Faster RCNN模型和SCP方法在肝包虫病病灶位置估计中的研究

Research on Faster Region-Convolutional Neural Networks Model and SCP Method in Estimating the Location of Hepatic Echinococcosis

临床医生凭借经验判断肝包虫病在肝脏区域常发病位置具有一定的主观性,因此,本研究将Faster RCNN目标检测模型应用于肝包虫病的检测,并与SCP方法结合对肝包虫病在肝脏区域的发病位置进行估计。本研究首先使用基于ResNet101网络的目标检测模型检测肝包虫病病灶,然后使用基于SCP方法建立的极坐标系统计肝包虫病在肝脏区域常发位置。实验结果表明基于ResNet101网络的目标检测模型能够有效提取目标的特征,检测准确率达到了89.6%,基于SCP方法统计出的肝包虫病常发病位置主要在肝脏的Ⅳ区、Ⅵ/Ⅶ区。本研究通过目标检测模型检测肝包虫病病灶,可以辅助医生诊断疾病,减少漏检、错检的发生。对于统计出的肝包虫病在肝脏区域的常发病位置,能够辅助医生早期发现疾病,做到早发现、早诊断、早治疗。

It is subjective for clinicians to judge the common location of hepatic echinococcosis by experience.Therefore,the Faster RCNN object detection model was applied to detect hepatic echinococcosis in this study,and the SCP method was combined to estimate the location of hepatic echinococcosis in the liver region.In this study,the object detection model based on ResNet101 network was used to detect the lesions of hepatic echinococcosis,and then used the polar coordinate system based on SCP method to calculate the common location of hepatic hydatid disease in the liver region.The experimental results showed that the object detection model based on ResNet101 network can effectively extract the features of the target,and the detection accuracy rate reached 89.6%.The frequent occurrence locations of hepatic echinococcosis based on SCP method are mainly in the IV andⅥ/Ⅶregions of the liver.In this study,the detection of liver hydatid lesions by object detection model can help doctors diagnose the disease and reduce the occurrence of missed detection and wrong detection.For the statistics of liver hydatid disease in the liver region of the common location,it can assist doctors in early detection,early diagnosis,and early treatment of the disease.