文献速递乳腺技术与安全US

超声射频时间序列用于组织分型:基于纹理优化特征与多源分类方法(MOMC)的在体乳腺样本实验

Ultrasound Radio Frequency Time Series for Tissue Typing: Experiments on In-Vivo Breast Samples Using Texture-Optimized Features and Multi-Origin Method of Classification (MOMC).
PubMed 文献 · J Ultrasound Med · 2025-06-30

研究纳入118例患者的220个乳腺病灶数据点,从超声射频时间序列中提取283个纹理特征,经粒子群优化筛选至131个,采用多源分类方法(MOMC)联合支持向量机与集成分类器进行良恶性及BI-RADS分类,5折交叉验证下2类、3类、4类分类准确率分别为98.57±1.09%、91.53±0.89%、83.71±1.30%。

要点

  • 基于超声射频时间序列的机器学习方法在2类分类中达到98.57±1.09%的准确率。
  • 3类和4类分类准确率分别为91.53±0.89%和83.71±1.30%,随分类类别增多而下降。
  • 纹理特征(Gabor滤波、GLCM、GLRLM、GLSZM、GLDM)经粒子群优化筛选后用于分类。

核心结论

  • 研究提出一种基于在体超声射频时间序列数据的机器学习方法,可区分不同乳腺病灶类型, findings 强调其在提高分类准确率方面的效力,有望推动乳腺癌筛查的计算机辅助诊断(CAD)进展。

研究设计与样本

  • 研究分析了118例患者的220个数据点,使用配备线性探头的SuperSonic Imagine Aixplorer®医学/研究系统采集射频时间序列数据集(RFTSBU),由专家放射科医生在B模式图像上手动选取感兴趣区(ROI),从每个ROI提取283个特征,经粒子群优化(PSO)筛选至131个高效特征,采用多源分类方法(MOMC)结合支持向量机(MOMC-SVM)和集成分类器(MOMC-ensemble)进行分类,采用5折交叉验证。

诊断标准与关键数值

  • 2类分类(良性、可能良性、可疑、恶性中的二分类)使用MOMC-SVM和MOMC-ensemble分类器,5折交叉验证准确率为98.57±1.09%。
  • 3类分类使用MOMC-SVM和MOMC-ensemble分类器,5折交叉验证准确率为91.53±0.89%。
  • 4类分类使用MOMC-SVM和MOMC-ensemble分类器,5折交叉验证准确率为83.71±1.30%。

局限性

  • 原文未报告局限性

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

OBJECTIVES: One of the most promising auxiliaries for screening breast cancer (BC) is ultrasound (US) radio-frequency (RF) time series. It has the superiority of not requiring any supplementary equipment over other methods. This article sought to propound a machine learning (ML) method for the automated categorization of breast lesions-categorized as benign, probably benign, suspicious, or malignant-using features extracted from the accumulated US RF time series.

METHODS: In this research, 220 data points of the categories as mentioned earlier, recorded from 118 patients, were analyzed. The RFTSBU dataset was registered by a SuperSonic Imagine Aixplorer® medical/research system fitted with a linear transducer. The expert radiologist manually selected regions of interest (ROIs) in B-mode images before extracting 283 features from each ROI in the ML approach, utilizing textural features such as Gabor filter (GF), gray-level co-occurrence matrix (GLCM), gray-level run-length matrix (GLRLM), gray-level size zone matrix (GLSZM), and gray-level dependence matrix (GLDM). Subsequently, the particle swarm optimization (PSO) narrowed the features to 131 highly effective ones. Ultimately, the features underwent classification using an innovative multi-origin method classification (MOMC), marking a significant leap in BC diagnosis.

RESULTS: Employing 5-fold cross-validation, the study achieved notable accuracy rates of 98.57 ± 1.09%, 91.53 ± 0.89%, and 83.71 ± 1.30% for 2-, 3-, and 4-class classifications, respectively, using MOMC-SVM and MOMC-ensemble classifiers.

CONCLUSIONS: This research introduces an innovative ML-based approach to differentiate between diverse breast lesion types using in vivo US RF time series data. The findings underscore its efficacy in enhancing classification accuracy, promising significant strides in computer-aided diagnosis (CAD) for BC screening.

来源:J Ultrasound Med,查看原文。版权归原作者所有。

引用

  1. Arab M, Fallah A, Rashidi S, et al. Ultrasound Radio Frequency Time Series for Tissue Typing: Experiments on In-Vivo Breast Samples Using Texture-Optimized Features and Multi-Origin Method of Classification (MOMC). J Ultrasound Med. 2025. PMID: 40586272. doi:10.1002/jum.16745

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。