Chinese General Practice ›› 2025, Vol. 28 ›› Issue (17): 2149-2155.DOI: 10.12114/j.issn.1007-9572.2024.0160

Special Issue: 乳腺癌最新文章合辑

• Original Research • Previous Articles     Next Articles

Predictive Value of S-Detect Combined with Virtual Touch Tissue Imaging Quantification in Axillary Lymph Node Metastasis of Breast Cancer

  

  1. 1Department of Ultrasound, First Affiliated Hospital, Shihezi University, Shihezi 832008, China
    2First Affiliated Hospital, Shihezi University, NHC Key Laboratory of Prevention and Treatment of Central Asia High Incidence Diseases, Shihezi 832008, China
  • Received:2024-04-21 Revised:2024-08-24 Published:2025-06-15 Online:2025-04-22
  • Contact: LI Jun

S-Detect联合声触诊组织成像定量技术对乳腺癌腋窝淋巴结转移的预测价值研究

  

  1. 1832008 新疆维吾尔自治区石河子市,石河子大学第一附属医院超声科
    2832008 新疆维吾尔自治区石河子市,石河子大学第一附属医院国家卫健委中亚高发病防治重点实验室
  • 通讯作者: 李军
  • 作者简介:

    作者贡献:

    邓雅倩负责文章的构思与设计,撰写论文;曹春莉负责研究的实施与可行性分析;邓雅倩、马金梅进行数据收集;李文肖、徐泽林负责数据整理;徐泽林负责统计学处理;邓雅倩、曹春莉负责结果的分析与解释;邓雅倩、马金梅进行论文的修订;成静负责文章的质量控制及审校;李军对文章整体负责,监督管理。

  • 基金资助:
    国家自然科学基金资助项目(82060318); 中国医学科学院中央级公益性科研院所基本科研业务费专项资金资助项目(2020-PT330-003); 兵团科技攻关计划项目(2019DB012); 石河子大学第一附属医院青年基金项目(QN202126,QN202107)

Abstract:

Background

Axillary lymph node metastasis in breast cancer is an important factor in assessing disease progression and prognosis, and there is an urgent need for non-invasive and accurate methods to assess axillary lymph node status. In recent years, the integration of artificial intelligence and medical imaging technology has shown great potential. Among them, S-Detect technology, with its powerful image analysis capability, combined with the accurate quantitative assessment of virtual touch tissue imaging quantification (VTIQ) technology, provides new possibilities for the prediction of axillary lymph node metastasis in breast cancer.

Objective

To explore the value of the conventional ultrasound-based artificial intelligence S-Detect combined with VTIQ technique to predict axillary lymph node metastasis in breast cancer.

Methods

Data of 148 female breast cancer patients (166 masses in total) who underwent surgery at the First Affiliated Hospital, Shihezi University from February 2022 to February 2024 were retrospectively collected and divided into metastatic group (n=61) and non-metastatic group (n=105) based on axillary lymph node pathology results. All patients underwent routine ultrasound, VTIQ and S-Detect examinations before surgery. Univariate and multivariate Logistic regression analyses were used to explore the influence of each observational index on axillary lymph node metastasis, screen meaningful indexes and establish a Logistic regression prediction model. The predictive value of the model was evaluated by the ROC curve.

Results

The results of multivariate Logistic regression analysis showed that breast cancer mass border (OR=0.619, 95%CI=0.540-0.693), margin (OR=0.563, 95%CI=0.484-0.640), calcification (OR=0.559, 95%CI=0.480-0.636), aspect ratio (OR=0.540, 95%CI=0.461-0.617) and SWVmean (OR=0.794, 95%CI=0.725-0.853) were independent influences in predicting axillary lymph node metastasis in breast cancer (P<0.05). Logistic equations were constructed: Logistic (P) =-14.293+1.664×border+1.315×margin+1.757×calcification+1.341×aspect ratio+1.196×blood flow classification+0.736×SWVmax-3.942×SWVcentre+0.710×SWVmean. The AUC of this joint prediction model was 0.902 (95%CI=0.847-0.943), which was greater than the AUCs of each independent influencing factor (all P<0.05), and the AUC values of the joint prediction model were greater than the AUCs of the independent prediction models of conventional ultrasound (AUC=0.605, 95%CI=0.526-0.680), S-Detect (AUC=0.672, 95%CI=0.595-0.743), and VTIQ (AUC=0.794, 95%CI=0.725-0.853). The agreement between this joint prediction model and pathological results was good (Kappa=0.732, P<0.05) .

Conclusion

The prediction model constructed by combining S-Detect and VTIQ techniques based on conventional ultrasound has certain clinical predictive value for axillary lymph node metastasis in breast cancer.

Key words: Breast cancer, Axillary lymph node metastasis, Ultrasound, S-Detect intelligent assisted diagnosis, Virtual touch tissue imaging quantification

摘要:

背景

乳腺癌腋窝淋巴结转移是评估疾病进展和预后的重要因素,目前亟需无创、准确的方法来评估腋窝淋巴结状态。近年来,人工智能与医学影像技术的融合展现出巨大潜力。其中,S-Detect技术具有强大的图像分析能力,结合声触诊组织成像定量(VTIQ)技术的精准量化评估,为乳腺癌腋窝淋巴结转移的预测提供了新的可能。

目的

探讨基于常规超声的人工智能S-Detect联合VTIQ技术预测乳腺癌腋窝淋巴结转移的价值。

方法

回顾性收集2022年2月—2024年2月于石河子大学第一附属医院进行手术的148例女性乳腺癌患者(共166个肿块)资料,依据腋窝淋巴结病理结果分为转移者(n=61)与未转移者(n=105)。所有患者术前行常规超声、VTIQ及S-Detect检查。采用单因素和多因素Logistic回归分析探讨各观察指标对腋窝淋巴结转移的影响,筛选有意义的指标并建立Logistic回归预测模型,采用受试者工作特征(ROC)曲线评价该模型的预测价值。

结果

多因素Logistic回归分析结果显示,乳腺癌肿块边界(OR=0.619,95%CI=0.540~0.693)、边缘(OR=0.563,95%CI=0.484~0.640)、钙化(OR=0.559,95%CI=0.480~0.636)、纵横比(OR=0.540,95%CI=0.461~0.617)及剪切波速度平均值(SWVmean)(OR=0.794,95%CI=0.725~0.853)是预测乳腺癌腋窝淋巴结转移的独立影响因素(P<0.05)。依此构建Logistic方程:Logistic(P)=-14.293+1.664×边界+1.315×边缘+1.757×钙化+1.341×纵横比+1.196×血流分级+0.736×SWV最大值(SWVmax)-3.942×SWV中间值(SWVcentre)+0.710×SWVmean。该联合预测模型的ROC曲线下面积(AUC)为0.902(95%CI=0.847~0.943),均大于各独立影响因素的AUC(P<0.05),且联合预测模型的AUC值均大于常规超声(AUC=0.605,95%CI=0.526~0.680)、S-Detect(AUC=0.672,95%CI=0.595~0.743)以及VTIQ(AUC=0.794,95%CI=0.725~0.853)各独立预测模型的AUC(P<0.05),该联合预测模型与病理结果的一致性良好(Kappa=0.732,P<0.05)。

结论

基于常规超声的S-Detect联合VTIQ技术构建的预测模型对乳腺癌腋窝淋巴结转移有一定临床预测价值。

关键词: 乳腺癌, 腋窝淋巴结转移, 常规超声, S-Detect智能辅助诊断, 声触诊组织成像定量技术

CLC Number: