通过PET-CT图像纹理特征预测软组织肉瘤转移性

提出了一种针对软组织肉瘤转移性预测的辅助诊断方法,该方法通过对患者的FDG-PET和CT诊断图像进行纹理特征分析,共提取了105个特征,其中包括灰度共生矩阵的24个特征和其他81个灰度等级的特征,分别利用支持向量机、K近邻和随机森林等机器学习算法建立预测模型,并采用网格搜索法对其参数进行优化.最后使用留一交叉验证法对各模型进行验证.通过评估各模型性能,选择支持向量机作为最终预测模型,得到了80%的平均精确度.此外,该模型的敏感度达到81%,特异性达到79%,表明该模型预测结果具有一定的可靠性,可以对STS进行辅助诊断并通过更好的适应性治疗来改善患者的预后.

Prediction of soft tissue sarcoma metastasis by PET-CT image texture features

This paper proposes an auxiliary diagnostic method for soft tissue sarcoma metastasis prediction. This method extracts 105 features which include 24 features of the Gray Level Co-occurrence Matrix(GLCM)and other 81 grayscale features by analyzing the texture features of FDG-PET and CT diagnostic images. Machine learning algorithms such as Support Vector Machine(SVM), K-Nearest Neighbor(KNN)and Random Forest(RF)are used to build prediction models, and their parameters are optimized by grid search method. Finally, the models are evaluated by the leave-one-out cross-validation method. By evaluating the performance of each model, support vector machine can be selected as the final prediction model, and the average accuracy of 80% is obtained. In addition, the sensitivity and specificity of this model reached 81% and 79% respectively, indicating that the predicted results of this model have certain reliability, which can be used to aid diagnosis of STS and improve patient outcomes through better adaptive treatment.