奇异值分解滤波在甲状腺结节微波消融围术期超声造影评估中的应用

Application of singular value decomposition filtering in the assessment of perioperative contrast-enhanced ultrasound of microwave ablation of thyroid nodules

目的:探讨奇异值分解(SVD)在优化超声造影(CEUS)图像质量以及甲状腺结节微波消融(MWA)围术期CEUS评估中的应用价值。方法:收集2023年4月在武汉科技大学附属华润武钢总医院2例行甲状腺结节MWA患者其围术期的二维超声、CEUS图像数据集,根据归一化互相关算法获取基于CEUS图像进行SVD滤波处理后的造影图像,比较原始CEUS图像和SVD滤波处理后的造影图像的信噪比(SNR)及微血管密度(MVD)的差异。结果:与甲状腺结节MWA术前和术后的原始CEUS图像相比,经SVD滤波处理后的造影图像能够更清晰显示结节内血管结构且检测出更多的血流信号。术后原始CEUS图像的平均SNR为(1.07±0.66)dB,术后经SVD滤波处理后的造影图像平均SNR为(6.21±1.01)dB(t=-6.015,P=0.038)。术后原始CEUS图像的平均MVD为(0.60±0.33)%,术后经SVD滤波处理后的造影图像平均MVD为(5.42±0.67)%(t=-9.076,P=0.029)。术后经SVD滤波处理后的造影图像SNR、MVD,显著高于术后原始CEUS图像的SNR、MVD,其研究结果均具有统计学意义。结论:SVD滤波可有效抑制噪声并增强微血流信号,通过提升造影图像质量以用于甲状腺结节MWA围术期的精准诊断,灵敏地检测出术后残留的消融不完全区域,以指导临床医师进行更彻底地消融治疗,为精准消融疗效评估提供决策支持。

Objective:To explore the application value of singular value decomposition(SVD)in optimizing the image quality of contrast-enhanced ultrasound(CEUS)and perioperative CEUS assessment of microwave ablation(MWA)for thyroid nodule.Methods:The data sets of two-dimensional ultrasound and CEUS images of two patients with thyroid nodules,who underwent MWA in China Resources&Wisco General Hospital affiliated to Wuhan University of Science and Technology from April 2023,were collected.The contrast images based on CEUS images after SVD filtering were obtained according to the normalized cross-correlation algorithm,and the differences of signal-to-noise ratio(SNR)and micro-vessel density(MVD)between the contrast images after SVD filtering and original CEUS images were compared.Results:Compared with the original CEUS images of thyroid nodules before and after MWA,the contrast images that were processed by SVD filtering can more clearly show the vascular structure with thyroid nodules,and can detect more signals of blood flow.The results showed that the average SNR of the original CEUS images after operation was(1.07±0.66)dB,and the average SNR of the contrast images by SVD filtering after operation was(6.21±1.01)dB(t=-6.015,P=0.038).The average MVD of the original CEUS images after operation was(0.60±0.33)%,and the average MVD of the contrast images of SVD filtering after operation was(5.42±0.67)%(t=-9.076,P=0.029).It can show that the SNR and MVD of the contrast images of SVD-filtering after operation were significantly higher than these of the original CEUS images after operation,and the results of these studies were significant.Conclusion:SVD filtering can effectively suppress noise and enhance the signals of micro-blood flow.The improved quality of contrast images can be used in the accurate diagnosis of perioperative MWA of thyroid nodules,it can sensitively detect the residually incomplete areas of ablation after operation,so as to guide clinicians to carry out more thorough ablation treatment and provide decision support for the assessment of curative effect of accurate ablation.