基于 T 灌注 MRI 优化特征和肿瘤成分体积的机器学习框架进行脑胶质瘤分级。
Glioma grading using a machine-learning framework based on optimized features obtained from T perfusion MRI and volumes of tumor components.
机构信息
Centre for Biomedical Engineering, Indian Institute of Technology Delhi, New Delhi, India.
Philips Innovation Campus, Philips India Limited, Bangalore, India.
出版信息
J Magn Reson Imaging. 2019 Oct;50(4):1295-1306. doi: 10.1002/jmri.26704. Epub 2019 Mar 20.
BACKGROUND
Glioma grading between intermediate grades (Grade II vs. III and Grade III vs. IV) as well as multiclass grades (Grade II vs. III vs. IV) is challenging and needs to be addressed.
PURPOSE
To develop an artificial intelligence-based methodology for glioma grading using T perfusion parameters and volume of tumor components, and validate the efficacy of the methodology by grading on a cohort of glioma patients.
STUDY TYPE
Retrospective.
POPULATION
The development set consisted of 53 glioma patients and validation consisted of 13 glioma patients.
FIELD STRENGTH/SEQUENCE: Conventional MRI images (2D T -W, dual PD-T -W, and 3D FLAIR) and 3D T perfusion MRI data obtained at 3 T.
ASSESSMENT
Enhancing and nonenhancing components of glioma were segmented out and combined to form the region of interest (ROI) for glioma grading. Prominent vessels were removed from the selected ROI. Different T perfusion parameters from the ROI were combined with volume of tumor components to form the feature set for glioma grading. Optimization was carried out for selection of the statistic of the T perfusion parameters and the features to be used for glioma grading using sequential feature selection and random forest-based feature selection method. An optimized support vector machine (SVM) classifier was used for glioma grading.
STATISTICAL TESTS
Mean ± SD, analysis of variance (ANOVA) followed by the Tukey-Kramer test, ROC analysis.
RESULTS
Classification error for Grade II vs. III was 3.7%, for Grade III vs. IV was 5.26%, and for Grade II vs. III vs. IV was 9.43% using the proposed methodology. The mean of the values above the 90 percentile value of T perfusion parameters provided a maximum area under the curve (AUC) for intermediate grade differentiation. Random forest obtained optimal feature set provided better grading results than other methods using the SVM classifier.
DATA CONCLUSION
It was feasible to achieve low classification error for intermediate as well as multiclass glioma grading using an SVM classifier based on optimized features obtained from T perfusion MRI and volumes of tumor components.
LEVEL OF EVIDENCE
4 Technical Efficacy: Stage 3 J. Magn. Reson. Imaging 2019;50:1295-1306.
背景
对中级胶质瘤（Ⅱ级与Ⅲ级和Ⅲ级与Ⅳ级）以及多类胶质瘤（Ⅱ级与Ⅲ级与Ⅳ级）进行分级具有挑战性，需要加以解决。
目的
开发一种基于人工智能的方法，利用 T 灌注参数和肿瘤成分体积对胶质瘤进行分级，并通过对一组胶质瘤患者进行分级来验证该方法的疗效。
研究类型
回顾性。
人群
开发集包括 53 名胶质瘤患者，验证集包括 13 名胶质瘤患者。
磁场强度/序列： 常规 MRI 图像（2D T1-W、双 PD-T1-W 和 3D FLAIR）和 3D T 灌注 MRI 数据在 3T 下获得。
评估
将胶质瘤的增强和非增强成分分割出来并组合形成感兴趣区（ROI）用于胶质瘤分级。从选定的 ROI 中去除突出的血管。从 ROI 中提取不同的 T 灌注参数并结合肿瘤成分的体积形成用于胶质瘤分级的特征集。使用顺序特征选择和基于随机森林的特征选择方法对 T 灌注参数的统计量和用于胶质瘤分级的特征进行优化选择。使用优化的支持向量机（SVM）分类器进行胶质瘤分级。
统计学检验
均值±标准差，方差分析（ANOVA）后采用 Tukey-Kramer 检验，ROC 分析。
结果
使用提出的方法，Ⅱ级与Ⅲ级之间的分级错误为 3.7%，Ⅲ级与Ⅳ级之间的分级错误为 5.26%，Ⅱ级、Ⅲ级与Ⅳ级之间的分级错误为 9.43%。T 灌注参数值超过 90 百分位数的平均值提供了中间分级差异的最大曲线下面积（AUC）。使用 SVM 分类器，随机森林获得的最优特征集比其他方法提供了更好的分级结果。
数据结论
基于 T 灌注 MRI 和肿瘤成分体积获得的优化特征，使用 SVM 分类器实现中间级和多类胶质瘤的低分类错误是可行的。
证据水平
4 技术功效：3 级 J. 磁共振成像 2019；50：1295-1306.
Zotero 插件