使用扩散加权成像的纹理特征进行膀胱癌分级的放射学评估(英文)

2017-02-15 11:40:46 admin 28

Radiomics assessment of bladder cancer grade using texture features from diffusion-weightedimaging.

J Magn Reson Imaging. 2017 Feb 15

Zhang X1, Xu X1, Tian Q2, Li B1, Wu Y1, Yang Z3, Liang Z4, Liu Y1, Cui G2, Lu H1.

Author information

    1.Department of Biomedical Engineering, Fourth Military Medical University, Xi'an, Shaanxi, P.R. China.

    2.Department of Radiology, Tangdu Hospital, Fourth Military Medical University, Xi'an, Shaanxi, P.R. China.

    3.Department of Urology, Tangdu Hospital, Fourth Military Medical University, Xi'an, Shaanxi, P.R. China.

    4.Departments of Radiology, Computer Science and Biomedical Engineering, State University of New York, Stony Brook, New York, USA.

Abstract

PURPOSE


To 1) describe textural features from diffusion-weighted images (DWI) and apparent diffusion coefficient (ADC) maps that can distinguish low-grade bladder cancer from high-grade, and 2) propose a radiomics-based strategy for cancer grading using texture features.

MATERIALS AND METHODS


In all, 61 patients with bladder cancer (29 in high- and 32 in low-grade groups) were enrolled in this retrospective study. Histogram- and gray-level co-occurrence matrix (GLCM)-based radiomics features were extracted from cancerous volumes of interest (VOIs) on DWI and corresponding ADC maps of each patient acquired from 3.0T magnetic resonance imaging (MRI). A Mann-Whitney U-test was applied to select features with significant differences between low- and high-grade groups (P < 0.05). Then support vector machine with recursive feature elimination (SVM-RFE) and classification strategy was adopted to find an optimal feature subset and then to establish a classification model for grading.

RESULTS


A total 102 features were derived from each VOI and among them, 47 candidate features were selected, which showed significant intergroup differences (P < 0.05). By the SVM-RFE method, an optimal feature subset including 22 features was further selected from candidate features. The SVM classifier using the optimal feature subset achieved the best performance in bladder cancer grading, with an area under the receiver operating characteristic curve, accuracy, sensitivity, and specificity of 0.861, 82.9%, 78.4%, and 87.1%, respectively.

CONCLUSION


Textural features from DWI and ADC maps can reflect the difference between low- and high-grade bladder cancer, especially those GLCM features from ADC maps. The proposed radiomics strategy using these features, combined with the SVM classifier, may better facilitate image-based bladder cancer grading preoperatively.

LEVEL OF EVIDENCE


3 J. Magn. Reson. Imaging 2016.

KEYWORDS


apparent diffusion coefficient; bladder cancer grade; radiomics; support vector machine; texture feature


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