光学分子影像手术导航系统中图像融合算法研究

手术中肿瘤的精确定位一直是国际上的研究热点及挑战性问题.对光学分子影像手术导航系统中的图像融合算法进行了研究,对图像数据进行预处理后,利用梯度信息确定荧光图像兴趣点,缩小配准点的搜索区域.配准过程采用穷尽搜索策略,以平均绝对差(MAD)、均方差(MSD)和归一化积相关(NPROD)作为相似性度量,并利用域法修正的方法对兴趣点坐标进行修正,避免了因兴趣点过度偏离图像中心而导致配准失败.该算法将32组图像数据的平均配准时间从原始算法的539.384 s缩短到43.200 6 s(MAD),46.685 3 s(MSD)及85.978 6 s(NPROD).采用这3种相似性度量算法的融合效果均优于原始算法.

Research on image fusion algorithm in optical molecular imaging surgical navigation system

Intraoperative precise iocalization of tumor had been an international research hotspot and challeng -ing issue .The image fusion algorithm in optical molecular imaging surgical navigation system was conducted . Preprocessed the image data , got the interest point of fluorescence image by using its gradient information , the search region of registration points was narrowed down .Registration process used exhaustion search strategy , Mean Absolute Deviation （ MAD ） , Mean Square Deviation （ MSD ） and Normalized Product Correlation （ NPROD） as the similarity metrics , and the coordinates of interest point were corrected by the domain correc-tion method to avoid registration failure caused by excessive deviation of the interest points from image center . The algorithm shortened the average registration time of 32 sets of image data from 539.384 s to 43.200 6 s （MAD）,46.685 3 s（MSD） and 85.978 6 s（NPROD）.The image fusion effect of algorithm using the three similarity measure was found better than that of the original algorithm .