AUTHOR=Han Yu , Wang Wen , Yang Yang , Sun Ying-Zhi , Xiao Gang , Tian Qiang , Zhang Jin , Cui Guang-Bin , Yan Lin-Feng 

TITLE=Amide Proton Transfer Imaging in Predicting Isocitrate Dehydrogenase 1 Mutation Status of Grade II/III Gliomas Based on Support Vector Machine

JOURNAL=Frontiers in Neuroscience

VOLUME=14

YEAR=2020

URL=https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2020.00144

DOI=10.3389/fnins.2020.00144

ISSN=1662-453X

ABSTRACT=<sec><title>Background</title><p>To compare the efficacies of univariate and radiomics analyses of amide proton transfer weighted (APT<sub>W</sub>) imaging in predicting isocitrate dehydrogenase 1 (<italic>IDH1</italic>) mutation of grade II/III gliomas.</p></sec><sec><title>Methods</title><p>Fifty-nine grade II/III glioma patients with known <italic>IDH1</italic> mutation status were prospectively included (<italic>IDH1</italic> wild type, 16; <italic>IDH1</italic> mutation, 43). A total of 1044 quantitative radiomics features were extracted from APT<sub>W</sub> images. The efficacies of univariate and radiomics analyses in predicting <italic>IDH1</italic> mutation were compared. Feature values were compared between two groups with independent <italic>t-</italic>test and receiver operating characteristic (ROC) analysis was applied to evaluate the predicting efficacy of each feature. Cases were randomly assigned to either the training (<italic>n</italic> = 49) or test cohort (<italic>n</italic> = 10) for the radiomics analysis. Support vector machine with recursive feature elimination (SVM-RFE) was adopted to select the optimal feature subset. The adverse impact of the imbalance dataset in the training cohort was solved by synthetic minority oversampling technique (SMOTE). Subsequently, the performance of SVM model was assessed on both training and test cohort.</p></sec><sec><title>Results</title><p>As for univariate analysis, 18 features were significantly different between <italic>IDH1</italic> wild-type and mutant groups (<italic>P</italic> &lt; 0.05). Among these parameters, <italic>High Gray Level Run Emphasis All Direction offset 8 SD</italic> achieved the biggest area under the curve (AUC) (0.769) with the accuracy of 0.799. As for radiomics analysis, SVM model was established using 19 features selected with SVM-RFE. The AUC and accuracy for <italic>IDH1</italic> mutation on training set were 0.892 and 0.952, while on the testing set were 0.7 and 0.84, respectively.</p></sec><sec><title>Conclusion</title><p>Radiomics strategy based on APT image features is potentially useful for preoperative estimating <italic>IDH1</italic> mutation status.</p></sec>