基于随机森林的肺部肿瘤PET/CT图像计算机辅助诊断方法研究

Study on computer aided diagnosis of lung tumors PET/CT images based on random forest

针对当前基于神经网络、聚类分析以及支持向量机三种辅助诊断方法存在的诊断准确性低的问题,本研究提出一种基于随机森林的肺部肿瘤PET/CT图像计算机辅助诊断新方法。该方法首先对PET/CT图像进行预处理,包括灰度化、平滑以及分割等,然后提取PET/CT图像的灰度、形态和纹理等特征,最后利用随机森林算法进行肺部肿瘤PET/CT的辅助识别,以实现肺部肿瘤的病理诊断。结果表明,本方法的ROC曲线结果优于上述三种方法,提高了诊断准确性,可为医生诊疗提供重要参考。

Accurate diagnosis of lung tumor is important to improve the cure rate.Aiming at the low diagnostic accuracy of three kinds of auxiliary diagnostic methods based on neural network,cluster analysis and support vector machine,we proposed a computer-assisted diagnosis method of lung tumor PET/CT images based on random forest.Firstly,we preprocessed PET/CT images,including graying,smoothing,segmentation etc.Then we extracted the gray scale,morphology,texture and other features of PET/CT images.Finally,the random forest algorithm was used to assist in PET/CT identification of lung tumors,to realize the pathological diagnosis of lung tumors.The results show that the ROC curve of this method is better than that of the above three methods,indicating that the diagnosis accuracy is higher,which provides an important reference for doctor's diagnosis and treatment.