基于超声图像纹理特征的RFA治疗无损测温技术研究

Research on Noninvasive Temperature Estimation Technology Based on Texture Features of Ultrasound Images for RFA

摘要: 在肿瘤热疗中，治疗效果与组织区域处的温度监控有直接的关系，超声可以作为热疗中实现组织无损测温的一种重要手段。本文提出了一种基于超声图像纹理分析的应用于肿瘤热疗的无损测温方法，对新鲜离体动物肾脏进行射频消融(Radiofrequency Ablation, RFA)实验并实时记录超声影像及对应温度数据，对消融前后的减影图像进行小波变换并提取处理后图像的灰度梯度共生矩阵中的特征参数与温度进行曲线拟合。结果表明，在射频消融实验中，处理后超声图像的特征参数混合熵与温度具有显著的线性相关性，验证了所提出方法应用于消融过程中温度监控的可行性。

Abstract: In tumor hyperthermia, effect is directly related to temperature monitoring during the therapy. Ultrasound can be used as one of the most important methods for noninvasive temperature meas-urement of tissues in hyperthermia. In this study, a noninvasive temperature estimation method for hyperthermia based on ultrasound image with wavelet transform and texture analysis was pro-posed. Radiofrequency ablation (RFA) was performed on animal kidneys in vitro, and ultrasound images and temperature data were collected in real time. With wavelet transform of subtraction images before and after ablation, texture features such as energy and hybrid entropy extracted from gray-level gradient co-occurrence matrix of the processed ultrasound images were linear fitted with temperature. Results demonstrated that texture features hybrid entropy obtained from im-ages processed with proposed method had high linear correlation with temperature, and verified the feasibility of the proposed approach of temperature monitoring during RFA.