改进的拉普拉斯金字塔多光谱术中引导快速融合算法

摘要

In this paper,the tumor was labeled by ICG,and the tumor details of the non-top image were enhanced while the non-tumor information of the top image was removed based on the improved Laplacian pyramid algorithm,so as to strengthen and distin-guish tumor information;the GPU technology was used to parallelize the fusion algorithm to improve the efficiency of the algorithm speed and real-time display the multi-spectral fusion image.Compared with the traditional imaging method,the imaging method pro-vided in this paper can provide more real-time and comprehensive information of the tumor,and improve the accuracy of tumor detec-tion,which will be useful for surgeon to perform the tumor resection.%本研究利用吲哚菁绿(indocyanine green,ICG)对肿瘤进行标记,通过改进的拉普拉斯金字塔算法对非顶层图像的肿瘤细节进行增强,同时去除顶层图像的非肿瘤信息,从而达到强化与区分肿瘤信息的目的;利用图形处理单元(graphics processing unit,GPU)技术将融合算法进行并行化处理,可以有效提高本研究算法的运算速度,实时显示多光谱融合图像.相对于传统成像方式,本研究提供的成像方法可以实时获取更为全面的肿瘤信息,提高了肿瘤检测的准确性,从而帮助外科医生更好地实施肿瘤切除手术.