,决策曲线显示在风险阈值0.3–0.9范围内有更优净获益。
要点
- 联合3D体积分割与大语言模型逐层分类可提高csPCa检出准确性,外部验证AUROC达0.900。
- 两个联合模型均显著优于单一V-Net或LLM指标,AUROC差值0.060–0.145,P均小于0.001。
- 在0.3–0.9风险阈值范围内联合模型的标准化净获益优于单一方法,更利于临床决策。
核心结论
- 联合3D体积分割与基于大语言模型的逐层分类可显著提升csPCa检测准确性和临床效用,优于单一模态方法。
- 该联合方法在具有明显扫描设备和扫描方案异质性的独立多中心外部验证队列中显示出较好的泛化能力。
研究设计与样本
- 回顾性多中心研究,纳入5050例行mpMRI患者,其中3896例用于模型开发,1154例用于外部验证。
- 采用3D V-Net进行体素级csPCa分割,生成阳性体积、阳性层数、阳性层率三个患者级指标。
- 外部验证通过AUROC、AUPRC、决策曲线分析、NRI和IDI进行性能评估。
诊断标准与关键数值
- 外部验证中,联合模型1的AUROC为0.900(95% CI:0.882–0.918)。
- 外部验证中,联合模型2的AUROC为0.885(95% CI:0.865–0.905)。
- 联合模型AUROC较所有单一V-Net和LLM指标的差值为0.060–0.145,所有P值小于0.001。
- 两个联合模型在多数比较中NRI和IDI改善显著,多数P值小于0.05。
- 决策曲线分析显示在临床相关风险阈值0.3–0.9范围内联合模型标准化净获益更优。
临床应用与阅片要点
- 联合模型1由V-Net阳性层率与LLM阳性层率经逻辑回归构建,输出患者级csPCa预测概率。
- 联合模型2由V-Net阳性体积与LLM阳性层率经零调整Gamma回归构建,整合体积与层面信息。
- 联合策略可缓解单一分割或单一分类方法在异构扫描条件下的性能波动。
局限性
- 本研究为回顾性设计,存在选择偏倚和标注偏倚的潜在可能。
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)。
英文原文(摘要)· 点击展开
RATIONALE AND OBJECTIVES: To develop and externally validate integrated models combining three-dimensional (3D) volumetric segmentation and large language model (LLM)-based slice-wise classification for improved detection of clinically significant Prostate Cancer (csPCa) on multiparametric Magnetic Resonance Imaging (mpMRI).
MATERIALS AND METHODS: This retrospective multi-institutional study included 5050 patients (3896 for model development and 1154 for external validation) who underwent mpMRI. A 3D V-Net was trained for voxel-wise csPCa segmentation, generating three patient-level metrics: positive volume, positive slice count, and positive slice rate. A four-billion-parameter MedGemma-IT LLM was fine-tuned for slice-level classification, producing positive slice count and rate metrics. The optimal V-Net and LLM metrics were integrated into two combined models: Combined Model 1 (logistic regression of V-Net and LLM positive slice rates) and Combined Model 2 (zero-adjusted Gamma regression of V-Net positive volume and LLM positive slice rate). Performance was evaluated in the external validation cohort using area under the receiver operating characteristic curve (AUROC), area under the precision-recall curve (AUPRC), decision curve analysis (DCA), net reclassification improvement (NRI), and integrated discrimination improvement (IDI).
RESULTS: In external validation, Combined Model 1 achieved an AUROC of 0.900 (95% CI: 0.882-0.918) and Combined Model 2 achieved 0.885 (95% CI: 0.865-0.905), significantly outperforming all individual V-Net and LLM metrics (ΔAUROC: 0.060-0.145; all P < 0.001). Both combined models demonstrated significant improvements in NRI and IDI for most comparisons (P < 0.05 for most) and provided superior standardized net benefit across clinically relevant risk thresholds (0.3-0.9) on DCA.
CONCLUSION: Integrating 3D volumetric segmentation with LLM-based slice-wise classification significantly improves csPCa detection accuracy and clinical utility compared to single-modality approaches, demonstrating promising generalizability in an independent, multi-institutional external validation cohort with substantial scanner and protocol heterogeneity.
引用
- Wang, Gao, Wang, et al. Integrating 3D Volumetric Segmentation and LLM-Based Classification for csPCa Detection on mpMRI: Multi-Institutional External Validation. Acad Radiol. 2026. PMID: 42425805. doi:10.1016/j.acra.2026.06.029
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。