Department of Ultrasound, Fujian Medical University Affiliated Union Hospital, No. 29 Xinquan Road, Fuzhou, 350001, Fujian, China.
J Med Ultrason (2001). 2024 Jan;51(1):71-82. doi: 10.1007/s10396-023-01373-0. Epub 2023 Oct 5.
This study aimed to use conventional ultrasound features, ultrasound radiomics, and machine learning algorithms to establish a predictive model to assess the risk of post-surgical recurrence of gastrointestinal stromal tumors (GISTs).
This retrospective analysis included 230 patients with pathologically diagnosed GISTs. Radiomic features were extracted from manually annotated images. Radiomic features plus conventional ultrasound features were selected using the SelectKbest analysis of variance and stratified tenfold cross-validation recursive elimination methods. Finally, five different machine learning algorithms (logistic regression [LR], support vector machine [SVM], random forest [RF], extreme gradient boosting [XGBoost], and multilayer perceptron [MLP]) were established to predict risk stratification of GISTs. The predictive performance of the established model was mainly evaluated based on the area under the receiver operating characteristic (ROC) curve (AUC) and accuracy, whereas the predictive performance of the optimal machine learning algorithm and a radiologist's subjective assessment were compared using McNemar's test.
Seven radiomics features and one conventional ultrasound feature were selected to construct the machine learning models for GIST risk classification. The mentioned five machine learning models were able to predict the malignant potential of GISTs. LR and SVM outperformed other classifiers on the test set, with LR achieving an accuracy of 0.852 (AUC, 0.881; sensitivity, 0.871; specificity, 0.826) and SVM achieving an accuracy of 0.852 (AUC, 0.879; sensitivity, 0.839; specificity, 0.870), and proved significantly better than the radiologist (accuracy, 0.691; sensitivity, 0.645; specificity, 0.813).
Machine learning-based ultrasound radiomics features are able to noninvasively predict the biological risk of GISTs.
本研究旨在利用常规超声特征、超声放射组学和机器学习算法建立预测模型，以评估胃肠道间质瘤（GIST）术后复发的风险。
本回顾性分析纳入了 230 例经病理诊断为 GIST 的患者。从手动标注图像中提取放射组学特征。使用方差分析的 SelectKbest 和分层十折交叉验证递归消除法选择放射组学特征和常规超声特征。最后，建立了五种不同的机器学习算法（逻辑回归[LR]、支持向量机[SVM]、随机森林[RF]、极端梯度提升[XGBoost]和多层感知机[MLP]）来预测 GIST 的风险分层。主要基于接受者操作特征（ROC）曲线下面积（AUC）和准确性来评估所建立模型的预测性能，而使用 McNemar 检验比较了最优机器学习算法和放射科医生主观评估的预测性能。
选择了 7 个放射组学特征和 1 个常规超声特征来构建用于 GIST 风险分类的机器学习模型。上述五种机器学习模型能够预测 GIST 的恶性潜能。LR 和 SVM 在测试集上的表现优于其他分类器，LR 的准确性为 0.852（AUC 为 0.881、敏感性为 0.871、特异性为 0.826），SVM 的准确性为 0.852（AUC 为 0.879、敏感性为 0.839、特异性为 0.870），明显优于放射科医生（准确性为 0.691、敏感性为 0.645、特异性为 0.813）。
基于机器学习的超声放射组学特征能够无创预测 GIST 的生物学风险。
