论著摘要 |【MR】通过筛选乳腺X线摄影中发现的可疑乳腺病变中造影剂自由扩散MRI的影像组学特征预测恶性肿瘤(双语版)

2017-09-02 10:25:28 admin 20

Prediction of malignancy by a radiomic signature from contrast agent-free diffusion MRI in suspicious breast lesions found on screening mammography.

发表日期:2017.8.11    来源:J Magn Reson Imaging.

作者:Bickelhaupt S1Paech D1Kickingereder P1,2Steudle F1Lederer W3Daniel H4Götz M5Gählert N5Tichy D6Wiesenfarth M6Laun FB7Maier-Hein KH5Schlemmer HP1Bonekamp D1.

作者介绍

    1.Department of Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany.

    2.Department of Neuroradiology, University of Heidelberg Medical Center, Heidelberg, Germany.

    3.Radiological Clinic at the ATOS Clinic Heidelberg, Heidelberg, Germany.

    4.Radiology Center Mannheim (RZM), Mannheim, Germany.

    5.Junior Group Medical Image Computing, German Cancer Research Center (DKFZ), Heidelberg, Germany.

    6.Division of Biostatistics, German Cancer Research Center (DKFZ), Heidelberg, Germany.

    7.Department of Medical Physics in Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany.

目的

评估影像组学是一种通过扩散加权成像和T2加权序列区分乳房X摄影中发现的可疑乳腺癌是恶性或良性病变的工具。

To assess radiomics as a tool to determine how well lesions found suspicious on breast cancer screening X-ray mammographycan be categorized into malignant and benign with unenhanced magnetic resonance (MR) mammography with diffusion-weighted imaging and T2 -weighted sequences.

材料和方法

从无症状筛查人群里,将乳房X线照相中有可疑发现的50例的妇女使用1.5T对比增强乳腺MRIceMRI)进行检查。除此以外,还进行了未增强,简称为缩写扩散加权成像(ueMRI),包括T2加权(T2w),扩散加权成像(DWI)和具有背景抑制DWI序列(DWIBS)和相应的表观扩散系数(ADC)图的方法。从ueMRI得到的影像组学特征中,构建了三个Lasso监督机器学习分类器:1)单变量平均ADC模型,2)无约束影像组学模型,3)强制包含平均ADC的受限影像组学模型,并将其与经验丰富的放射科医师的临床表现进行比较。



From an asymptomatic screening cohort, 50 women with mammographically suspicious findings were examined with contrast-enhanced breast MRI (ceMRI) at 1.5T. Out of this protocol an unenhanced, abbreviated diffusion-weighted imaging protocol (ueMRI) including T2 -weighted, (T2 w), diffusion-weighted imaging (DWI), and DWI with background suppression (DWIBS) sequences and corresponding apparent diffusion coefficient (ADC) maps were extracted. From ueMRI-derived radiomic features, three Lasso-supervised machine-learning classifiers were constructed and compared with the clinical performance of a highly experienced radiologist: 1) univariate mean ADC model, 2) unconstrained radiomic model, 3) constrained radiomic model with mandatory inclusion of mean ADC.

结果

无约束和限制性影像组学分类器由11个参数组成,并且实现恶性与良性病变的区分,a.632 +引导程序受体操作特征(ROC),曲线下面积(AUC)为84.2/ 85.1%,对比经验丰富的放射科医师为95.9/ 95.9%,平均ADC77.4%



The unconstrained and constrained radiomic classifiers consisted of 11 parameters each and achieved differentiation of malignant from benign lesions with a .632 + bootstrap receiver operating characteristics (ROC) area under the curve (AUC) of 84.2%/85.1%, compared to 77.4% for mean ADC and 95.9%/95.9% for the experienced radiologist using ceMRI/ueMRI.

结论

在这项初步研究中,我们确定了两种在良性病变中区分恶性分化表现良好的ueMRI影像组学分类器,其性能优于单纯平均ADC参数。分类能力比经验丰富的乳腺放射科医生完美表现低一些,指出影像组学作为一种不需要培训来诊断决策的工具的潜力。当性能达到人类专家水平时将非常可取,并且基于我们的结果被认为是可能的,当技术的进一步发展和验证后该概念将在更大的人群中推广。



In this pilot study we identified two ueMRI radiomics classifiers that performed well in the differentiation of malignant from benign lesions and achieved higher performance than the mean ADC parameter alone. Classification was lower than the almost perfect performance of a highly experienced breast radiologist. The potential of radiomics to provide a training-independent diagnostic decision tool is indicated. A performance reaching the human expert would be highly desirable and based on our results is considered possible when the concept is extended in larger cohorts with further development and validation of the technique.

关键词

具有背景抑制扩散加权成像序列(DWIBS),表观扩散系数,具有背景抑制的扩散加权成像,磁共振,乳房X光摄影,影像组学



DWIBS; apparent diffusion coefficient; diffusion-weighted imaging with background suppression; magnetic resonance; mammography; radiomics


阅读原文:10.1002/jmri.25606

慧影医疗科技(北京)有限公司

地点:北京市海淀区中关村东升科技园B2-C103

电话:400-890-9020

邮箱:radcloud@huiyihuiying.com

关闭
图片
图片
  • 人工智能诊断云平台